MLIB/MTRACE

MATLAB-dSPACE
Interface Libraries
Release 7.1 May 2011

How to Contact dSPACE

Mail:

dSPACE GmbH
Rathenaustrae 26
33102 Paderborn
Germany

Tel.:

+49 5251 1638-0

Fax:

+49 5251 16198-0

E-mail:

[email protected]

Web:

http://www.dspace.com

How to Contact dSPACE Support

There are different ways to contact dSPACE Support:
Visit our Web site at http://www.dspace.com/goto?support
Send an e-mail or phone:
General Technical Support:
[email protected]
+49 5251 1638-941
Use the dSPACE Installation Manager:
On your dSPACE DVD at \Tools\InstallationManager
Via Start Programs dSPACE Installation Manager (after installation of the dSPACE
software)
At http://www.dspace.com/goto?im
You can always find the latest version of the dSPACE Installation Manager here.
dSPACE recommends that you use the dSPACE Installation Manager to contact dSPACE
Support.

Software Updates and Patches

dSPACE strongly recommends that you download and install the most recent patches for
your current dSPACE installation. Visit http://www.dspace.com/goto?support for software
updates and patches.

Important Notice
This document contains proprietary information that is protected by copyright. All rights
are reserved. Neither the documentation nor software may be copied, photocopied,
reproduced, translated, or reduced to any electronic medium or machine-readable form, in
whole or in part, without the prior written consent of dSPACE GmbH.
Copyright 1998 - 2011 by:
dSPACE GmbH
Rathenaustrae 26
33102 Paderborn
Germany
This publication and the contents hereof are subject to change without notice.
AutomationDesk, CalDesk, ConfigurationDesk, ControlDesk, SystemDesk and TargetLink
are registered trademarks of dSPACE GmbH in the United States or other countries, or
both. Other brand names or product names are trademarks or registered trademarks of
their respective companies or organizations.

Contents

About This Document

7
Document Symbols and Conventions ....................................... 7
Accessing Online Help and PDF Files......................................... 8
Related Documents .................................................................. 9

Introduction to MLIB/MTRACE

11
Basics of MLIB/MTRACE.......................................................... 11

Installation

13
System Requirements ............................................................. 13
Installation Process ................................................................. 14
Demo Files ............................................................................. 15

Getting Started

17
Preparing Real-Time Applications ........................................... 18
Typical Calling Sequences of MLIB/MTRACE Commands.......... 19

Using MLIB/MTRACE for Multiprocessor

Systems

23
Basic Principles of Using Multiprocessor Systems .................... 24
Data Capture With Different Sampling Periods or
Asynchronous ........................................................................ 27

MLIB/MTRACE Functions

31
General Information on the Function Reference............................... 32
Naming Conventions.............................................................. 32
MLIB/MTRACE Function Reference .................................................. 33
CaptureState.......................................................................... 34
FetchData .............................................................................. 36
Get ........................................................................................ 40
GetApplInfo ........................................................................... 42
GetBoardInfo ......................................................................... 44
GetMapVar ............................................................................ 45
GetTrcVar ............................................................................... 47
Intrpt ..................................................................................... 51

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Contents

IsApplRunning ....................................................................... 52
LoadFile ................................................................................. 53
LoadVarFile ............................................................................ 55
LockProgram.......................................................................... 56
Multiprocessing ..................................................................... 57
Read ...................................................................................... 57
SearchMap ............................................................................ 59
SearchTrc ............................................................................... 60
SelectBoard............................................................................ 61
Set......................................................................................... 62
SetVarProperty ....................................................................... 67
StartCapture .......................................................................... 69
StopCapture .......................................................................... 71
TriggerState ........................................................................... 72
UnlockProgram ...................................................................... 73
Write ..................................................................................... 73

MLIB/MTRACE Programming

75
Programming Instructions ............................................................... 76
Multiple MLIB/MTRACE Data Captures in One Real-Time
Application ............................................................................ 76
Locking the Real-Time Program.............................................. 77
Accessing Matrices with MLIB/MTRACE........................................... 78
Built-in Simulink Blocks .......................................................... 79
2-D Look-Up Tables................................................................ 79
S-Functions ............................................................................ 79
mlib_1dlookup ...................................................................... 80
mlib_2dlookup ...................................................................... 83
mlib_matrix............................................................................ 86
MLIB/MTRACE Programming Examples ........................................... 88
MLIB/MTRACE with Handcoded Applications ......................... 89
MLIB/MTRACE with dSPACE RTI ............................................. 93
Additional MLIB/MTRACE Programming Examples.................. 96
MLIB/MTRACE with dSPACE RTI-MP ....................................... 97

File Reference

101
MAT File Generated by MLIB/MTRACE.................................. 101

Utilities

103
Basics on the TrcView Utility ................................................. 103
How to Use TrcView............................................................. 104

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Contents

Limitations

107
Accessing Variables with Read/Write Functions..................... 108
Number of Applications Simultaneously Accessing the
Hardware............................................................................. 108
Number of Platforms Simultaneously Accessed by
MLIB/MTRACE...................................................................... 108

Troubleshooting

109
MLIB/MTRACE Error Messages.............................................. 109
Invalid Variable Connections When Migrating to MATLAB R2011a
125

Appendix

127
MLIB/MTRACE Benchmarks .................................................. 127
Impact of MLIB/MTRACE on Execution Time ......................... 130

Index

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

133

May 2011

Contents

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

About This Document

This document contains detailed information on the MATLAB-dSPACE

Interface Libraries.
Where to go from here

Information in this section

Document Symbols and Conventions

Accessing Online Help and PDF Files

Related Documents

Document Symbols and Conventions

Symbols

The following symbols may be used in this document.

Indicates a general hazard that may cause personal injury
of any kind if you do not avoid it by following the
instructions given.
Indicates the danger of electric shock which may cause
death or serious injury if you do not avoid it by following
the instructions given.
Indicates a hazard that may cause material damage if
you do not avoid it by following the instructions given.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

About This Document

Indicates important information that should be kept in

mind, for example, to avoid malfunctions.
Indicates tips containing useful information to make your
work easier.

Naming conventions

The following abbreviations and formats are used in this document:

%name%
Names enclosed in percent signs refer to environment
variables for file and path names, for example, %DSPACE_PYTHON25%
folder of the Python installation.
< > Angle brackets contain wildcard characters or placeholders for
variable file and path names, etc.
Precedes the document title in a link that refers to another
document.
ST M

Indicates that a link refers to another document, which is available

in dSPACE HelpDesk.

Accessing Online Help and PDF Files

Objective

After you install your dSPACE software, the documentation for the
installed products is available as online help and Adobe PDF files.

Online help

You can access the online help, dSPACE HelpDesk, as follows:

Windows Start menu Click Start Programs <product name>
dSPACE HelpDesk (<product name>)
If there is no product-specific dSPACE HelpDesk:
Click Start Programs dSPACE HelpDesk.
Context-sensitive
dSPACE software.

Press the F1 key or click the Help button in the

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

PDF Files

About This Document

You can access the PDF files as follows:

dSPACE HelpDesk
document:

Click the PDF link at the beginning of a

Related Documents
Below is a list of documents that you are recommended to also read
when working with MLIB/MTRACE.
Information in other documents
Interface libraries
MLIB/MTRACE MATLAB-dSPACE Interface Libraries

Contains detailed reference information and instructions on

the experiment features of MLIB/MTRACE.

CLIB C Interface Library

Contains detailed reference information on the C Interface
Library CLIB, which contains C functions to access the
dSPACE processor and controller boards from the host PC.

ControlDesk
ControlDesk Next Generation
AutomationDesk
RTI and RTLib

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

10

About This Document

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Introduction to MLIB/MTRACE

Basics of MLIB/MTRACE
The MATLAB-dSPACE Interface Libraries give you access to dSPACE
real-time processor hardware from the MATLAB workspace.
MLIB/MTRACE functions can be called from the MATLAB Command
Window or from MATLAB M files. Thus the powerful numerical tools
running under MATLAB can be used in combination with
MLIB/MTRACE for:
n

Analyzing real-time data

Test automation

Optimizing control algorithms

MLIB/MTRACE provides basic functions for reading and writing data to

dSPACE processor boards and other functions for generating
interrupts, setting the processor state, and getting processor status
information.
MLIB/MTRACE provides real-time data capture capabilities, including
the following features:
n

Free-running or level-triggered mode

Pre- and post-trigger

Selection between single_shot and continuous mode

Simultaneous start of multiple data captures

Distinction between double, float and integer (signed and

unsigned) variables

Adjustable trace capture sampling rate (downsampling)

Direct data transfer to the MATLAB workspace

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

11

Introduction to MLIB/MTRACE

Data storage on the hard disk drive in continuous mode (optional)

Specification of data capture parameters by property/values pairs.

MLIB/MTRACE functions are used to modify parameters online and to

send sequences of test data to real-time applications. To get access to
any variable on a real-time processor with MLIB/ MTRACE, you do not
need to know its address because there are several functions that
return the required addresses when the symbolic name of a variable is
specified.
MLIB/MTRACE ideally complements the software environment, which
consists of MATLAB, Simulink, Real-Time Workshop, and the dSPACE
tools RTI and ControlDesk 3.x / ControlDesk Next Generation.
MLIB/MTRACE is not only used for programs generated by RTI, but
also for any other application program written in C.
The following illustration shows how a real-time processor is accessed
from MATLAB M files via MLIB/MTRACE.

12

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Installation

To get help on the installation process of MLIB/MTRACE, refer to the

list below:
n

To check the required dSPACE boards and the versions currently

available, refer to System Requirements on page 13

For information about the created directories, refer to Installation

Process on page 14

For M files information about the copied demo files, refer to Demo
Files on page 15

System Requirements
Host PC and operating system MLIB/MTRACE runs on IBMPC/AT or compatible machines under the following operating systems:
n

Windows XP

Windows Vista

Windows 7

MATLAB Release The functions of the MATLAB-dSPACE Interface

and Trace Libraries are available for the MATLAB versions that are
supported by the current dSPACE Release.
dSPACE processor boards
MLIB/MTRACE currently supports the
following dSPACE processor boards:
n

DS1103

DS1104

DS1401

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

13

Installation

DS1005

DS1005 multiprocessor systems

DS1006

DS1006 multiprocessor systems

dSPACE software products

dSPACE software products:

Related topics

MLIB/MTRACE requires the following

Board

RTLIB

DS1103

RTLIB1103

DS1104

RTLIB1104

DS1401

RTLIB1401

DS1005

RTLIB1005

DS1006

RTLIB1006

When using RTI or RTI-MP for multiprocessor systems,

you need the RTI version from the related dSPACE
Release.

When MLIB/MTRACE is used to access a dSPACE

processor board via network, make sure that the
TCP/IP Network software is installed (delivered with the
Windows operating system).

It is recommended to install all the software related to

MLIB/MTRACE from the same dSPACE DVD.

Basics
Installation on page 13

Installation Process
The installation program performs all necessary steps automatically.
During installation you will only be asked for the MATLAB root folder.
When MLIB/MTRACE is installed, files are copied to
%DSPACE_ROOT%\Matlab\Mlib.
In addition, the file %DSPACE_ROOT%\Matlab\Local\Dspacerc.m is copied
to \Matlab\Toolbox\Local.

14

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Installation

A call to dspacerc.m is then appended to the MATLAB startup file

Matlabrc.m. As a result, the dSPACE configuration file dspace_setup.m
residing in %DSPACE_ROOT%\Matlab\Local is executed each time
MATLAB is started.
All necessary configurations for dSPACE products related to MATLAB
are set automatically by dspace_setup.m.
Related topics

Basics
Installation on page 13

Demo Files
The files necessary to run the MLIB/MTRACE demos supplied by
dSPACE are installed in the following folders:
n

%DSPACE_ROOT%\demos\dsXXXX\mlib,

%DSPACE_ROOT%\demos\dsXXXXMP\dualXXXX\mlib.
dsXXXX denotes the respective processor board name, for
example, DS1103, DS1104, DS1401, DS1005 or DS1006.

Related topics

Basics
Installation on page 13

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

15

16

Installation

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Getting Started

A good way to become familiar with MLIB/MTRACE is to take a look at

the demo examples installed in the following folders
n

%DSPACE_ROOT%\demos\dsXXXX\mlib for MLIB/MTRACE

%DSPACE_ROOT%\demos\dsXXXXMP\dualXXXX\mlib for
MLIB/MTRACE-MP.

The sample M files can also be used as templates for new applications.
n

Before using MLIB/MTRACE functions, you must select a real-time

processor board for use with MLIB/MTRACE. The processor board
must be already registered in ControlDesk 3.x's Platform Manager
or in ControlDesk Next Generation's Platform/Device Manager.

Board info

To obtain a list of the boards currently registered in ControlDesk 3.x's

Platform Manager or in ControlDesk Next Generation's
Platform/Device Manager, include the command boards_info =
mlib('GetBoardInfo') in the MATLAB window. The boards_info will
then contain basic information about registered boards. If boards_info
is empty, no boards are registered.

mlibini.m

It is recommended to use the mlibini.m M file supplied with

MLIB/MTRACE to select the first processor board registered in
ControlDesk 3.x's Platform Manager or in ControlDesk Next
Generation's Platform/Device Manager. For most MLIB/MTRACE
applications, it is sufficient to include the line
% select the first processor board registered
% in the Platform Manager for use with
% MLIB/MTRACE
mlibini

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

17

Getting Started

at the beginning of M files that call MLIB/MTRACE functions.

Connection via bus or
network

MLIB/MTRACE can access locally installed boards via a bus connection,

and other boards via a network connection. The choice of bus or
network connection is automatic. The first time an MLIB/MTRACE
function is called, the last connection that was set by Control Desks
Platform Manager is used.

MLIB/MTRACE DLL file

MLIB/MTRACE is implemented as a MATLAB MEX DLL file, which is

loaded into the PC memory the first time it is called during a MATLAB
session. MLIB/MTRACE remains loaded until a clear mex, clear
functions, clear dsmlib, or clear all command is invoked, or until
MATLAB is closed. If you clear MLIB/MTRACE, you must call
SelectBoard before proceeding with other MLIB/MTRACE functions.
The following topics contain further information on:
n

Preparing Real-Time Applications on page 18

Typical Calling Sequences of MLIB/MTRACE Commands on page 19

Preparing Real-Time Applications

MLIB/MTRACE needs special services to be executed in the real-time
application, to perform Read/Write operations and carry out real-time
data acquisition:
n

host_service(hostsvc_no, ts): where the hostsvc_no can be an

integer number between 1 and 31

master_cmd_server()

RTI generates the necessary function calls of host_service() and

master_cmd_server() automatically. If you program hand-coded
applications, refer to the RTLib Reference of your processor board.
host_service (hostsvc_no,
ts)

This service, called the data capture service in this manual, is

responsible for real-time data acquisition. It collects one set of data per
call and writes it to a buffer reserved in the real-time processor
memory, called the trace buffer below. Thus, the rate at which the
data is sampled depends on the frequency at which this service is
executed.
The second parameter, ts, is a pointer to the time stamp structure that
represents the absolute time when the set of trace data was collected.

18

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Getting Started

For further information, refer to Data Capture Block (

RTI and RTIMP Implementation Reference) or host_service in the RTLib Reference
of your processor board.
ST M

master_cmd_server()

This service reads and evaluates commands sent by MLIB/MTRACE to

the real-time program. It has to be called in the background loop.
You can only read and write data from the local memory of the
DS1103 and the DS1401 boards, local variables residing on the
DS1104 board, and the cached memory of DS1005 and DS1006, if
this service is executed.

Example

while(1)
{
...
master_cmd_server();
...
}

Related topics

Basics
Getting Started on page 17

Typical Calling Sequences of MLIB/MTRACE Commands

MLIB/MTRACE functions always relate to the most recently selected
board. Repeated selection of the same board has no effect.
For example, if a DS1005 board is used, the typical sequence of
MLIB/MTRACE commands is as follows:

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

19

Getting Started

Example

This example shows how to modify parameters online:

% select DS1005 for use with MLIB/MTRACE
mlib('SelectBoard', 'DS1005');
% obtain the descriptors of selected variables
var_names = {'x_par'; 'y_par'};
xy_desc = mlib('GetMapVar', var_names);
% read data from DS1005
[x_val, y_val] = mlib('Read', xy_desc);
% display the results
fprintf("x_par = %f\ny_par = %f\n", x_val, y_val);
% write new value to DS1005
data = {0.125; 4.5};
mlib('Write', xy_desc, 'Data', data);
This example shows the online real-time data acquisition:
% connect to processor board DS1005
mlib('SelectBoard', DS1005);
% find address of variable in processor memory
x_var = mlib('GetTrcVar', 'User Variables/x');
% configure MLIB for data capture:
% select variable for trace capture,
% configure the time frame for data acquisition
mlib('Set', ...
'TraceVars', x_var, ...
'Delay', 0, ...
'NumSamples', 500);
% start data acquisition;
% default service number for DS1005 is 1
mlib('StartCapture');
% wait until capture is finished
while mlib('CaptureState')~=0, end;
% transfer real-time data to MATLAB workspace
x_data = mlib('FetchData');
% plot results
plot(x_data');

For further information, refer to the demo applications in

%DSPACE_ROOT%/demos/dsxxxx/mlib.
For a detailed description of the MLIB/MTRACE commands, refer to
MLIB/MTRACE Functions on page 31.

20

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Related topics

Getting Started

Basics
Getting Started on page 17

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

21

22

Getting Started

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Using MLIB/MTRACE for

Multiprocessor Systems

MLIB/MTRACE supports access to variables of real-time applications

running on dSPACE single-processor and multiprocessor systems.
Typical dSPACE multiprocessor systems are:
n

Two or more DS1005 boards linked by the GigaLink connection

Two or more DS1006 boards linked by the GigaLink connection

For basic information, refer to Basic Principles of Using Multiprocessor

Systems on page 24.
For information on data acquisition on multiprocessor systems, refer to
Data Capture With Different Sampling Periods or Asynchronous
on page 27.
For further information about dSPACE multiprocessor
systems, refer to the RTI and RTI-MP Implementation
Guide, DS1005 RTLib Reference and DS1006 RTLib
Reference.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

23

Using MLIB/MTRACE for Multiprocessor Systems

Basic Principles of Using Multiprocessor Systems

Each member board of a multiprocessor system based on DS1005 or
DS1006 is equipped with a host interface and can be accessed directly
by MLIB/MTRACE in single-processor mode. However, a single
MLIB/MTRACE instance can operate only with one currently selected
board. To access all boards simultaneously in single-processor mode,
you must start several instances of MATLAB.
The other way is to use MLIB/MTRACE in multiprocessor mode. After
the Multiprocessing function is invoked, all global variables residing on
any processor that is part of the selected multiprocessor system are
accessible with MLIB/MTRACE. Read/write access is similar to
read/write access in single-processor mode. Appreciable differences
occur in the data capturing (tracing) mechanism. A distributed tracing
method is used for multiprocessor systems. The graphic below explains
this idea:

In this method, variables are traced locally on each processor board.

MLIB/MTRACE accesses them separately and tries to give the data a
common time frame. This strategy has the following requirements:
n

24

Information on the time stamp when the data set was collected
must be available.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Using MLIB/MTRACE for Multiprocessor Systems

Data acquisition on the single processors must be started

synchronously.

The first condition is fulfilled if the time stamps at which the data has
been collected are also traced. The second condition is met by using a
trigger mechanism: One of the processors is used to synchronize the
start of data acquisition on all the other processors. This processor is
called the synchronizing processor below. If triggered mode has been
selected, the synchronizing processor is the processor on which the
trigger variable resides. For non triggered mode, any processor can be
the synchronizing processor. In this case MLIB/MTRACE chooses the
synchronizing processor according to certain rules, which are not of
importance for an understanding of the distributed tracing method.
The synchronized processors always run in triggered mode.
The trigger signal which starts data acquisition on these processors is a
32-bit word sent by the synchronizing processor.
Assuming triggered mode has been selected, as soon as the trigger
event is recognized by the synchronizing processor, a trigger word is
changed according to predefined rules and sent to the synchronized
processors. The synchronized processors react to the change in the
trigger word and start data acquisition.
Data Capture in
Continuous Mode

If the data is captured in continuous mode, the start of data

acquisition on the single processors is not synchronized. In this case
MLIB/MTRACE returns the data to the MATLAB workspace in the form
of a vector of structures: one structure for each processor. Since the
time stamps are traced throughout the capture process, it is possible to
give the data a common time frame.

Service Group

As shown in the example, data acquisition on the master processor is

carried out by host service number 1. On the slave processor, it is
carried out by host service number 2. These two services form the
service group. It is possible to build several service groups for the
multiprocessor system, for example: CPU master, service number 1;
CPU slave, service number 1; or CPU master, service number 2; CPU
slave, service number 2, etc.
To carry out data acquisition in single processor mode,
you must specify a service number; to carry out data
acquisition in MP1005 or MP1006 mode, you must
specify a service group.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

25

Using MLIB/MTRACE for Multiprocessor Systems

The service group is identified by its name. To specify the service group
as above, the following MLIB/MTRACE functions must be called
(assuming Task1 on the master is called with 0.00012 sampling period,
and Task2 on the slave is called with a sample period of 0.00036)
mlib('Set', 'ServiceGroup','Master1_Slave2_Group',....
'CPU','Master',...
'Service', 1,....
'StepSize', 0.00012);
mlib('Set','ServiceGroup','Master1_Slave2_Group',....
'CPU','Slave',....
'Service',2,....
'StepSize', 0.00036);
Each time the StartCapture function is called with this name as an
input parameter: mlib('StartCapture','Master1_Slave2_Group);, the
data capture is carried out by service 1 on the master processor and by
service 2 on the slave processor.
The TRC files generated by RTI-MP contain entries describing the
available host services in the following form:
sampling_period[1]
{
value:
0.0
alias:
"AsynchronousTask"
}
The index of the sampling_period vector denotes the host_service
number, the service name is specified under alias and the step size (0.0
= asynchronous task) under value. If TRC files exist, MLIB/MTRACE
combines services that are on different processors but have the same
name to form one service group. The name of the service group
corresponds to the name of the services. You do not have to specify
the service group as shown above. However, you can use this service
group, for example, with StartCapture function.
Related topics

Basics
Data Capture With Different Sampling Periods or Asynchronous on page 27

26

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Using MLIB/MTRACE for Multiprocessor Systems

Data Capture With Different Sampling Periods or

Asynchronous
If the services carrying out data acquisition on the individual member
processors are called with the same sampling period, in most cases
MLIB/MTRACE is able to find a common time axis for data collected on
different CPUs. Therefore it makes no difference whether the trace
interval has been specified by the number of samples to be collected
or by the duration of data acquisition. Since the sampling period is the
same for all processors, the N samples on processors P1, P2, P3 etc.,
corresponds to (N * sampling period) T seconds on processor P1, P2,
P3 etc. and T seconds on processor P1 correspond to
(T/sampling_period) N samples on processor P2, P3.
However, if the services carrying out data acquisition on the individual
member processors are called with different sampling periods or
asynchronously, MLIB/MTRACE returns different results depending on
how the trace interval is defined. The examples below explain the
differences.
The following assumptions apply to both examples:
n

Sampling period on processor P1 = tp

Sampling period on processor P2 = 3 * tp

Synchronizing processor: P1

1. The trace interval is specified by the duration of the data

acquisition:
mlib('Set',...
'Start', t0,....
'Stop',tN,...
);

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

27

Using MLIB/MTRACE for Multiprocessor Systems

The graph below shows the result of the data acquisition:

Only points which lie exactly on or after the lower time limit, and
points which lie before or exactly on the upper time limit, are
collected.
The number of samples collected with processor P1 differs from the
number of samples collected with processor P2.
2. The trace interval is specified by the number of samples to be
collected:
mlib('Set',...
'Delay', -2,....
'NumSamples',5,...
);

28

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Using MLIB/MTRACE for Multiprocessor Systems

The graph below shows the result of data acquisition:

Since the samples are collected with a different sampling period on

each processor, the data is spread over different time intervals.

Related topics

Basics
Basic Principles of Using Multiprocessor Systems on page 24

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

29

30

Using MLIB/MTRACE for Multiprocessor Systems

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

This topic presents a brief overview of the MLIB/MTRACE functions. All

functions are listed alphabetically and described in detail with relevant
examples:
n

General Information on the Function Reference on page 32

MLIB/MTRACE Function Reference on page 33

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

31

MLIB/MTRACE Functions

General Information on the Function

Reference
Naming Conventions
In the chapter MLIB/MTRACE Function Reference on page 33, the
following conventions for parameter names are used:
var Variable descriptor a structure containing information about a
variable as returned by the following address evaluation functions:
Function

Description

var.name

Name of the variable

var.cpu

Name of the CPU on which the variable resides

var.type

Data type

var.addr

Variable base address

var.offset

Offset of the first element from the base address

var.length

Number of elements

var_vec
ival
uival

A vector of structures described above

Matrix of integer values in the MATLAB workspace

Matrix of unsigned integer values in the MATLAB workspace

fval Matrix of single-precision floating-point values in the MATLAB

workspace
dval
Matrix of double-precision floating-point values in the
MATLAB workspace
data A cell array in the MATLAB workspace, where each cell can be
a scalar, a vector, or a matrix of any data type, supported by a selected
processor board

32

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

MLIB/MTRACE Function Reference

The following table gives an overview of the implemented
MLIB/MTRACE functions.
Purpose

Refer to

Administrative Functions
To get information about a real-time application

GetApplInfo on page 42

To get information about boards currently registered in ControlDesk 3.x's Platform Manager or
in ControlDesk Next Generation's Platform/Device Manager.

GetBoardInfo on page 44

To select a real-time processor board for use with MLIB/MTRACE

SelectBoard on page 61

To check if a real-time application is running

IsApplRunning on page 52

To set a lock on the real-time program

LockProgram on page 56

To select single-processor or multiprocessor mode

Multiprocessing on page 57

To release a lock on the real-time program

UnlockProgram on page 73

To load a variable description file

LoadVarFile on page 55

Address Evaluation Functions

To obtain the descriptor for a global variable from the linker map file

GetMapVar on page 45

To obtain the descriptor for a global variable specified within the TRC file

GetTrcVar on page 47

To search the linker map file for variables whose names match the specified regular expression

SearchMap on page 59

To search the TRC file for variables whose names match the specified regular expression

SearchTrc on page 60

To set the properties stored in a variable descriptor

SetVarProperty on page 67

Read/Write Functions
To read data from a real-time processor board

Read on page 57

To write date to a real-time processor board

Write on page 73

Interrupt Functions
To generate an interrupt on a real-time processor

Intrpt on page 51

Data Acquisition Functions

To start a data capture on the processor board

StartCapture on page 69

To stop a data capture on the processor board

StopCapture on page 71

To get trigger status information

TriggerState on page 72

To get capture status information

CaptureState on page 34

To transfer data from the processor board memory to the MATLAB workspace

FetchData on page 36

To set data acquisition options

Set on page 62

To get data acquisition options

Get on page 40

Functions Supporting MLIB/MTRACEGenerated MAT Files

To load selected data from a MAT file into the MATLAB workspace

Processing mode

LoadFile on page 53

If the syntax or description depends on the processing mode, the

mode is indicated:
n

SP = single-processor mode

MP = multiprocessor mode with DS1005-based or DS1006-based

multiprocessor systems

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

33

MLIB/MTRACE Functions

Information on how to select a processing mode is given in the

description of the Multiprocessing function.
n

To get online help for MLIB/MTRACE functions, type

help mlib or help <function_name> in the MATLAB
Command Window.

To get HTML online help, open the dSPACE HelpDesk

or type mlib_help in the MATLAB Command Window.

If you are using MLIB/MTRACE with a multiprocessor

system based on DS1005 or DS1006, we strongly
recommend that you familiarize yourself with the topic
Using MLIB/MTRACE for Multiprocessor Systems
on page 23 before reading the reference part for the first
time.

CaptureState

34

Purpose

To get capture status information.

Syntax SP

state = mlib('CaptureState')
state = mlib('CaptureState', services)

Syntax MP

state = mlib('CaptureState')
state = mlib('CaptureState', serviceGroups)

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Description

MLIB/MTRACE Functions

When used without an input parameter, this function returns the

current state of the data capture process performed by the default
service for SP or the service group for MP1005 or MP1006.
In single-processor mode, the default service is service
number 1.
In multiprocessor mode, the defaults are as follows:
n

If the TRC file exists:

o

The Service group collecting services with the same

name as the service called with the fastest rate on
the master processor is used as the default.
If all services are called asynchronously, the service
group collecting services with the same name as the
service first found on the master processor is used
as the default.

If the TRC file does not exist:

o

The service group named ServiceName1 is used as

the default. This collects all services with the
number 1 on member processors.

Otherwise, you must specify the service/service group

services/serviceGroups in the parameter.
The return values indicate the current state of the data capture:
Value

Description

The buffer is empty.

The buffer is neither empty nor full (data acquisition is in

progress).

The buffer is full (data acquisition is finished).

If multiple data captures performed by different data capture

services/services groups were started simultaneously, the states of all of
them by calling the CaptureState function with the additional input
parameter services/serviceGroups. This must be a vector containing
the numbers of services or a cell array containing the names of service
groups carrying out data acquisition. In this case, state is a vector
whose elements indicate the state of the corresponding capture
process.
Result

The state of the capture process is returned.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

35

MLIB/MTRACE Functions

Example

Example 1:
% wait for a data acquisition to finish
while mlib('CaptureState')~=0, end
disp('Capture finished.');
Example 2:
% wait for a data acquisition performed
% by services 1 and 2 to be finished
while any(mlib ('CaptureState', [1 2])) ~= 0, end
disp('Both captures finished.');
Example 3:
% wait for a data acquisition performed
% by the service groups HostService1 and HostService2
while any(mlib ('CaptureState', ...
{HostService1; HostService2})) ~= 0, end
disp('Both captures finished.');

Related topics

References
StartCapture on page 69
TriggerState on page 72

FetchData

36

Purpose

To fetch data from the real-time processor buffer and transfer it to the
MATLAB workspace.

Syntax

traced_data = mlib('FetchData')
traced_data = mlib('FetchData', [count ,], ...
property_name, property_value, ...

Description

Depending on the settings made in the MLIB/MTRACE function Set,

the FetchData function returns the traced data in one of the following
formats:
n

Matrix

Matrix with time

Structure

Structure with time

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

Matrix The number of rows corresponds to the number and length

of variables that have been traced. Each row represents a time series of
a variable (or of one vector element if the variable is a vector). Each
column represents all variables at the same specific time. The rows in
the output matrix are assigned to the traced variables in the order in
which the variables were selected with the TraceVars property.
Matrix with time
The format is the same as the matrix format,
except that the first row represents time stamps at which the data was
collected. The number of rows therefore corresponds to the number
and length of variables that have been traced + 1.
Structure with time
n

.time

.signals

This structure has the following fields:

The .time field contains a [N x 1] vector of the time stamps, where N is

the number of collected samples (a sample contains a set of values of
all the selected variables collected in one sampling step). The .signals
field contains a vector of substructures, each of which corresponds to
one trace variable.
Each substructure has two fields:
n

.values

.label

The .label field specifies the variable name, the .values field contains a
[N x variable length] matrix, with values of the corresponding variable
at time stamps specified in the .time field.
Structure
The format is the same as the structure with time, except
the .time field remains empty.
Structures for
multiprocessor systems

If the structure format has been selected in the Set function, the
FetchData function returns a vector of structures as described above,
each of which corresponds to one processor with traced variables. To
indicate which structure corresponds to which processor, an additional
.name field is added to hold the processor name.
Note that each processor has its own time vector.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

37

MLIB/MTRACE Functions

The data can be fetched from the processor memory even if the
capturing process is still in progress. If the count parameter is omitted,
all the data already captured and written to the trace buffer is
returned. It is recommended to call the CaptureState function
beforehand to ensure that the data capturing process is finished. The
count parameter can be used to fetch the exact number of samples. In
this case the number of columns in the output matrix and the number
of rows in the .values matrices in the output structure equals the
count. If there is not enough data to complete the output
matrix/structure (for example, the FetchData function was already
called several times and the remaining number of samples to be read
from the trace buffer is smaller than count), zeros are written instead.
By default, MLIB/MTRACE assumes that the data to be fetched was
captured by default capture service (SP) or service group (MP1005 or
MP1006).
If the data capture was started in the continuous mode of
MLIB/MTRACE with the StreamToDisk option set to ON
(see the Set function), each call of the FetchData function
also streams the data into a MAT file on the hard disk.
The data is always stored in the MAT file (matrix/matrix
with time), independently of the output format of the
FetchData function. The structure and structure with time
formats are not supported.
The possible properties and their descriptions are as follows:
SP

Service If a service other than the default is used, you must specify
its number via this property. This property can be used only in singleprocessor mode.

MP

ServiceGroup

If a service group other than the default is used, its

name must be explicitly specified by via this property.

This property is used only in multiprocessing mode with multiprocessor
systems based on DS1005 or DS1006.
StartSample By default, MLIB/MTRACE reads the data from the
trace buffer, beginning with the current MLIB/MTRACE read pointer.
The MLIB/MTRACE read pointer is the position in the trace buffer up to
which data was fetched into the MATLAB workspace. To change this
order, the StartSample property.

38

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

Result

The data captured for each variable selected with the Set function is
transferred into the MATLAB workspace.

Example

Example 1:

uy_var = {'User Variables/u'; 'User Variables/y'}

uy_addr = mlib('GetTrcVar',uy_var)
% capture 2000 samples of each variable by service 2
mlib('Set', 'Service', 2, ...
'TraceVars', uy_addr, ...
'NumSamples', 2000);
mlib('StartCapture', 2);
% wait for a data acquisition to finish
while mlib('CaptureState', 2) ~= 0, end
% fetch all data for variables u and y
matdata = mlib('FetchData' , 2);
plot(matdata');
Example 2:
uy_var = {'User Variables/u'; 'User Variables/y'}
uy_addr = mlib('GetTrcVar',uy_var)
% capture 2000 samples of each variable by
% service 1(default)
mlib('Set', 'TraceVars', uy_addr, ...
'NumSamples', 2000);
mlib('StartCapture');
% fetch data while capture process is still running
while mlib('CaptureState') ~= 2
data = mlib('FetchData', 200)
% edit data etc.
...
end
Related topics

References
Set on page 62

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

39

MLIB/MTRACE Functions

Get
Purpose

To get current data acquisition options.

Syntax SP

info = mlib('Get')
info = mlib('Get', property_name)
info = mlib('Get', 'Service', service_number)
info = mlib('Get', 'Service', service_number, ...
property_name)

Syntax MP

info = mlib('Get')
info = mlib('Get', property_name)
info = mlib(Get,ServiceGroup, service_group_name)
info = mlib(Get,ServiceGroup, service_group_name, ...
property_name)

Result

The current MLIB/MTRACE settings, such as the time frame, trigger

options, and variables to be traced, are returned. By default, if no
service number/service group name or property name is specified, the
function returns information about all the settings of the data capture
process performed by the default service (SP) or service group
(MP1005 or MP1006).
Another service number/service group name can be selected with the
service_number/service_group_name parameter.
If any property is explicitly specified by its name, the info structure
contains this single field only.
Depending on the selected processing mode, the returned info
structure can have the following fields.
SP Mode

40

MP Mode

.Service

.ServiceGroup

.Trigger

.Trigger

.TriggerVariable

.TriggerVariable

.TriggerLevel

.TriggerLevel

.TriggerEdge

.TriggerEdge

.StepSize

.Downsampling

.Downsampling

.TimeStamping

.TimeStamping

.UseAbsoluteTime

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

SP Mode

MP Mode

.UseAbsoluteTime

.TraceVars

.Start

.AcquisitionMode

.Stop

.StreamToDisk

MLIB/MTRACE Functions

.Delay

.Output format

.NumSamples

.AutoClearBuffer

.TraceVars

.InterpolateBoundaryPoints

.AcquisitionMode

.CPU

.StreamToDisk

.TimeOutValue

.FileName
.OutputFormat
.AutoClearBuffer
.TimeOutValue
.CPU
The .CPU field is a vector of structures, where each structure
contains information about settings specified for one CPU. Each
structure has the following subfields:
n

.name (name of the CPU)

.service (number of the data acquisition service on the current CPU,

which belongs to the specified service group)

.stepSize (the sampling period of the task where the data

acquisition service is called)

.start (the start of the trace interval for which the data is to be
collected, stated in the units of the step size)

.stop (the end of the trace interval for which the data is to be
collected, stated in the units of the step size)

.numSamples (length of the trace interval specified in the number

of samples)

.delay (delay related to the trigger event specified in the number of

samples)

.fileName (name of the MAT file to store the data collected in

StreamToDisk mode)

For further information on the fields and the returned info structure,
refer to the Set function.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

41

MLIB/MTRACE Functions

Example
% display the settings of the data capture process
% performed by service 2
mlib('Get', 'Service', 2)
...
% display the value of the NumSamples property of the data
% capture process performed by service 1(default)
mlib('Get', 'NumSamples')
Related topics

References
Set on page 62

GetApplInfo

42

Purpose

To get information about the real-time application.

Syntax

info = mlib('GetApplInfo')
info = mlib('GetApplInfo', 'property_name', property_value...)

Result

By default, if SP mode is selected, GetApplInfo returns information

about the real-time application currently running on the selected
processor board.

Description

The returned information is stored as a structure containing the

following fields:
Field

Description

.name

Name of the real-time program

.date

Creation date of the real-time program

.board

Name of the board on which the program is running

.type

Type of the board on which the program is running

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

If the MP mode is selected, the info parameter is an M-by-1 vector of

structures, where M equals the number of CPUs in the currently
loaded MP application. Each structure contains information about the
program running on the corresponding CPU as described above plus
the following fields:
Field

Description

.master_cpu

Name of the master CPU

.master_board

Name of the board on which the master CPU

resides

.reachable

An integer value that indicates if the global

variables residing on the CPU are accessible
via MLIB/MTRACE or not
0: the variables are not accessible
1: the variables are accessible

The permissible property/value pairs are listed below:

SDFFile Name of the SDF file created by RTI-MP. If this property is
defined (SP or MP mode), the GetApplInfo function returns
information about the MP application specified in the SDF file,
independently of the application currently loaded on the processor
board(s).
CPU Name of the CPU for which the info structure is to be
returned. In this case, info is a simple structure.
Example

This example refers to single-processor systems (SP)

% which application is running on the selected board?

appl = mlib('GetApplInfo');
disp(['The current real-time application is' appl.name]);
This example refers to multiprocessor systems (MP)
% Get the information which program is running on the
% CPU "MASTER"
appl = mlib('GetApplInfo', 'CPU', 'Master');
fprintf('The program running on the CPU MASTER is: %s', ...
appl.name);
Related topics

References
IsApplRunning on page 52

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

43

MLIB/MTRACE Functions

GetBoardInfo
Purpose

To get information about boards currently registered in

ControlDesk 3.x's Platform Manager or in ControlDesk Next
Generation's Platform/Device Manager.

Syntax

board_info = mlib('GetBoardInfo')

Description

This result is returned as an M-by-1 vector of structures, where M is

equal to the number of boards currently registered in
ControlDesk 3.x's Platform Manager or in ControlDesk Next
Generation's Platform/Device Manager and where each structure
contains the following fields:
Field

Description

.name

Name of the board

.type

Type of the board

.index

Board's registration index at the device driver

This function can be invoked at any time, which means without calling
the SelectBoard function first.
Example

% Get the board info

board_info
= mlib('GetBoardInfo');
% get the number of boards currently registered
num_of_board = size(board_info,1);
% display info
printf('Registered processor boards\ttype\tindex\n');
for i = 0:num_of_board,
fprintf('\t\t%s\t%s\t%d\n', board_info(i).name, ...
board_info(i).type, board_info(i).index);
end
% select first registered processor board
mlib('SelectBoard', baord_info(1).name);

Related topics

References
SelectBoard on page 61

44

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

GetMapVar
Purpose

To obtain the descriptor for a global variable.

Syntax

var_vec = mlib('GetMapVar', variable_names)

var_vec = mlib('GetMapVar', variable_names, ...
'property_name', property_value, ...)
[var1, ..., varN] = mlib('GetMapVar', variable_names)
[var1, ..., varN]= mlib('GetMapVar', variable_names, ...
'property_name', property_value, ...)

Description

This result returns descriptors for the variables specified in

variable_names. The returned variables must be declared as global
symbols in the currently running real-time application.
The variable_names parameter must be an M-by-1 cell array, where M
is the number of variables whose descriptor are to be obtained and
where each cell contains a symbolic name for one variable. var_vec is
therefor an M-by-1 vector of descriptors. If M=1 (a descriptor for only
one variable is to be obtained), the function also accepts a string as an
input parameter. var_vec is then a simple structure.
If multiple output arguments are specified (var1,..., varN), the
descriptors corresponding to the specified variables are assigned to the
separate output parameters.
The number of parameters on the left-hand side must
match the number of variable names on the right-hand
side (N = M).
The property/value pairs apply to all specified variables. If the
properties are not explicitly specified in the function call,
MLIB/MTRACE uses default values:
CPU

Name of the CPU on which the variables reside.

If you select MP mode, you must specify this property.
Otherwise, an error message is displayed.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

45

MLIB/MTRACE Functions

Type Data type of the specified variable(s). Use the abbreviations

from the following table to specify the data type:
Abbreviation

Description

Int8

8-bit signed char

UInt8

8-bit unsigned char

Int16

16-bit signed short

UInt16

16-bit unsigned short

Int32

32-bit signed integer

UInt32

32-bit unsigned integer

FloatIEEE32

32-bit single-precision floating-point type in IEEE 754 format

FloatIEEE64

64-bit double-precision floating-point type in IEEE 754 format,

default for DS1103, DS1104, DS1401, DS1005 and DS1006 boards

Int8Ptr

Pointer to 8-bit signed char

UInt8Ptr

Pointer to 8-bit unsigned char

Int16Ptr

Pointer to 16-bit signed short

UInt16Ptr

16-bit unsigned short

Int32Ptr

Pointer to 32-bit signed integer

UInt32Ptr

Pointer to 32-bit unsigned integer

FloatIEEE32Ptr

Pointer to 32-bit single-precision floating-point type in IEEE 754 format

FloatIEEE64Ptr

Pointer to 64-bit double-precision floating-point type in IEEE 754 format

Offset
Offset of the first element relative to the basic address. The
default value is 0.
The calculation of the actual memory address depends on
the data type of the variable:
n

For directly addressed variables:

addr = var.addr + sizeof(var.type)*offset

For variables referenced by pointers:

addr = *var.addr +
sizeof(referenced.type)*offset

Length
value is 1.

Number of elements as an integer number. The default

You can change the default properties of each variable descriptor by

using the SetVarProperty function.

46

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

The variable addresses are retrieved from the linker map file generated
by the linker. The MAP file must reside in the same folder as the
associated object files (PPC or X86). A variable declared as static, or
local variables within the function body, cannot be accessed.
Example

Example 1:
% obtain the descriptors for variables declared
% in the real-time application running on DS1005 board:
% int simState;
% double * currentTime;
% double finalTime;
var_names = {'simState'; 'currentTime'; 'finalTime'};
var
= mlib('GetMapVar',vars_names);
% set the appropriate data types
var(1)
= mlib('SetVarProperty', var(1), 'type', 'Int32');
var(2)
= mlib('SetVarProperty', var(2), 'type', ...
'FloatIeee64Ptr');
% var(3)
= mlib('SetVarProperty', 'var(3)', 'type, ...
% 'FloatIeee64'); not necessary, FloatIeee64 is default type
% for DS1005 board
Example 2:
% obtain the descriptors for variables declared
% in the real-time application running on DS1005 board:
% float mass, damping, spring_coefficient;
var_names = {'mass'; 'damping'; 'spring_coefficient'};
var
= mlib('GetMapVar', var_names, 'type', ...
'FloatIeee32');

Related topics

References
GetTrcVar on page 47
SetVarProperty on page 67

GetTrcVar
Purpose

To obtain the descriptor for a variable specified within the TRC file.

Syntax SP

var_vec
= mlib('GetTrcVar', variable_names)
[var1, ..., varN] = mlib('GetTrcVar', variable_names)

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

47

MLIB/MTRACE Functions

Syntax MP

var_vec
= mlib('GetTrcVar', variable_names, 'CPU', cpu_name)
[var1, ..., varN] = mlib('GetTrcVar', variable_names, 'CPU', cpu_name)

Description

If a real-time application is generated from a Simulink model using the

Real-Time Workshop and Real-Time Interface, a TRC file (TRC) is
generated automatically. The TRC file maps Simulink variable names to
the corresponding variables in the generated code (source code
variables). Variables in real-time applications can therefore be accessed
via the names of Simulink blocks, or via the labels of the
corresponding signal lines.
To describe accessible variables within an application program, you can
also create the TRC file manually. You can structure the information
hierarchically by assigning variables to groups and subgroups.
GetTrcVar can be used to find the descriptors of one or more variables
renamed in the TRC file (referred to as trace variables below). Specify
the names of the variables in the variable_names input parameter.
The variable_names parameter must be an M-by-1 cell array, where M
is equal to the number of trace variables whose descriptor are to be
obtained and where each cell contains the name of one variable.
var_vec is an M-by-1 vector of structures described in the chapter
Naming Conventions on page 32. If M=1, which means that the
descriptor of only one trace variable is to be obtained, the function
also accepts a simple string as an input parameter. var_vec is then a
simple structure.
If multiple output arguments are specified, var1,..., varN, the
variable descriptors corresponding to the specified variables are
assigned separately to each output argument.
The number of variable descriptors on the left-hand side
must match the number of variable names on the right
hand-side (N = M).
The properties of each variable descriptor (such as the address, data
type, length) are retrieved from the TRC file.
You can also change the properties by using the SetVarProperty
function.

48

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

If the trace variables are identical to the Simulink variables, the syntax
for their names is as follows:
Syntax

Description

Model Root/<subsystem>/<blockName>/<parameterName>

For block parameters

Model Root/<subsystem>/<blockName>/Out<i>

For i-th block outputs

Model Root/<subsystem>/<blockName>/ContState

For continuous states of a block

Model Root/<subsystem>/<blockName>/DiscState

For discrete states of a block

Model Root/<subsystem>/<blockName>/ContStateDeriv

For state derivatives of a block

Model Root/<subsystem>/<outportName>

For outputs of a subsystem

Labels/<label>

For labeled signals

For the parameters of the Look-Up tables the following syntax applies:
Syntax

Description

Model Root/<subsystem>/<blockName>/InputValues

For the vector with input values of the 1-D Look-Up

table block

Model Root/<subsystem>/<blockName>/OutputValues

For the vector with output values of the 1-D Look-Up

table block

Model Root/<subsystem>/<blockName>/Look-Up\nTable (2D)/Row Index

For the row vector of the 2-D Look-Up table block

Model Root/<subsystem>/<blockName>/Look-Up\nTable (2D)/ColumnIndex

For the column vector of the 2-D Look-Up table

block

Model Root/<subsystem>/<blockName>/Look-Up\nTable (2D)/OutputValues For the table matrix of the 2-D Look-Up table block

subsystem is the path name of the Simulink system where the

corresponding block resides, but with the model name omitted. Refer
to the Simulink Users Guide by MathWorks for Simulink system
identifiers. You can specify output of the block gain block Kp, which
resides in the subsystem controller of the model vcfp, with Model
Root/controller/Kp/Out1. You can specify Gain parameter with Model
Root/controller/Kp/Gain.
Labeled signals are block outputs in the Simulink model which are
represented by labeled signal lines. label denotes the label of the
signal to be specified.
In many cases, the subsystem, block or label string does not exactly
match the name (with model name omitted) of the associated
Simulink block or signal label, for example, if carriage returns or
slashes are involved, or if labels are identical. In ambiguous cases, it is
recommended to use the TrcView utility to obtain variable names.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

49

MLIB/MTRACE Functions

Description of vectorized
variables in TRC file

The vectorized variables can be described in the TRC file by declaring

of the first and last index:
myVector[10..99]
{
type: flt(64,IEEE)
alias: myVector
flags: PARAM
}
The vectorized variables can also be described in the TRC file by
specifying the vectorized data type:
typedef ArrayFlt_90 flt(64,IEEE)[10..99]
myVector[10]
{
type: ArrayFlt_90
alias: myVector
flags: PARAM
}
In the first case, the GetTrcVar function can only access single vector
elements in which the vector is disassembled by the TRC file parser.
Assuming that this vector is placed in the User Variables group, the call
to the GetTrcVar function looks like this:
elem_10 = mlib(GetTrcVar, User Variables/myVector[10])
to get the descriptor for the 10th element, or
elem_34 = mlib(GetTrcVar, User Variables/myVector[34])
to get the description for the 34th element, etc.
To get the descriptor for the whole vector, use the following calling
sequence:
myVectorDesc = mlib(GetTrcVar,UserVariables/myVector[10]);
myVectorDesc = mlib(SetVarProperty, myVectorDesc, Length, 90);

In the second case, the GetTrcVar function returns the descriptor for
the whole vector. Since the vectorized parameters, outputs, states and
state derivatives of the Simulink block are generated in the TRC file in
this manner, the retrieved descriptor always describes all of the vector
elements.
If you create TRC files manually, the following general syntax applies to
variable names:
group/subgroup1/.../subgroupN/name.

50

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

For example, variable Kp in the subsystem controller of the top level

group ABS system can be accessed with ABS system/controller/Kp.
Example

% set the parameter 'Amplitude' and 'Frequency'

% of the Simulink block named 'Signal Generator'
par_names
= { ...
Model Root/Signal Generator/Amplitude ; ...
Model Root/Signal Generator/Frequency ; ...
};
param
= mlib('GetTrcVar', par_names);
% read the value for amplitude and frequency
param_val
= mlib('Read', param);
% change the value of amplitude
param_val{1} = param_val{1} + 0.1;
% change the value of frequency
param_val{2} = param_val{2} + 10;
% write new value into real-time processor
mlib('Write', param, 'Data', param_val);

Related topics

Basics
Accessing Matrices with MLIB/MTRACE on page 78
References
GetMapVar on page 45
SetVarProperty on page 67

Intrpt
Purpose

To generate an interrupt on the real-time processor.

Syntax

mlib('Intrpt')

Result

A hardware interrupt is generated on the real-time processor. The

names of the interrupt service routines in C code are:
Board

Name

DS1103

Name as registered by DS1103_set_interrupt_vector()

DS1104

Name as registered by DS1104_set_interrupt_vector()

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

51

MLIB/MTRACE Functions

The interrupt service routine must be present in the

real-time program. The respective interrupt vector
must be initialized and the interrupt enabled (see the
demo application program adc_XXXX_hc.c in the folder
%DSPACE_ROOT%\demos\dsXXXX\mlib).

The generated interrupt is processed as the Host

interrupt in your application. This interrupt can only be
made available as trigger source in an application that
is generated with RTI1103 or RTI1104 (see
DS1103MASTER_HWINT_Ix (
DS1103 RTI
Reference) or DS1104MASTER_HWINT_Ix
(
DS1104 RTLib Reference)). RTI1005, RTI1006 and
RTI1401 do not support the host interrupt. If you want
to use the application on a DS1005, DS1006 or
MicroAutoBox, you must handcode the application
using the board-specific RTLib functions to get access
to this interrupt.
ST M

ST M

Example

% request a hardware interrupt on a DS1005

% processor board
mlib('SelectBoard', 'DS1005');
...
mlib('Intrpt');

IsApplRunning

52

Purpose

To check if an application is currently running on the real-processor

board(s).

Syntax

state = mlib('IsApplRunning')

Result

state will show if an application is running

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Description

MLIB/MTRACE Functions

If the SP mode is selected (default), state is an integer value that

equals zero or 1. If the MP mode is selected, state is an M-by-1 vector
of structures, where M is equal to the number of CPUs in an MP
application and where each structure contains the following fields:
Name

Description

.board

Board name

.cpu

CPU name

.appl_state

State of the program loaded on the CPU. This is an

integer value equal to 0 or 1.

In both SP and MP mode, the meanings of the state values are:

n

1 application is running

0 application is not running

Example

if mlib('IsApplRunning')
str = 'Currently running real-time application is:';
appl = mlib('GetApplInfo');
disp([str appl.name]);
else
disp('No real-time application running\n');
end

Related topics

References
GetApplInfo on page 42

LoadFile
Purpose

To load selected portions of data from the MAT file generated by

MLIB/MTRACE during continuous data acquisition with the
StreamToDisk option.

Syntax

data = mlib('LoadFile', mat_file_name)

data = mlib('LoadFile', mat_file_name, ...
property_name, property_value)

Result

The data you selected is loaded.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

53

MLIB/MTRACE Functions

Description

mat_file_name is a string containing the name of the file generated by

MLIB/MTRACE in MAT format. The file name extension .mat can be
omitted. By default, all data stored in the specified MAT file is returned
to the data output matrix. Each row of this matrix represents a time
series of one variable. If the data stored in the MAT file was collected
without time stamps, its number of rows corresponds to the number
of selected trace variables; otherwise, the number of rows corresponds
to the number of selected trace variables +1 (the first row is for the
time stamps). The rows in data are assigned to the trace variables in
the order in which the variables were selected with the TraceVars
property.
In many cases the amount of collected data can be too large to load it
all into the MATLAB workspace. Here you can specify a portion of data
to be loaded by setting appropriate property/value pairs:
Variables An M-by-1 cell array of variable names specifying the
variables for which the data will be loaded. The number of rows of the
data matrix corresponds to M, or to M +1 if the time stamps were
stored in the MAT file. The rows in the data are assigned to the
specified variables in the order in which the variables were selected. If
data for only one variable is to be loaded, the function also accepts a
string as a parameter. By default, the data for all variables is returned.
CPUs An M-by-1 cell array of CPU names. You must set this
property if data acquisition was performed in the multiprocessor mode
and some of the variables residing on the different CPUs have the
same name. To distinguish between variables, the CPUs must be
specified in addition to variable names (property Variables). By default,
MLIB/MTRACE assumes that all variable names are unique and loads
the data corresponding to the first variable found with the specified
name.
Samples
A2-by-1 vector of integer values specifying the first and
last samples of the interval for which the data is to be loaded. The
number of columns of the data matrix results from the interval length.
By default, the whole time series is loaded.
Interval A2-by-1 vector of double values specifying the interval (in
the units of the step size) from which the data is to be loaded. The
number of columns of the data matrix results from the interval length
divided by the step size and downsampling factor. By default, the
whole time series is loaded.

54

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

Example

% The data acquisition was performed in SP mode

% for variable Model Root/Signal Generator/Out1,
% Model Root/Plant/y/Out1 and Model Root/Controller/out/out1
% and stored into MAT file exp_data.mat
% load the whole time series for variables Model Root/Plant/y/Out1
% and Model Root/Signal Generator/Out1
variables = {'Model Root/Plant/y/Out1'; ...
Model Root/Signal Generator/Out1}; ...
data = mlib('LoadFile', 'exp_data', ...
'Variables', variables);
...
% load the data for the selected variable in the
% specified interval
data = mlib('LoadFile', 'exp_data', ...
'Variables', variables, ...
'Samples', [1000; 2000]);
% data is a 2 x 1001 matrix
...

Related topics

Basics

MLIB/MTRACE Benchmarks on page 127

LoadVarFile
Purpose

To load the variable description file.

Syntax

mlib('LoadVarFile', var_file_name)

Description

For single-processor systems, the variable description file is identical to

the TRC file. For multiprocessor systems, the variable description file is
identical to the SDF file. For further information on the TRC and SDF
files, refer to Creating System Description Files (
ControlDesk
Experiment Guide).
ST M

By default, MLIB/MTRACE assumes single-processor mode and that

the name of the TRC file matches the name of the application
currently running on selected board. If the name of the TRC file differs
from the name of the application program or if there are several
variable description files, you must use this function to specify the one
required for MLIB/ MTRACE. Only variables available in a currently
loaded TRC file can be accessed by the GetTrcVar function. Loading a
TRC file while the multiprocessor mode is active switches
MLIB/MTRACE to single-processor mode.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

55

MLIB/MTRACE Functions

If an SDF file is loaded, the multiprocessor mode is switched on. In this

case the function is similar to the Multiprocessing function.
The var_file_name must be a string with the file name extension TRC
or SDF. If the variable file resides in the MATLAB working folder, only
the file name of the relative path is sufficient, otherwise the absolute
path must be specified.
Example

% Select DS1005 board for use with MLIB/MTRACE.

% Note, that the application smdtf_1005_sl.ppc
% is currently running on selected board.
mlib('SelectBoard','DS1005');
% load the variable file smdtf_1005_sl_set1.trc
mlib('LoadVarFile','smdtf_1005_sl_set1.trc');

LockProgram
Purpose

To set a lock on an application program.

Syntax

mlib('LockProgram')

Result

The application program currently running on the selected processor

board is locked. The application program lock ensures that the realtime program cannot be mistakenly aborted or restarted by other host
applications while MLIB/MTRACE functions are in progress.
The real-time program remains locked until the MLIB
function UnlockProgram is explicitly called.

Example

56

% lock real-time processor

% to prevent other host applications
% from interfering
mlib('LockProgram');
% some calls to MLIB/MTRACE functions
...
% MLIB/MTRACE has finished, give way to other
% host applications
mlib('UnlockProgram');

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Related topics

MLIB/MTRACE Functions

References
UnlockProgram on page 73

Multiprocessing
Purpose

To set MLIB/MTRACE to multiprocessor mode.

Syntax

mlib('Multiprocessing', 'sdf_file_name')
mlib('Multiprocessing', 'off')

Result

MLIB/MTRACE is in the multiprocessor mode.

Description

Because MLIB/MTRACE is in single-processor mode by default, you

must activate the multiprocessor mode if MLIB/MTRACE is used in a
multiprocessor environment. Sdf_file_name is the name of the system
description file (SDF) of the currently running multiprocessor
application. YOu must specify the full path name if the M file calling
MLIB/MTRACE resides in a folder other than the files associated with
the application (PPC, X86, MAP, etc.). Invoking Multiprocessing with
the keyword off returns MLIB/MTRACE to the single-processor mode.

Example

% select board
mlib('SelectBoard', 'DS1005');
mlib('Multiprocessing', 'pipt1_dual1005_sl.sdf');
% access variables in the multiprocessor system
...
mlib('Multiprocessing', 'off');
% single-processor on DS1005...

Read
Purpose

To read data from the real-time processor.

Syntax

data
= mlib('Read', var_vec)
[data1, ..., dataN] = mlib('Read', var_vec)

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

57

MLIB/MTRACE Functions

Result

The data is read from the real-time processor.

Description

The var_vec parameter is an M-by-1 vector of variable descriptors as

returned by the GetTrcVar or GetMapVar function (see Naming
Conventions on page 32). A vector of data is read for each variable
descriptor. The data type and the number of elements are contained in
the variable descriptor (var_vec(i).type and var_vec(i).length,
respectively). The data vector is returned in the ith cell of the M-by-1
cell array data.
If you specify multiple output arguments (data1, ..., dataN), each
data vector read from the real-time hardware is assigned to a separate
output.
The number of arguments on the left-hand side must
match the number of variable descriptors on the righthand side (N = M).

Example

Example 1:
% get the descriptor of the variables simState,
% currentTime and read their values
% obtain the addresses
variables = {'simState'; 'currenTime'}
vars
= mlib('GetTrcVar', variables);
% read the values
data
= mlib('Read', vars);
simState
= data{1};
currentTime = data{2};
The last 3 statements can optionally be replaced by
[simState, currentTime] = mlib('Read', vars);
Example 2:
% read the 3 values of the variable int_vector and 5
% values of the variable dbl_vector
% get the variable descriptors
int_vector_desc = mlib('GetMapVar', 'int_vector', 'type', ...
'Int32', 'length', 3);
dbl_vector_desc = mlib('GetMapVar', 'dbl_vector', 'type', ...
'FloatIeee64', 'length', 5);
[int_vector, dbl_vector] = mlib('Read', ...
[int_vector_desc; dbl_vector_desc]);

58

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Related topics

MLIB/MTRACE Functions

References
Write on page 73

SearchMap
Purpose

To search the linker map file for variables whose names match the
specified regular expression.

Syntax SP

vars = mlib('SearchMap', regexp)

Syntax MP

vars = mlib('SearchMap', regexp, 'CPU', cpu_name)

Result

The linker map file corresponding to the currently running real-time

application is searched for variables whose names match the specified
pattern regexp. These names are stored in a column cell array vars,
where each cell contains one variable name as a string.
The pattern regexp must be a regular expression. The regular
expression allowed by MLIB/MTRACE is constructed as follows:
n

A period (.) matches any character.

An asterik (*) matches zero or more occurrences.

A plus (+) matches one or more occurrences.

A question mark (?) matches zero or one occurrence of the previous

pattern.

A caret (^) indicates the beginning of a line.

A dollar sign ($) indicates the end of a line.

A non empty string enclosed in ([]) brackets matches any single

character in that string. ASCII intervals such as [A-Z] are allowed.
The class of characters that are not expected is indicated by a caret
after a left bracket, for example, [^0-9].

Any special character such as ., *, [, ] etc. must be preceded by a

backslash (\) if you want it match itself.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

59

MLIB/MTRACE Functions

Example

% Assumed the user specific global variables are defined

% in the real-time application by
% float usr_var1, usr_var2, usr_var2 etc.
% get the names of all user specific variables
usr_vars
= mlib('SearchMap', '^usr_.*');
% obtain the descriptors for each of them
usr_var_desc = mlib('GetMapVar', usr_vars);

Related topics

References
SearchTrc on page 60

SearchTrc
Purpose

To search the TRC file for variables whose names match the specified
regular expression.

Syntax SP

var_names = mlib('SearchTrc', regexp)

Syntax MP

var_names = mlib('SearchTrc', regexp, 'CPU', cpu_name)

Result

The TRC file corresponding to the currently running real-time

application is searched for variables whose names match the specified
pattern regexp. These names are stored in a column cell array
var_names, where each cell contains one variable name as a string.
The pattern regexp must be a regular expression. The regular
expression allowed by MLIB/MTRACE is constructed as follows:

60

A period (.) matches any character.

An asterik (*) matches zero or more occurrences.

A plus (+) matches one or more occurrences.

A question mark (?) matches zero or one occurrence of the previous

pattern.

A caret (^) indicates the beginning of a line.

A dollar sign ($) indicates the end of a line.

A non empty string enclosed in ([]) brackets matches any single

character in that string. ASCII intervals as [A-Z] are allowed. The
class of characters that are not expected is indicated by a caret after
a left bracket: [^09].

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

The special character (as ., *, [, ] etc.) must be preceded by a

backslash (\) if you want it match itself.

Example

% Example
% Get the names of all Simulink block outputs
% contained in the subsystem PID-Controller
output_names = mlib(SearchTRC,.*PID-Controller/.*/Out1);
% Obtain the descriptors for found block outputs
output_desc = mlib(GetTrcVar,output_names);

Related topics

References
SearchMap on page 59

SelectBoard
Purpose

To select a processor board for use with MLIB/MTRACE.

Syntax

mlib(SelectBoard)
mlib('SelectBoard', 'board')

Description

If the board parameter is omitted, the working board set by

ControlDesk 3.xs Platform Manager is used.
Using ControlDesk Next Generation, you cannot specify a
working board.

The optional board parameter can be either an ASCII string or an

integer value. If it is an ASCII string, the board that is registered with
this name in ControlDesk 3.x's Platform Manager or in
ControlDesk Next Generation's Platform/Device Manager is selected. If
an integer value is specified, the board registered with that index is
selected. The selected real-time processor board is used for
subsequent MLIB/MTRACE calls.
To obtain a list of the boards currently registered in ControlDesk 3.x's
Platform Manager or in ControlDesk Next Generation's
Platform/Device Manager, enter the following in the MATLAB
Command Window:
boards = mlib('GetBoardInfo')

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

61

MLIB/MTRACE Functions

You can access only one platform (single-processor board

or multiprocessor system) at the same time.

Example

% select first registered board

boards = mlib('GetBoardInfo');
mlib('SelectBoard', boards(1).name);
...
% switch to second one
mlib('SelectBoard', boards(2).name);
...

Set
Purpose

To set the data acquisition options.

Syntax

mlib('Set', property_name, property_value, ...)

Result

The data acquisition options, such as the time frame, triggering and
variables to be traced, are set by defining the appropriate
property/value pairs.
The permissible property/value pairs for data acquisition are listed
below:
ServiceGroup (MP1005 or MP1006) Name of the service group
used for data acquisition on multiprocessor systems based on DS1005
or DS1006.
Service An integer number (1 31) specifying which data capture
service in the real-time program is to perform data acquisition
(MLIB/MTRACE can distinguish between up to 31 different services).
The default is service number 1.
Trigger

The trigger mode is turned on or off.

ON trigger mode enabled

OFF free-running mode (default value)

TriggerVariable
TriggerLevel

Value of the trigger level. Default value: 0.0

TriggerEdge

The trigger edge is specified:

62

Descriptor of the trigger variable

rising trigger on rising edge (default value)

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

falling trigger on falling edge

TraceVars Variables to be traced specified as an M-by-1 vector of

variable descriptors
StepSize The step size of the task containing the corresponding
host_service macro. For example, if the selected (or default) service is
contained in the timer task routine, the step size is identical to the
sampling period.
If the step size is defined, you can specify the trace interval either by
the Start/Stop properties or by Delay/NumSamples, otherwise only by
Delay/NumSamples.
Downsampling A positive integer value used to reduce the
amount of data per capture. A downsampling factor of n forces
MLIB/MTRACE to store a set of data only every n-th time the service
macro is called in the real-time program. Default value: 1.
TimeStamping Specifies if the time stamps at which the data was
collected are to be traced or not
n

ON - time stamps are traced

OFF- time stamps are not traced (default)

In MP1005 or MP1006 mode, this property is set to ON,
and you cannot modify it.

UseAbsoluteTime Specifies if the time stamps are to be stated as

absolute times or relative to the absolute zero point of the trace shot.
In triggered mode, the time at which the trigger event occurred is
taken as absolute zero, while in the free-running mode, the time
stamp of the data set first collected is absolute zero.
ON - use the absolute time
OFF- use the time relative to absolute zero (default value)
Start The start of the trace interval for which data is to be
collected, specified in units of the step size. If the trigger mode is set to
ON, a negative value indicates the pre-trigger mode, and a positive
value indicates the post-trigger mode (you must define the step size
beforehand).
Stop The end of the trace interval for which the data is to be
collected, specified in the units of the step size (you must define the
step size beforehand).

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

63

MLIB/MTRACE Functions

Delay Delay related to the trigger event specified in the number of

samples. A negative value indicates the pre-trigger mode, and a
positive value indicates the post-trigger mode (default value: 0).
NumSamples
Length of the trace interval specified in the number
of samples. Default value: 200.
AcquisitionMode

The mode of the data capture is specified.

continuous the data is collected until StopCapture is called. The

trigger must be set to OFF. This mode is useful if you want to
acquire large amounts of data over a long time period.

single_shot a single data capture whose length is specified by

the Start/Stop or Delay/NumSamples properties (default value)

StreamToDisk In continuous mode the collected data can

optionally be saved on the hard disk:
n

ON data is saved in a MAT file on the hard disk

OFF data is not saved in a MAT file (default value) on the hard
disk

OutputFormat Specifies the format in which the traced data is

available in the MATLAB workspace.
n

matrix - the data format is a matrix (default value). If

TimeStamping has been set to ON, the format is matrix with
time.

structure - the data format is a structure. If TimeStamping has

been set to ON, the format is structure with time.

For further information about the format, refer to function FetchData

on page 36.

64

The data is always stored in the MAT file as a matrix or

a matrix with time, independently of the output
format selected with this property.

In MP1005 or MP1006 mode, the matrix format

means that the matrix with time is to be returned
(because of the time stamps). However, the common
time axis for the data collected on different processors
can be found only if the data was traced with the
same sampling period. If the data on different
processors is traced by services called with different
sampling periods or asynchronously, the output format
must be set to structure

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

FileName If StreamToDisk is set to ON, you must specify the name

of the MAT file where the collected data is to be stored. You do not
need to type the file name extension MAT. For more information, see
MLIB/MTRACE Benchmarks on page 127.
MP1005 and MP1006 only:
The data collected on different processors contained in
the multiprocessor system is saved to separate MAT files
whose names must be unique. You can explicitly specify a
MAT file name for each processor by using this property
together with the CPU property. If the CPU property is not
specified, the names of the MAT files are built according
to the rule: <specifiedFileName>_<processorName>.mat
AutoClearBuffer
Specifies if the trace buffer is to be cleared and
freed after the last portion of collected data is fetched into the
MATLAB workspace:
n

ON the buffer is automatically cleared and freed (default value)

OFF the buffer is not cleared and the data can be read multiple
times.

TimeOutValue Integer value specifying the time in seconds after

which a timeout is reported during execution of the FetchData
function. TimeOut can occur if MLIB/MTRACE has to wait until the
desired is reached (default value: 5).
InterpolateBoundaryPoints (MP1005 and MP1006 only)
Specifies if points are to be added (if necessary) on the boundaries of
the trace interval. If yes, the values for the boundary points are
obtained using the interpolation method. This property can be useful if
data is captured asynchronously, for example, using the data
acquisition service from an asynchronous task, during the specified
time interval <t1,t2>. In this case only points with time stamps t>=t1
or t<=t2 are used. Because data acquisition is asynchronous, it can
happen that there are no points exactly on the boundaries.
ON - add points to the boundaries of the trace interval
OFF- do not add boundaries to the trace interval (default value)
This property takes effect only if you specify the trace
interval in the time unit, and not in samples.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

65

MLIB/MTRACE Functions

MaxSamplesPerSecond Specifies the estimated number of calls

per second to the asynchronous task containing the corresponding
host_service macro.
CPU Name of the CPU. If you specify this property, the following
properties (if specified) apply only to the selected CPU:
n

Service

FileName

MaxSamplesPerSecond

StepSize

Otherwise the above properties apply to all processors in the DS1005

or DS1006 multiprocessor application.
Example

Example 1:
% obtain the variable address
variables = {'Model Root/Signal Generator/Out1'; ...
'Model Root/Plant/out'};
vars_desc = mlib('GetTrcVar', variables);
% specify the data capture options:
% triggering on the variable Model Root/Signal Generator/out1,
% trigger level 0.0, rising edge;
% 500 samples with 10 samples pre-trigger;
% variables to be traced:
% 'Model Root/Signal Generator/Out1' and,
% 'Model Root/Plant/out';
% single_shot mode, downsampling 1
% service number: 1
mlib('Set', 'Trigger', 'ON', ...
'TriggerVariable', vars_desc(1), ...
'TraceVars', vars_desc, ...
'Delay', -10, ...
'NumSamples', 500);
Example 2:

66

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

% MP1005 mode has been activated

% get the addresses of the trace variables
sigGen = mlib(GetTrcVar,Model Root/Signal Generator/Out1,...
CPU,Master) ;
enablsysOut = mlib(GetTrcVar,Model Root/enablsys/output,...
CPU,Slave) ;
% For RTI-MP-generated applications, the available service groups
% are automatically obtained from TRC files.
% If TRC files do not exist, or any other service group is desired as
% described in TRC files, the service group
% can be specified by the user.
% in the example below, a new service group is created.
% It allows that data can be traced simultaneously with
% synchronous (CPU Master) and asynchronous (CPU Slave) tasks
mlib(Set,ServiceGroup,myNewServiceGroup,...
CPU,Master,...
Service, 1,...
StepSize, 0.00012);
mlib(Set,ServiceGroup,myNewServiceGroup,...
CPU,Slave,...
Service, 2,...
MaxSamplesPerSecond, 100);
% This new service group can now be used for data acquisition
% The data is to be captured during the time interval <-0.00036; 0.048>
% The values for the asynchronously
% traced variable Model Root/enablesys/output
% are to be interpolated on boundaries of this trace interval
mlib(Set,ServiceGroup,myNewServiceGroup,...
Trigger,on,...
TriggerVariable, sigGen,...
Start, -0.00036,...
Stop, 0.048,...
TraceVariables,[sigGen; enablsysOut];
InterpolateBoundaryPoints,on);

Related topics

References
Get on page 40

SetVarProperty
Purpose

To set the properties stored in the variable descriptor.

Syntax

new_var_vec = mlib('SetVarProperty', old_var_vec, ...

'property_name', property_value, ...)

Result

The new property values are stored in the variable descriptor.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

67

MLIB/MTRACE Functions

Description

The permissible properties are listed in the table below:

Name

Description

Name

Name of the variable

CPU

Name of the CPU on which the variable resides.

In MP mode, this property must not be changed
after the variable address was obtained with
GetTrcVar or GetMapVar.

Type

Data type of the variable. For possible data type

specifications, see the description of the
GetMapVar function.

Addr

The variables base address

Offset

Offset of the first element from the base address

as an integer number

RelativeOffset

Offset of the first element from the current

memory address

Length

Number of elements as an integer number

The calculation of the current memory address depends

on the data type of the variable:
n

For directly addressed variables:

addr = var.addr * sizeof(var.type) * offset

For variables referenced by pointers:

addr = *var.addr + sizeof(referenced.type) *
offset

The specified property/value pairs apply to all variable descriptors

stored in the column vector old_var_vec. The modified variable
descriptors are returned to the column vector new_var_vec. If
old_var_vec is empty, a variable descriptor with specified properties is
created. To create a descriptor, you must define the following
properties Name, Addr, Type and CPU (for MP mode). The defaults are
used for the others: Offset = 0, Length = 1.

68

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

Example

% obtain the descriptors of variables declared

% in the DS1005 real-time application by:
% int
int_vect[5];
% float flt_vect[10];
% double dbl_vect[10];
var_names
= {'int_vect'; 'flt_vect'; 'dbl_vect'};
[int_vect_var, flt_vect_var, dbl_vect_var] = ...
mlib('GetMapVar', var_names);
% set the appropriate data types and lengths
int_vect_var = mlib('SetVarProperty', int_vect_var, ...
'type', 'Int32', 'length', 5);
flt_vect_var = mlib('SetVarProperty', flt_vect_var, ...
'type', 'FloatIeee32', 'length', 10);
dbl_vect_var = mlib('SetVarProperty', dbl_vect_var, ...
'type', 'FloatIeee64', 'length', 10);
% read the values of the variable float_vect
float_vect = mlib('Read', flt_vect_var);

Related topics

References

GetMapVar on page 45
GetTrcVar on page 47

StartCapture
Purpose

To start a new capture on the real-time processor board

Syntax SP

mlib('StartCapture')
mlib('StartCapture', services)

Syntax MP

mlib('StartCapture')
mlib(StartCapture,serviceGroups)

Result

The data capture process is initialized.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

69

MLIB/MTRACE Functions

Description

If there is an appropriate service in the real-time application, a set of

data (sample) is collected once per call of that service (host_service:
see the chapter Preparing Real-Time Applications on page 18).
Multiple services can be used simultaneously in one real-time
application, for example, in different tasks.
With the services/serviceGroups parameter, you can specify which
of the services will perform the data acquisition process. If the
services/serviceGroups parameter is omitted, the default service or
service group is used. Otherwise, MLIB takes the service number or the
name of the service group which is specified by the
services/serviceGroups parameter.
In some cases it is desirable to have multiple data captures performed
by different data capture services or service groups that start
simultaneously. The options of the data capture process traced
variable, trigger, time frame, etc. - for each of the services or service
groups must already be defined. For further information, refer to
Multiple MLIB/MTRACE Data Captures in One Real-Time Application
on page 76. In such cases
n

The services parameter is a vector containing the service numbers.

The serviceGroups parameter is a cell array containing the names

of the service groups.

Example

% start a capture for an application generated by the RTI

mlib('StartCapture');
while mlib('CaptureState')~=0, end
mat_data = mlib('FetchData');
...
% start capturing data using the data capture service
% number 2
mlib('StartCapture', 2);

Related topics

References
CaptureState on page 34
StopCapture on page 71

70

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

StopCapture
Purpose

To stop the data capture on the selected processor board.

Syntax SP

mlib('StopCapture')
mlib('StopCapture', services)

Syntax MP

mlib('StopCapture')
mlib('StopCapture', serviceGroups)

Result

The data capture process started with the function StartCapture is

stopped.

Description

If the additional service/serviceGroups parameter is not specified,

MLIB stops the data acquisition performed by the default service or
service group. Otherwise, you must specify the service number or the
name of the service group with the services/serviceGroups
parameter.
If multiple data captures performed by different services or service
groups were started simultaneously, the services/serviceGroups
parameter must be:
n

A vector containing the numbers of services (SP) or

A cell array containing the names of the service groups (MP1005

and MP1006)

that are to be stopped.

If the data capture was started in single_shot mode, it will be stopped
automatically after the desired number of samples is collected. If the
data capture was started in continuous mode, you must invoke
StopCapture to stop it.
Example

% start data capture in 'continuous' mode

mlib('StartCapture');
TIC;
while TOC < 600
data = mlib('FetchData', 300);
plot(data')
...
end
% stop the capture after 10 minutes (600 sec)
mlib('StopCapture');

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

71

MLIB/MTRACE Functions

Related topics

References
CaptureState on page 34
StartCapture on page 69

TriggerState
Purpose

To get trigger status information.

Syntax SP

status = mlib('TriggerState')
status = mlib('TriggerState', services)

Syntax MP

status = mlib('TriggerState')
status = mlib('TriggerState', serviceGroups)

Result

The trigger status information is obtained.

Description

When used without an input parameter, this function returns the

trigger status information for the capture process performed by the
default service or the service group. Otherwise, you must specify the
service or service group via the services/serviceGroup parameter.
The possible values for status are:
Value

Description

Trigger is disabled

Trigger armed, waiting for a trigger event

Trigger event has occurred

-1

Trigger not armed

If multiple data captures performed by different services or service

groups are started simultaneously, you can check the trigger states of
all of them by calling the TriggerState function with the additional
services/serviceGroups input parameter, which is then
n

A vector containing the service numbers (services)

A cell array containing the names of the service groups

(serviceGroups)

carrying out the data acquisition.

72

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Functions

In this case, state is a vector whose elements indicate the trigger state
of the corresponding capture process.
Example

% waiting for a trigger event

while mlib('TriggerState', 2) ~= 0, end;
disp('Trigger event occurred');

Related topics

Basics
Additional MLIB/MTRACE Programming Examples on page 96

UnlockProgram
Purpose

To release a lock on the application program.

Syntax

mlib('UnlockProgram')
mlibn('UnlockProgram')

Result

A previously set lock on the real-time program is released.

Example

% lock application program to prevent other host

% applications from interfering
mlib('LockProgram');
% some calls to MLIB/MTRACE functions
...
% MLIB/MTRACE has finished, give way to other
% host applications
mlib('UnlockProgram');

Related topics

References
LockProgram on page 56

Write
Purpose

To write data to the real-time processor.

Syntax

mlib('Write', var_vec, 'Data', data)

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

73

MLIB/MTRACE Functions

Result

The data is written to the real-time processor.

Description

The var_vec parameter is an M-by-1 vector of variable descriptors as

returned by GetTrcVar or GetMapVar (see Naming Conventions
on page 32). The data parameter must be an M-by-1 cell array. A
vector matrix of data contained in the corresponding cell of the data
parameter is written for each variable descriptor. The data type and
the number of elements are specified in the variable descriptor
(var_vec(i).type and var_vec(i).length, respectively). The data
contained in a matrix is written column-wise.
If M = 1, data can be a simple vector or matrix. It does not need to be
a 1-by-1 cell array in this case.
To reduce execution times, MLIB/MTRACE does not
provide any range checking (depending on the data type)
for specified values. It is your responsibility to check
variable ranges, for example, for integer data types.

Example

% write a vector of parameters in double precision format

% and a vector of parameters in single precision format
% to memory of the DS1005 board
dbl_data
= [0.543212 pi];
flt_data
= [12.12 8.21 0.12 0.36];
% get the variable descriptors; note that the default
% value of the data type is FloatIeee64 for DS1005 board
[dbl_vec_var, flt_vec_var] = mlib('GetMapVar', ...
{'dbl_vector'; 'flt_vector'});
% set the proper data type
flt_vec_var = mlib('SetVarProperty', flt_vec_var, ...
'Type', 'FloatIeee32', 'Length', size(flt_data, 1));
dbl_vec_var = mlib('SetVarProperty', dbl_vec_var, ...
''Length', size(dbl_data, 1));
% write the data into real-time processor,
mlib('Write', [dbl_vec_var; flt_vec_var], 'Data', ...
{dbl_data; flt_data});

Related topics

References
Read on page 57

74

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Programming

This topic contains the following sections:

n

For some instructions on how to program MLIB/MTRACE, refer to

Programming Instructions on page 76

For information on accessing matrices with MLIB/MTRACE, refer to

Accessing Matrices with MLIB/MTRACE on page 78

For detailed demonstrations of programming with MLIB/MTRACE,

refer to MLIB/MTRACE Programming Examples on page 88

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

75

MLIB/MTRACE Programming

Programming Instructions
For details on how to program multiple MLIB/MTRACE captures in one
real-time application and how to lock the real-time application, refer
to:
n

Multiple MLIB/MTRACE Data Captures in One Real-Time Application

on page 76

Locking the Real-Time Program on page 77

Multiple MLIB/MTRACE Data Captures in One Real-Time

Application
MLIB/MTRACE needs special service code to be executed in the realtime application to enable real-time data capture. This code collects
one set of data for each sampling step.
You can call multiple service codes in different tasks of the real-time
application to trace variables of periodic timer tasks and asynchronous
interrupt-driven tasks simultaneously. MLIB/MTRACE can handle up to
10 services per real-time application.
The real-time program communicates with MLIB/MTRACE via the
host_service(nr, 0) macro. To distinguish between multiple service
routines, this macro must be called in the real-time program:
host_service(1,0);
host_service(2,0);
etc.
In the real-time applications generated by RTI, the service macros are
automatically inserted into desired tasks if a DataCapture block is
used.
Each service routine requires additional execution time in the
application program. If many service macros are placed in a timer task
on a processor that is already very busy, an overload may occur. The
execution time needed by a service routine depends mainly on the
number of variables and pointers to be traced.
With multiple MLIB/MTRACE captures, each service routine in a realtime program requires one separate set of data acquisition options.

76

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Example

mlib('SelectBoard');
mlib('Set', 'Service', 1, ...
'Trigger', 'ON', ...
'TriggerVariable', trg_addr, ...
'NumSamples', 1000, ...
'TraceVars', trc_vars_0);
mlib('Set', 'Service', 2, ...
'Trigger', 'ON', ...
'TriggerVariable', trg_addr, ...
'NumSamples', 1000, ...
'Downsampling', 10, ...
'TraceVars', trc_vars_1);
mlib('StartCapture', [1; 2]);

Related topics

Basics

MLIB/MTRACE Programming

Impact of MLIB/MTRACE on Execution Time on page 130

Locking the Real-Time Program

The currently running real-time program can be locked with the
MLIB/MTRACE function LockProgram to protect it from being restarted
or corrupted by other host applications. All dSPACE PC software relies
on the device driver which provides this locking mechanism. A realtime program can be locked by one or more host programs at a time.
To unlock the real-time program, each host program must release its
lock. MLIB/MTRACE releases a previously set lock when
mlib('UnlockProgram') is invoked.
If a program is downloaded to a processor board via ControlDesk 3.x's
Platform Manager or ControlDesk Next Generation's Platform/Device
Manager after an MLIB/MTRACE instance locked the currently running
real-time program, the error message 'Another program has locked
the real-time program.' will appear. Return to the MATLAB Command
Window and type mlib('UnlockProgram') to make the MLIB/MTRACE
instance unlock the real-time program. If no other host programs have
locked the program, the next download via ControlDesk 3.x's Platform
Manager or ControlDesk Next Generation's Platform/Device Manager
should be successful.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

77

MLIB/MTRACE Programming

Accessing Matrices with MLIB/MTRACE

Objective

In the C code generated by the MATLABs Real-Time Workshop and

dSPACEs RTI, parameter matrices are represented by simple doubleprecision arrays. The index for finding a specific matrix element mainly
depends on whether the matrix was written out row-wise or columnwise. In some cases, there are additional rules that you have to comply
with if you wish to modify online transfer function parameters or lookup table data with MLIB/MTRACE. A good way to become familiar
with the rules that apply when MATLABs Real-Time Workshop maps
parameter matrices to C code is to see the associated model.prm file.
For detailed information about the rules, refer to:
n

Built-in Simulink Blocks on page 79

2-D Look-Up Tables on page 79

S-Functions on page 79

MLIB/MTRACE provides some M files that perform read and write

operations on the matrices and look-up tables parameters. For
details, see below.
Tools

MLIB/MTRACE supplies tools that let you perform Read/Write

operations on the matrices and lookup-tables parameters. These
tools are written as the M files mlib_1dlookup.m, mlib_2dlookup.m,
and mlib_matrix.m, and reside in %DSPACE_ROOT%\matlab\mlib. They
can be called from the MATLAB Command Window or embedded in
other M files.
Refer to:

78

mlib_1dlookup on page 80

mlib_2dlookup on page 83

mlib_matrix on page 86

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Programming

Built-in Simulink Blocks

Matrices are written out row-wise (C-style). The index of the (p, q)
element of an (m, n) matrix is thus (p-1)n+q. In linear time-invariant
(LTI) systems, optimization algorithms omit matrix elements equal to 0
or 1 in the generated code. Transfer functions must be transformed
into state space systems before code can be generated, so polynomial
coefficients (or poles and zeros) cannot be represented directly.

2-D Look-Up Tables

The following rules apply when parameters of 2-D look-up tables are
stored:
The n-vector elements are stored first, followed by the m-elements of
the y-axis vector.
The elements of the (m,n)-table data matrix are stored
n

Row-wise up to MATLAB 5.3.x (R11)

Column-wise since MATLAB 6.0 (R12)

S-Functions
Matrices are written out column-wise (MATLAB or Fortran-style). The
index of the (p, q) element of an (m, n) matrix is thus (q-1)m+p.
For more information, refer to the Real-Time Workshops Guide by
Mathworks.
MLIB/MTRACE always writes data vectors to the associated real-time
memory column-wise. For parameter matrices of built-in Simulink
blocks, the MATLAB transpose operator can be used to make sure that
matrices are written in the right order.
The following code shows how the parameters that represent state
space matrices of the Simulink compensator, observer or State Space
block can be modified online by using MLIB/MTRACE.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

79

MLIB/MTRACE Programming

Example

This example works only if the matrices do not contain zeros or ones,
and if the sizes of the matrices equal the sizes of the matrices in the
Simulink block that the code was generated from.
...
% Design the estimator (as an example for a LTI system).
% a, b, c, d are the resulting state space matrices.
[a,b, c, d] = my_estimator_design_routine;
% Select and lock DSP board
mlib('SelectBoard', 'DS1005');
mlib('LockProgram');
% Obtain addresses
var_names = { ...
['Model Root/compensator/estimator/' ...
'State Space/A']; ...
['Model Root/compensator/estimator/ ...
'State Space/B']; ...
['Model Root/compensator/estimator/ ...
'State Space/C']; ...
['Model Root/compensator/estimator/ ...
'State Space/D']; ...
};
[a_var, b_var, c_var, d_var] = mlib('GetTrcVar', var_names);
% the GetTrcVar function already obtains the number of elements
% in each of specified state space
% Write down the new estimator matrices. Use the transpose
% operator to make sure the matrix elements are stored in the
% right order.
data_to_writing = [a'; b'; c'; d'];
mlib('Write', [a_var; b_var; c_var; d_var], ...
'Data', data_to_writing);

If zeros or ones are involved, omitted elements must also be taken care
of, and the code will not be as simple. An easy way to avoid zeros and
ones is to add eps (machine precision) to the matrices before
generating code with RTI, though this method involves additional
computational loads and may change the system dynamics slightly.

mlib_1dlookup

80

General syntax

output = mlib_1dlookup(Command, ParameterList)

Description

Depending on the selected command, mlib_1dlookup obtains the

descriptors of the vectors of input and output values (input vector and
output vector) of the specified 1-D look-up table, and performs
Read/Write operations.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Init

MLIB/MTRACE Programming

myTable = mlib_1dlookup(Init,lutName [,length][,procName])

The descriptors of the 1-D look-up table parameters, input vector and
output vector, are obtained and returned in the myTable structure.
The myTable structure contains the following fields:
Field

Description

length

Length of the 1-D table vectors

inputDesc

Descriptor of the input vector

outputDesc

Descriptor of the output vector

name

1-D table name, set by lutName

cpu

Processor name, if specified, otherwise an empty

string

The syntax for the lutName parameter is the same as for the Simulink
block parameters, whose addresses are to be obtained with the
GetTrcVar function Model Root/<subsystem>/<blockName>, where
<subsystem> is the path name of the Simulink system containing the
block called <blockName>, but with the model name omitted.
n

Use the trcview utility to get the correct lutName.

Length is only supported for downward compatibility

with previous versions of this tool. It is not necessary to
specify it.

If this function is called in multiprocessor mode, you must specify the

processor name.
ReadInput

data = mlib_1dlookup(ReadInput,myTable [,indices])

This call reads the values of the input vector. If the indices vector
parameter is specified only the values of the elements that it specifies
are read; otherwise, the whole vector is returned.
myTable must be the structure returned by the Init function.

ReadOutput

data = mlib_1dlookup(ReadOutput,myTable [,indices])

This call reads the values of the output vector. If the indices vector
parameter is specified only the values of the elements that it specifies
are read; otherwise, the whole vector is returned.
myTable must be the structure returned by the Init function.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

81

MLIB/MTRACE Programming

WriteInput

mlib_1dlookup(WriteInput,myTable, [,indices], data)

This call writes data to the vector of input values. If the indices vector
parameter is specified data is written only to the elements that it
specifies; otherwise to the whole input vector.
myTable must be the structure returned by the Init function. The data
parameter must be a vector containing the data to be written.
The number of elements in data must match the number
of elements to be written.

WriteOutput

mlib_1dlookup(WriteOutput, myTable [,indices], data)

This call writes data to the vector of output values. If the indices
vector parameter is specified data is written only to the elements that
it specifies; otherwise, to the whole output vector.
myTable must be the structure returned by the Init function. The data
parameter must be a vector containing the data to be written.
The number of elements in data must match the number
of elements to be written.

Example

The 1-D look-up table block named 1DTable resides in the Simulink
subsystem Linearization. The number of elements is 11.
% Init the mlib_1dlookup function
myTable = mlib_1dlookup(Init,...
Model Root/Linearization/1DTable );
% Read all elements of the vector of input values - the
% indices specification is superfluous
inputData = mlib_1dlookup(ReadInput, myTable);
% Read the elements 2, 3 and 4 of the vector of output values
outputData = mlib_1dlookup(ReadOutput, myTable, [2:4])
% Write the new data to the area read before
newOutputData = [1 2 3];
mlib_1dlookup(WriteOutput, myTable, [2:4], newTableData);

82

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Programming

mlib_2dlookup
General syntax

output = mlib_2dlookup(Command, ParameterList)

Description

Depending on the selected command, mlib_2dlookup obtains the

descriptors of the row vector, column vector and table of the specified
2-D look-up table, and performs Read/Write operations.

Init

myTable = mlib_2dlookup(Init,lutName [,rows,cols]...

[,procName])
The descriptors of the 2-D look-up table parameters row vector,
column vector and table values are obtained and returned in the
myTable structure.
The myTable structure contains the following fields:
Field

Description

rows

Number of rows in the 2-D table

cols

Number of columns in the 2-D table

rowDesc

Descriptor of the row vector

colDesc

Descriptor of the column vector

tableDesc

Descriptor of the table values

name

2-D table name, set by lutName

cpu

Processor name, if specified, otherwise empty

string

The syntax for the lutName parameter is the same as for the Simulink
block parameters whose addresses are to be obtained with the
GetTrcVar function Model Root/<subsystem>/<blockName>, where
<subsystem> is the path name of the Simulink system containing the
block named <blockName>, but with the model name omitted.
n

Use the trcview utility to get the correct lutName.

Rows and cols are only supported for downward

compatibility with previous versions of this tool. It is
not necessary to specify them.

If this function is called in multiprocessor mode, you must specify the

processor name.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

83

MLIB/MTRACE Programming

ReadRow

data = mlib_2dlookup(ReadRow,myTable [,indices])

This call reads the values of the row vector. If the indices vector
parameter is specified, only the values of the elements that it specifies
are read; otherwise, the whole vector is returned.
myTable must be the structure returned by the Init function.

ReadCol

data = mlib_2dlookup(ReadCol,myTable [,indices])

This call reads the values of the column vector. If the indices vector
parameter is specified, only values of the elements that it specifies are
read; otherwise, the whole vector is returned.
myTable must be the structure returned by the Init function.

ReadTable

data = mlib_2dlookup(ReadTable, myTable [,rows, cols])

This call reads the table values. If the vector parameters rows and cols
are specified, only the rows and columns that they specify are read;
otherwise, the whole table is returned.
myTable must be the structure returned by the Init function.

WriteRow

mlib_2dlookup(WriteRow,myTable, [,indices], data)

This call writes data to the row vector. If the vector indices parameter
is specified, data is only written to the elements that it specifies;
otherwise, to the whole row vector.
myTable must be the structure returned by the Init function. The data
parameter must be a vector containing the data to be written.
The number of elements in data must match the number
of elements to be written.

WriteCol

mlib_2dlookup(WriteCol, myTable [,indices], data)

This call writes data to the column vector. If the indices vector
parameter is specified, data is written only to the elements that it
specifies; otherwise, to the whole column vector.
myTable must be the structure returned by the Init function. The data
parameter must be a vector containing the data to be written.

84

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Programming

The number of elements in data must match the number

of elements to be written.

WriteTable

mlib_2dlookup(WriteTable, myTable, [,rows, cols], data)

This call writes the table data. If the vector parameters rows and cols
are specified, data is written to the rows and columns that they
specify; otherwise to the whole table.
myTable must be the structure returned by the Init function. The data
parameter must be a matrix containing the data to be written.
The size of the data matrix must match the lengths of
rows and cols vectors, or the size of the whole table.

Example

The 2-D look-up table block named 2DTable resides in the Simulink
subsystem Linearization. The number of rows is 3 and the number of
columns is 5.
% Init the mlib_2dlookup function
myTable = mlib_2dlookup(Init,...
Model Root/Linearization/2DTable );
% Read all elements of the row vector - the indices
% specification is superfluous
rowData = mlib_2dlookup(ReadRow, myTable);
% Read the elemnts 2,3 and 4 of the column vector
colData = mlib_2dlookup(ReadCol, myTable, [2:4])
% Read the table values: columns 1-3 from rows 2 and 3.
% The dimension of the output data matrix will be 2 x 3.
tableData = mlib_2dlookup(ReadTable,myTable, [2:3], [1:3]);
% Write the new data to the area read before
newTableData = [1 2 3; 5 6 7];
mlib_2dlookup(WriteTable, myTable, [2:3], [1:3],...
newTableData);

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

85

MLIB/MTRACE Programming

mlib_matrix
General syntax

output = mlib_matrix(Command, ParameterList)

Description

Depending on the selected command, mlib_matrix obtains the

descriptor of the specified matrix and performs Read/Write operations.

Init

myMatrix = mlib_matrix(Init,matName, rows, cols,...

alignment [,procName])
The matrix descriptor is obtained and returned in the myMatrix
structure.
The myMatrix structure contains the following fields:
Field

Description

rows

Number of matrix rows, set by rows parameter

cols

Number of matrix cols, set by cols parameter

matDesc

Matrix descriptor

name

Name, set by matName

alignment

Describes how the matrix elements are stored in

real-time memory;
possible values: Row-Wise, Col-Wise

cpu

Processor name, if specified, otherwise empty

string

The name of the first matrix element must be specified as the matName
parameter. Apply the same syntax as for the Simulink block vectorized
parameters, whose addresses are to be obtained with the GetTrcVar
function Model Root/<subsystem>/<blockName>/<matrixParameter>,
where <subsystem> is the path name of the Simulink system
containing the block named <blockName>, but with the model name
omitted.
Use the trcview utility to get the correct syntax of the
matName.

If this function is called in multiprocessor mode, you must specify the

processor name.

86

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Read

MLIB/MTRACE Programming

data = mlib_matrix(Read, myMatrix [,rows, cols])

This call reads the matrix values. If the vector parameters rows and
cols are specified, only the rows and columns that they specify are
read; otherwise the whole matrix is returned.
myMatrix must be the structure returned by the Init function.

Write

mlib_matrix(Write, myMatrix, [,rows, cols], data)

This call writes the data matrix. If the vector parameters rows and cols
are specified, the data is written to the rows and columns that they
specify; otherwise to the whole matrix.
myMatrix must be the structure returned by the Init function. The
data parameter must be a matrix containing the data to be written.
The size of the data matrix must match the lengths of
rows and cols vectors, or the size of the whole matrix.

Example

The State Space block State-Space resides in the Simulink subsystem

Controller. The dimension is 3 x 3.
% Init the mlib_matrix function for A matrix
myMatrix = mlib_matrix(Init,...
Model Root/Controller/State-Space/A,...
3, 3, Row-Wise);
% Read the values of the A matrix.
matrixData = mlib_matrix(Read,myMatrix);
% Write the new data to the following area: rows 2,3,
% columns 1:2
newData = [1 2; -0.5 1];
mlib_matrix(Write, myMatrix, [2:3], [1:2], newData);

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

87

MLIB/MTRACE Programming

MLIB/MTRACE Programming Examples

For a demonstration of MLIB/MTRACE programming, refer to:

88

MLIB/MTRACE with Handcoded Applications on page 89

MLIB/MTRACE with dSPACE RTI on page 93

Additional MLIB/MTRACE Programming Examples on page 96

MLIB/MTRACE with dSPACE RTI-MP on page 97

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Programming

MLIB/MTRACE with Handcoded Applications

The following M file (simple_xxxx.m) is an example of a program that
demonstrates MLIB/MTRACE capture capabilities with a handcoded
application. It is located in %DSPACE_ROOT%\demos\dsxxxx\mlib.
% Syntax: simple_1005;
% Purpose: Demonstrates MLIB/MTRACE on a DS1005
% This file needs the demo application
% '%DSPACE_ROOT%\demos\ds1005\GettingStarted\HandCode\
% smd_1005_hc.obj' running.
%
% The input signal 'x_disp', and the states 'x' and 'x_dot' of
% a second order lag system are traced. Data is plotted in a
% figure window.
% Copyright (c) 1996-1999 by dSPACE GmbH
% select processor board 'ds1005'
mlib('SelectBoard',DS1005);
% select variables to be traced and obtain their descriptors
var_names = {'Model/x_disp';...
'Model/x';...
'Model/x_dot'...};
var = mlib('GetTrcVar',var_names)
% set the options of data acquisition performed by service
% number 1 (default)
mlib('Set','Trigger','ON', ...
'TriggerLevel',0, ...
% default, can be omitted
'TriggerEdge','rising', ...
% default, can be omitted
'TriggerVariable', var(1), ... % trigger on variable 'Model/x_disp'
'TraceVars', var, ...
'NumSamples', 500, ...
'Delay',0);
% default, can be omitted
% start capture on DS1005
mlib('StartCapture');
% wait until capture is done
while mlib('CaptureState')~=0,end
% fetch after capture is complete
out_data = mlib('FetchData');
% plot results
clf;
subplot(2,1,1);
plot(out_data(1:2,:)');
title('x\_disp, x');
subplot(2,1,2);
plot(out_data(3,:)');
title('x\_dot');
set(gcf,'NumberTitle','off');
set(gcf,'Name','A Simple MLIB/MTRACE Example');
figure(gcf);

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

89

MLIB/MTRACE Programming

Running simple_XXXX.m should result in plots as shown below:

MLIB/MTRACE can be used for real-time data acquisition. It also

provides asynchronous reading and writing of parameters in the realtime application. The example file chmass_xxxx.m demonstrates the
use of both features. It is contained in
%DSPACE_ROOT%\demos\dsxxxx\mlib.

90

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Programming

function chmass_1005(mass);
% Syntax: chmass_1005(mass)
% Purpose: Demonstrates MLIB/MTRACE on a DS1005
% This file needs the demo application
%'DSPACE_ROOT\demos\ds1005\GettingStarted\HandCode\smd_1005_hc.obj' running.
% The input signal 'x_disp', and the state 'x' of a second
% order lag system are traced.
% Data is plotted in a figure window.
% The chmass_1005.m demo is augmented by MLIB/MTRACE
% functions, which are used to change the mass 'm'
% of the spring-mass-damper system.
% Copyright (c) 1996-2000 by dSPACE GmbH
if margin ~= 1,
mass=[0.002 0.005 0.01 0.02];
end
if mass<0.001 | mass>0.1,
error('*** mass must be in the range 0.001 < mass < 0.1')
end
% select DS1005 board for use with MLIB/MTRACE
mlib('SelectBoard', 'DS1005');
% select the variable used by MLIB/MTRACE and get their
% descriptors
variables = {'Model/Model Parameters/m';...
'Model/x_disp';...
'Model/x'};
[m_desc, x_disp_desc, x_desc] = mlib('GetTrcVar', ...
variables)
% set the option for data acquisition performed
% by service number 1 (default)
mlib('Set', 'Trigger', 'ON', ...
'TriggerLevel', 0, ...
% default, can be omitted
'TriggerEdge', 'rising', ... % default, can be omitted
'TriggerVariable', x_disp_desc, ...
'TraceVars', [x_disp_desc; x_desc], ...
'NumSamples', 100, ...
'Delay', -10, ...
'DownSampling', 1);
% default, can be omitted
for m = mass,
% lock program during capture
mlib('LockProgram');
% write new mass value to DS1005 via MLIB/MTRACE
mlib('Write',
m_desc, 'Data', m);
% start capture and wait until it is complete
mlib('StartCapture');
while mlib('CaptureState')~=0,end
% transfer data when capture is complete and release lock
out_data = mlib('FetchData');
mlib('UnlockProgram');
% plot results
clf;
plot('out_data');
title(['x_disp, x (mass=' num2str(m) ')']);
set(gcf,'NumberTitle', 'off');
set(gcf,'Name', 'MLIB/MTRACE');
figure(gcf);
if (m ~= mass(end)),
fprintf ('-- press any key --\n');
pause;
end;
end
mclear;

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

91

MLIB/MTRACE Programming

The following plot contains the results of the chmass_XXXX sample M

file.

(chmass_XXXX(0.01))

(chmass_XXXX(0.005))

92

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Programming

MLIB/MTRACE with dSPACE RTI

The following examples refer to the demo program for PPC boards
smdtf_XXXX_sl.ppc located in %DSPACE_ROOT%\demos\dsXXXX\mlib.
Although smdtf_xxxx_sl.ppc was generated from a Simulink model
using dSPACEs Real-Time Interface (RTI) to MATLAB/Simulink, neither
Simulink nor RTI is needed to run the MLIB/MTRACE sample programs.
You can use ControlDesk 3.x's Platform Manager or ControlDesk
Next Generation's Platform/Device Manager to download
smdtf_xxxx_sl.ppc to the desired processor board.

The Simulink model smdtf_XXXX_sl shown above represents the

transfer function of a spring-mass-damper system. A signal generator
is part of smdtf_XXXX_sl. The following M file, siggen_XXXX.m, is used
to display and change the signal generator parameters online.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

93

MLIB/MTRACE Programming

Example

This example refers to a DS1005 board.

function siggen_1005 (amplitude,period);
% ----------------------------------------------------------% function siggen_1005 (amplitude, period);
% SIGGEN_1005() sets the parameters (amplitude and period)of a
% signal generator using the dSPACE MATLAB Interface Library.
% The signal generator is part of the application
% 'smdtf_1005_sl'.
% It is assumed that the unit of the parameter 'frequency' of
% the signal generator is 'Hertz'.
% amplitude
peak value of the signal
% period
period length of the signal
%
(not used for noise signals)
% When called without input parameters, SIGGEN_1005 simply
% displays the current settings of the signal generator.
% -----------------------------------------------------------% Copyright (c) 1996-2000 by dSPACE GmbH
% Initialize dSPACE MLIB/MTRACE
mlib('SelectBoard', 'DS1005');
% Get descriptors for signal generator parameters
variables = {...
'Model Root/Signal\nGenerator/Amplitude'; ...
'Model Root/Signal\nGenerator/Frequency' ...
};
[amplit_var, freq_var] = mlib('GetTrcVar', variables);
if margin = 0,
% Print out command syntax
fprintf ('\n Usage: siggen(amplitude, period)\n\n');
% Print out signal generator parameters
[amplit,freq] = mlib('Read', [amplit_var; freq_var]);
frpintf ('Signal generator parameters:' ...
'Amplitude=%g, Period=%g sec\n'], amplit,1/freq');
return;
end
% Set period and amplitude
values = {1/period; amplitude};
mlib('Write', [freq_var; amplit_var], 'Data', values);

The ch_damp_XXXX.m M file is used to modify the damping factor of

smdtf_XXXX_sl. If you type RETURN at the input prompt of the
ch_damp_XXXX function to accept the defaults, the damping factor will
be changed from 0.5 to 1.0, in steps of 0.05 performed every second.
The results can be displayed easily with instruments in ControlDesk 3.x
or ControlDesk Next Generation.
Example

94

%
%
%
%
%
%
%
%
%
%
%
%

-----------------------------------------------------------Modify the damping factor of the system 'smdtf_1005_sl' at

certain time steps
Transfer function of the spring-mass-damper system
omega*omega
b0
G(s) = ---------------------------------- = --------------s*s + 2*damp*omega s + omega*omega s*s + a1 s + a0
Parameters stored in C Code:
P:Spring-Mass-Damper System.A(1)

-a1

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Programming

%
P:Spring-Mass-Damper System.A(2)
-a0
%
P:Spring-Mass-Damper System.C(2)
b0
%
% Before invoking this M file the real-time processor
% application smdtf_1005_sl.ppc must be
% loaded to the DS1005 processor board.
%
% !! To view the results,open the experiment
% file smdtf_1005_sl.cdx
% -----------------------------------------------------------% Initialize dSPACE MLIB
mlib('SelectBoard','DS1005');
% check if the application smdtf_1005_sl.ppc is running
DemoApplName = lower([pwd '\smdtf_1005_sl.ppc']);
if mlib('IsApplRunning'),
ApplInfo = mlib('GetApplInfo');
if strcmp(DemoApplName,lower(ApplInfo.name)) ~= 1,
err_msg = sprintf(['*** This MLIB demo file needs ',...
'the real-time processor application\n***',...
' ''%s'' running!'], DemoApplName);
error(err_msg);
end
else
err_msg = sprintf(['*** This MLIB demo file needs ',...
'the real-time processor application\n***',...
' ''%s'' running!'], DemoApplName);
error(err_msg);
end;
% Get descriptors for the state space matrices
matrices = {'Model Root/Spring-Mass-Damper System/A';...
'Model Root/Spring-Mass-Damper System/C'};
[Amatrix_desc, Cmatrix_desc] = mlib('GetTrcVar',matrices);
% Read current parameter values from real-time processor
Amatrix = mlib('Read',Amatrix_desc);
Cmatrix = mlib('Read',Cmatrix_desc);
a1 = -Amatrix(1);
a0 = -Amatrix(2);
b0 = Cmatrix(2);
OMEGA = sqrt(a0);
% bandwidth
DAMP = a1/(2*OMEGA); % damping factor
fprintf ('\n %8.1e \n',b0);
fprintf (' G(s) = -------------------------\n',b0);
fprintf (' s*s %+8.1e s %+8.1e\n\n',a1,a0);
fprintf (' Bandwidth: %6.0f\n',OMEGA);
fprintf (' Damping factor: %6.2f\n',DAMP);
% Input vector of new damping factors
damp_values = ...
input(' New damping factor(s) [0.5:0.05:1.0] ?');
if isempty(damp_values), damp_values = 0.5:0.05:1; end;
% Input times when parameters are updated
update_time = input(' Update interval 1 sec] ? ');
if isempty(update_time), update_time=1; end;
fprintf (' ---------------------------------------------\n');
% Start MATLAB stopwatch timer (min. timer resolution 54 ms)
TIC; T=0;
for damp=damp_values,
% Compute transfer function parameters
% of spring-mass-damper system
amatrix = [ -2*damp*OMEGA -OMEGA*OMEGA ]; % [ -a1 -a0 ]
% Write new transfer function parameters
% to real-time processor
mlib('Write',Amatrix_desc,'Data', amatrix);

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

95

MLIB/MTRACE Programming

fprintf ('T=%5.2f sec. Damping factor is now %5.2f,...\n', ...

T, damp);
% Let background processes become active while waiting
T = T + update_time;
while (TOC < T), end;
end
fprintf (' ---------------------------------------------\n');
% Reset gain to original value
mlib('Write',Amatrix_desc,'Data',Amatrix);
fprintf (' T=%5.2f sec. Damping factor reset to %5.2f\n',... T,DAMP);

Additional MLIB/MTRACE Programming Examples

There are additional MLIB/MTRACE programming examples in the
MLIB/MTRACE demo directories:

96

The M fileexectime_XXXX.m displays the execution time of the

current application.

chkerr_XXXX.m reads the error flag of the real-time program and

prints the error code.

trigger_xxxx.m shows how to start the data capture using the

trigger signal and how to cause the trigger event by means of
MLIB/MTRACE.

Another M file, cont_XXXX.m, demonstrates the use of

MLIB/MTRACE in continuous mode.

For an interrupt-driven A/D conversion, look at adc_XXXX.m.

You can find the corresponding C-sources in adc_XXXX_hc.c

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Programming

MLIB/MTRACE with dSPACE RTI-MP

The following illustration shows the Simulink model
pipt1_dual1005_sl, which can be used to generate the multiprocessor
application with dSPACE RTI-MP.

The following example, mlib_dual1005.m, demonstrates how to use

MLIB/MTRACE to change a parameter of the application and displays
the results:

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

97

MLIB/MTRACE Programming

Example

98

% mlib_dual1005.m - This is an example M file for

% demonstrating the multiprocessing feature of MLIB.
%
% Before invoking this M file the application
% 'pipt1_dual1005_sl' must be build with RTI-MP and downloaded
% to the dual DS1005 MP-System.
%
% The period of the signal generator is changed by MLIB and
% the result is plotted in a MATLAB figure window
% Copyright (c) 2000 by dSPACE GmbH
%
% $Workfile: mlib_dual1005.m $ $Revision: 1 $
% $Date: 10.06.03 15:45 $
% $Archive: /sw/MATLAB/MLIB and MTRACE/Installation/MLIB_4.4.7
% /demos/ds1005MP/dual1005/mlib/mlib_dual1005.m $
% Initialize dSPACE MLIB
mlib('SelectBoard','ds1005');
% DS1005 multiprocessor
% applications are based
% on DS1005 boards
% select MP-application configuration file
sdfFile = [pwd '\pipt1_dual1005_sl.sdf'];
mlib('MultiProcessing',sdfFile);
% get descriptors for variables residing on processor 'MASTER'% and 'SLAVE'
master_var = {'Model Root/Signal\ngenerator/Amplitude';...
'Model Root/Signal\ngenerator/Frequency';...
'Model Root/Signal\ngenerator/Out1'};
slave_var = 'Model Root/First Order Lag/y/Out1';
[ampl, freq, siggen] = mlib('GetTrcVar', master_var, 'CPU',...
'MASTER');
sys_out
= mlib('GetTrcVar', slave_var, 'CPU',...
'SLAVE');
% read the current values of amplitude and frequency
[org_ampl_val, org_freq_val] = mlib('Read', [ampl;freq]);
% set the option of the data acquisition performed by service
% group HostService
mlib('Set','Trigger','ON',...
'TriggerLevel',0,...
% default, can be omitted
'TriggerEdge','rising',... % default, can be omitted
'TriggerVariable',siggen,...
'TraceVars',[sys_out;siggen],...
'NumSamples',400,...
'Delay',-3);
% set the amplitude of the Signal Generator to 0.5
mlib('Write',ampl,'Data',0.5);
clf;
ah = 0;
for per = [0.002:0.002:0.02],
mlib('Write',freq,'Data',1/per);
% change the period of
% the signal genarator
mlib('StartCapture');
% start capture on the
% DS1005, default
% service number 1
while mlib('CaptureState')~=0,end % wait until data
% acquisition finished
out_data = mlib('FetchData');
if ah == 0,
% in DS1005-MP mode time stamps are always collected
t = out_data(1,:);
axis([t(1) t(end) -0.9 0.9]);
ah = gca;
l1 = line(t,out_data(2,:)');

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE Programming

l2 = line(t,out_data(3,:)');
set(l1,'Color','r');
grid;
else
set(l1,'YData',out_data(2,:),'EraseMode','Xor');
set(l2,'YData',out_data(3,:),'EraseMode','Xor');
end
figure(gcf);
drawnow;
pause(1);
end
% reset the amplitude and frequency
mlib('Write',[ampl;freq],'Data',{org_ampl_val;org_freq_val});
% return to single processor mode
mlib('MultiProcessing','OFF');

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

99

100

MLIB/MTRACE Programming

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

File Reference

MAT File Generated by MLIB/MTRACE

When the StreamToDisk option is enabled, the data collected in
continuous mode is stored in a MAT file on the hard disk. Even if the
format of the collected data has been set to structure, it is stored as a
matrix. To avoid large matrices, which are difficult to manage in the
MATLAB workspace, MLIB/MTRACE splits the data in the MAT file into
multiple matrices for long data captures (2 MB at the most).
If the data has been collected in SP mode, the resulting matrices are
named rt_data, rt_data_2, etc.
If the data has been collected in MP1005 or MP1006 mode, the
resulting matrices are named <processorName>_rt_data,
<processorName>_rt_data_1, <processorName>_rt_data_2, etc.
The whole MAT file and the desired matrices can be loaded in the
MATLAB workspace with the standard MATLAB load function. Any
selected portion of data can be loaded with the MLIB/MTRACE
LoadFile function.
In addition to the data matrices, the structures
n

experiment_info for SP

<processor_name>_experiment_info for MP1005 or MP1006

are also stored.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

101

File Reference

This structure contains fields providing information about the

performed data acquisition:
Field

Information

.date

Date when the MAT file was created (date of

the real-time experiment)

.step_size

Step size with which the data was collected

timestamping

102

ON: Time stamps at which data was

collected are also traced. In this case MAT
file matrices are matrices with time
OFF: Time stamps are not traced

.downsampling

Downsampling factor

.applications

Cell array of the name(s) of the application(s)

running while data acquisition was performed

.variables

Cell array of names of the traced variables

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Utilities

Basics on the TrcView Utility

Objective

The utility %DSPACE_ROOT%\exe\TrcView.exe provides a user interface

to TRC files (generated or user-written). By using this tool you can
enter variable names more conveniently and without type mismatches
in the GetTrcVar function

Possible problems when

entering variable names

Descriptors for global variables in a real-time application generated

with RTI can be obtained with the GetTrcVar function. Every time a
real-time application is built, RTI generates a TRC file that maps
Simulink variable names to the corresponding global variables in the
generated code. Thus variables can be specified with the names of
Simulink blocks whose outputs, parameters, states, or state derivatives
they represent, or with the labels of the corresponding signal lines. For
more information about the syntax used when a variable is specified
with GetTrcVar, refer to the description GetTrcVar on page 47.
The names of variables defined in a TRC file cannot always exactly
match the names of the associated Simulink blocks, their parameters
or signal labels. There are some cases when there are deviations; for
example, if block or subsystem names contain carriage returns or
slashes, or if identical signal labels exist. Moreover, it is not very
convenient to enter lengthy strings manually, especially if the Simulink
system features numerous hierarchies with long subsystem names.
Because descriptor evaluation with Simulink variable names is based
on string comparison, a call to GetTrcVar always results in an error if
only one character in a variable specification does not comply.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

103

Utilities

How to Use TrcView

Method 1

To use TrcView
1 Enter the trcview command in the MATLAB Command Window or
call the program directly from a DOS window. Then open the
desired TRC file. A model browser and a variable list make it easy to
search for a specific variable.
2 Double-click in the list to paste the full variable name, with the
syntax required by GetTrcVar, into an edit field.
3 From the edit field, copy and paste the variable name into any M
file editor or the MATLAB Command Window with the standard
shortcut keys Ctrl-C, Ctrl-V
The illustration below shows the Simulink system SimModel. Suppose
that the Constant Value parameter of the Simulink block Discrete PID
for realization/Anti-Windup/zero is to be changed with MLIB/MTRACE.
The name of the subsystem contains a carriage return.

104

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Utilities

For example, to obtain a descriptor for the parameter Constant Value,

the GetTrcVar function call is
mlib(GetTrcVar, ...
['Model Root/Discrete PID\nfor realization/Anti-Windup/'...
'zero.Value'];

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

105

Utilities

The following illustration shows the TrcView window after the

associated TRC file SimModel.trc has been opened. It shows a search
for the Constant Value parameter of the Simulink block Discrete PID
for realization/Anti-Windup/zero.

Method 2

To copy and paste a variable name to the M file-editor of the

MATLAB Command Window
1 In the hierarchy browser on the left, click the node of the desired
subsystem to display its available variables in the list box on the
right.
2 Double-click the desired variable name to paste it, with the syntax
required by GetTrcVar, to the edit field at the bottom of the
window.
3 From here, copy and paste the variable name to any M file editor or
the MATLAB Command Window.

Syntax errors in the TRC

file

If a TRC file contains syntax errors, you cannot open it with TrcView,
and the trace error file is displayed instead. The error positions are
marked and appropriate messages are also output.
For details on the TRC file syntax, refer to Syntax of the TRC File
(
RTI and RTI-MP Implementation Reference).
ST M

106

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Limitations

For information about MLIB/MTRACE limitations, refer to the topics

below:
n

Accessing Variables with Read/Write Functions on page 108

Number of Applications Simultaneously Accessing the Hardware

on page 108

Number of Platforms Simultaneously Accessed by MLIB/MTRACE

on page 108

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

107

Limitations

Accessing Variables with Read/Write Functions

Variables located in the local SRAM memory on the DS1103 and
DS1401, and the cached memory on DS1104, DS1005 and DS1006
boards, can be accessed by MLIB/MTRACE Read/Write functions via
the host interface and the special service, master_cmd_server. This
service must run in the background routine of the application
program.

Number of Applications Simultaneously Accessing the

Hardware
The number of applications that simultaneously perform real-time data
acquisition on PPC and X86 boards is limited to eight. ControlDesk can
represent more than one application provided multithreading is used
in the automation scripts. For example, while instrumentation is in
animation mode, automation scripts can access the real-time hardware
via the RTPLib.

Number of Platforms Simultaneously Accessed by

MLIB/MTRACE
You can access only one platform (single-processor board or
multiprocessor system) at the same time.

108

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Troubleshooting

Where to go from here

Information in this section

MLIB/MTRACE Error Messages

109

Invalid Variable Connections When Migrating to

MATLAB R2011a

125

MLIB/MTRACE Error Messages

Shows MLIB/MTRACEs error messages in numerical order and offers
solutions where applicable.
MLIB 1

No dSPACE processor board selected. A dSPACE processor

board registered in ControlDesk 3.xs Platform Manager must be
selected before any MLIB/MTRACE function other than SelectBoard
and GetBoardInfo can be performed.
1. Use mlib( SelectBoard ,<board>) to select a processor board.

MLIB 2

MLIB function <name> not implemented. The first parameter

of mlib() call does not specify an implemented MLIB/MTRACE
function.
1. Type help mlib in the MATLAB Command Window to get
information about the implemented functions.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

109

Troubleshooting

MLIB 3

Improper number of input arguments. The number of input

arguments does not match the number of arguments needed by the
specified function.
1. Type help <function_name> in the MATLAB Command Window to
get information about the function function_name.

MLIB 4

RESERVED

MLIB 5

Boards of type <boardType> cannot be selected

Only dSPACE
processor boards DS1103, DS1104, DS1401, DS1005 and DS1006 can
be selected for direct access with MLIB. Other dSPACE boards cannot
be accessed at all.

MLIB 6

RESERVED

MLIB 7

Variable descriptor expected. A variable descriptor as returned

by one of the address evaluation functions or created by
SetVarProperty is required as an argument of the MLIB/MTRACE
function.
1. Type help <function_name> in the MATLAB Command Window to
get information about the function function_name that caused this
error message.
See Naming Conventions on page 32 for information on variable
descriptors.

MLIB 8

Argument count is of improper type.

parameter must be scalar.

The specified count

MLIB 9

Number of output arguments must agree with the size of input

argument. The address evaluation functions GetTrcVar, GetMapVar
and the Read function allow a vectorized input parameter. If more
than one output parameter is specified, their number must match the
size (number of rows) of the input vector.

MLIB 10

Input argument must be of string type. An input argument of

non-string type (type: double, struct, cell, etc.) was specified.
1. Type help <function_name> in the MATLAB Command Window to
get information about the respective MLIB/MTRACE function.

110

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB 11

Syntax error in variable name specification.

variable name is not valid.

Troubleshooting

The syntax of a

1. Refer to the syntax of the appropriate address evaluation function

shown in MLIB/MTRACE Function Reference on page 33.
MLIB 12

RESERVED

MLIB 13

RESERVED

MLIB 14

Input argument is of improper type or size.

1. Type help <function name> in the MATLAB Command Window to
get information about the function <function name> that issued
this error message.

MLIB 15

RESERVED

MLIB 16

Only <maxnumber> variables allowed. The number of

traceable variables that can be accessed directly is limited to
<maxnumber>.

MLIB 17

Only <maxnumber> pointers allowed. The number of traceable

variables that can be referenced by pointers is limited to
<maxnumber>.

MLIB 18

Variable name missing. The first argument of one of the address

evaluation functions must be the name(s) of variables() for which the
descriptor is to be obtained.
1. Type help <function_name> in the MATLAB Command Window to
get information about the respective MLIB/MTRACE function.

MLIB 19

MLIB 21

Processor specification missing or incorrect. If the

multiprocessing mode is selected, you have to specify the processor for
some functions.
n

Type mlib(MultiProcessing, OFF) if you want to proceed with

MLIB/MTRACE in the single-processing mode.

Type help <function_name> in the MATLAB Command Window to

get information about the respective MLIB/MTRACE function and
how to specify the processor.

RESERVED

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

111

Troubleshooting

MLIB 22

RESERVED

MLIB 23

Error while accessing the field <field_name>. Field missing.

The function that issued this error message requires a variable
descriptor (a structure) as an input parameter.
1. Use only descriptors that are returned by one of the address
evaluation functions or generated by the SetVarProperty function.

MLIB 24

RESERVED

MLIB 25

RESERVED

MLIB 26

File name extension missing. SDF or TRC expected. The

extension of the specified variable description file name must be TRC
or SDF. In the second case the multiprocessing mode will be activated.

MLIB 27

Size mismatch between input parameter. The Write function

requires that a set of data is specified for each var_vec. The number of
elements to be written at memory address var_vec(i).addr must
match the length given under var_vec(i).length.
1. Type help <function_name> in the MATLAB Command Window to
get more information.

MLIB 28

Invalid property <property_name>.

1. Type help <function name> in the MATLAB Command Window to
get information about the function <function name> that issued
this error message.

MLIB 29

Bad value for property <property_name>

1. Type help <function name> in the MATLAB Command Window to
get information about the function that issued this error message.

MLIB 30

Invalid property/value pair arguments.

Type help <function name> in the MATLAB Command Window to get
information about the function that issued this error message.

112

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB 31

Troubleshooting

Multiprocessor application not specified. If the function

GetApplInfo is called in SP mode and the CPU-property is specified,
you must also specify the name of the SDF file describing the
multiprocessor application.
For further information, type help GetApplInfo in your MATLAB
Command Window.

MLIB 32

Can't create a variable descriptor. <property_name> missing.

1. If you create a variable descriptor using the SetVarProperty
function, you must also specify the following properties name, addr,
type. In MP mode, the CPU name is also required.

MLIB 33

Trigger mode cannot be set. Trigger variable not specified.

Before the trigger mode can be turned on, the trigger variable must be
specified.

MLIB 34

Illegal type of variable on which to trigger.

types for the trigger variable are:
n

The permitted data

Int8, Int8Ptr, Int16, Int16Ptr, Int32, Int32Ptr, UInt8, UInt8Ptr,

UInt16, UInt16Ptr, UInt32, UInt32Ptr, FloatIeee32,
FloatIeee32Ptr, FloatIeee64, FloatIeee64Ptr

MLIB 35

Illegal file extension.

Type help <function_name> in the MATLAB
Command Window to get information on the respective function that
issued this error message.

MLIB 36

Variable <variable_name> cannot be traced. Illegal data

type. The permitted data types for the trace variable are:
n

MLIB 37

Int8, Int8Ptr, Int16, Int16Ptr, Int32, Int32Ptr, UInt8,

UInt8Ptr, UInt16, UInt16Ptr, UInt32, UInt32Ptr, FloatIeee32,
FloatIeee32Ptr, FloatIeee64, FloatIeee64Ptr

No trace variables specified. Before the StartCapture function

can be called, the trace variables must be specified. See the description
of the Set function for information on how to specify the variable to
be traced.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

113

Troubleshooting

MLIB 38

Set the time frame for capturing real-time data first. Before
the function StartCapture can be called, you must specify the time
frame. See the description of the Set function for information on how
to specify the data capturing time frame.

MLIB 39

RESERVED

MLIB 40

The collected data is stored directly into the MAT file. Please
change the file extension to .mat.

MLIB 41

StepSize not specified in the <file_name> file. Please use the

'Samples' property instead of 'Interval'.
The information about
StepSize is not saved in the MAT file <file_name> generated during
continuous data acquisition (see experiment_info structure). For this
reason the interval at which the data is to be loaded into the MATLAB
workspace can only be specified by using the Samples property. See
the description of the LoadFile function for further information.

MLIB 42

Continuous mode cannot be activated. Set the 'Trigger'

property to 'OFF' first. The continuous data capture mode
requires that triggering is turned off.

MLIB 43

Inconsistent specifications of the capture frame. If the capture

frame is defined by Delay/NumSamples as well as by Start/Stop
properties during one call of the Set function, MLIB/MTRACE checks if
the specifications are consistent. The delay and number of samples
resulting from Start, Stop and StepSize must match the values of Delay
and NumSamples.

MLIB 44

Time interval cannot be specified! Unknown sample time.

The capture frame can be specified by using the Start and Stop
properties only if the StepSize or MaxSamplesPerPeriod was defined
beforehand.

MLIB 46

114

No data to be read in the trace buffer. Call the function

StartCapture first.
Before the FetchData function can be called,
data capturing must have been completed. If all the data was read
from the trace buffer and the AutoClearBuffer property was set to
'ON', any call to FetchData without first capturing new data results in
this error message.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Troubleshooting

MLIB47

TriggerState cannot be obtained. Call the StartCapture function

first.
The actual trigger state can only be obtained after the start of
the data capture process.

MLIB 48

Data acquisition options not specified for service/service group

<service number>/<service group>. Before the StartCapture
function is called with the specified service number (SP) or service
group (MP1005, MP1006), the Set function must be called to initialize
the options of the data acquisition performed by the specified
service/service group.

MLIB 49

Improper value for service number. MLIB/MTRACE supports up

to 10 services concurrently running within the real-time application.
The valid service numbers are 1-10.

MLIB 50

Acquisition service <service_number> not available or not

active. The specified service <service_number> either is not
available within the real-time application or is not used for real-time
data acquisition.

MLIB 51

Trace buffer overloaded. Select less variables or increase the

trace interval. If MLIB is running in continuous mode, the
collected data is written to the ring buffer. The size of this ring buffer is
set by NumSamples or StepSize/Start/Stop. After the last position of
the buffer is reached, the acquisition service continues writing the
collected data at the beginning of the ring buffer, etc. The existing,
previously written data will be overwritten only if it was read (fetched)
in time by MLIB. Otherwise this error message is displayed.

MLIB 52

RESERVED

MLIB 53

RESERVED

MLIB 54

RESERVED

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

115

Troubleshooting

MLIB 55

In the continuous mode with DS1005-MP or DS1006-MP systems

the format of the output data must be set to structure. If
continuous mode is selected for multiprocessor systems based on
DS1005 or DS1006, the data collected on different processors is not
put in a single time correlation. It is therefore not possible to return the
data in the MATLAB workspace in a matrix with a common time axis.
However, in the generated MAP files (if StreamToDisc mode is
activated) the data is still stored as a matrix, independently of the
selected format of the output data.
Use the Set function property OutputFormat to set the format for the
output data.

MLIB 56

The trace interval can be specified either by Start/Stop/StepSize

or by NumSamples/Delay properties combination
In
multiprocessing mode (DS1005, DS1006) it is possible to trace data
with different sampling periods (step sizes) on different processors or
trace data asynchronously. If the trace interval is specified by
NumSamples and Delay, MLIB/MTRACE collects the desired number of
samples on each processor.
If the trace interval is specified by Start, Stop and StepSize,
MLIB/MTRACE finishes the data acquisition after a specified time.
Different numbers of samples may be collected depending on each
processors sampling periods.

116

MLIB 57

StreamToDisk mode cannot be activated. Set the name of the

MAT file first.
Before StreamToDisk mode can be selected, you
must specify the name of the MAT file where the collected data is to
be stored. Use the Set function property FileName to specify the
name of the MAT file.

MLIB 100

Specified SDF file does not describe a multiprocessor system

based on DS1005 or DS1006 boards. The SDF file which is
specified as a parameter of the multiprocessing function must describe
a multiprocessor system. Check your SDF file.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB 101

Troubleshooting

No processor board registered with index <board_index>.

There is no board registered in ControlDesk 3.xs Platform Manager or
in ControlDesk Next Generations Platform/Device Manager with the
index <board_index>.
1. Invoke mlib(GetBoardInfo) to get information on which boards
are registered in ControlDesk 3.xs Platform Manager or in
ControlDesk Next Generations Platform/Device Manager.

MLIB 102

Processor board <board_name> not registered. The board is

not registered in ControlDesk 3.xs Platform Manager or in
ControlDesk Next Generations Platform/Device Manager.
1. Invoke mlib(GetBoardInfo) to get information which boards are
registered in ControlDesk 3.xs Platform Manager or in ControlDesk
Next Generations Platform/Device Manager.

MLIB 103

No real-time program running on board <name>.

A program
must be downloaded to the processor board before you can work with
MLIB/MTRACE functions.
1. Use ControlDesk 3.xs Platform Manager or ControlDesk
Next Generations Platform/Device Manager to download an
application to the selected processor board.

MLIB 104

Board is locked by another host application.

This error
message occurs if MLIB/MTRACE tries to access a board simultaneously
with another host program. This can happen only if another host
application crashed when it was about to access a board and was not
able to finish the operation.
1. If continuous attempts to call the respective MLIB/MTRACE function
fail, refresh the hardware connection by using ControlDesk 3.xs
Refresh Platform Connection command or ControlDesk
Next Generation's Refresh Interface Connection command.

MLIB 105

Can't reset/restart, real-time application program is locked.

As long as a real-time application is locked it cannot be reset or
restarted.

MLIB 106

Processor <processor_name> not found.

The processor does
not belong to the multiprocessor system specified in the SDF file.
1. Check if <processor_name> is spelled correctly.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

117

Troubleshooting

MLIB 107

RESERVED

MLIB 108

The board <board_name> cannot be initialized The file

<file_name> is missing. A processor board can only be initialized
if a real-time application is currently running on it and the matching
MAP file exists.

MLIB 109

Can't find processor board with index <index>. There are no

boards with the specified index registered in ControlDesks Platform
Manager.
1. Invoke mlib('GetBoardInfo') to get information on which boards are
currently registered with which index in ControlDesks Platform
Manager.

MLIB 110

Multiprocessor system is not running or it is running the wrong

application.
1. Check whether all processor boards related to the multiprocessor
application (as defined in the system description file, SDF) are
registered in ControlDesk 3.xs Platform Manager or in ControlDesk
Next Generations Platform/Device Manager, and whether the
correct real-time applications are running on them.

118

MLIB 111

RESERVED

MLIB 112

RESERVED

MLIB 113

RESERVED

MLIB 114

RESERVED

MLIB 115

master_cmd_server missing or not initialized properly. To read

data from or write data to the DS1103, DS1104, DS1401, DS1005 or
DS1006, the master_cmd_server function must be called in the
background task of the real-time application. This service call is
inserted automatically for applications generated by RTI.

MLIB 116

Time out. Maybe master_cmd_server is not called.

The
application is probably stopped or has crashed. Reload the real-time
application and try again.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Troubleshooting

MLIB 117

Communication error. There is a failure in the communication

between the host PC and the real-time processor board. The possible
reasons for this are: master_cmd_server is not initialized, the
application crashed or illegal data access was performed, such as
read/write from a wrong memory address.

MLIB 118

RESERVED

MLIB 119

RESERVED

MLIB 120

No free entry available in the instances list of service

<service_number>.
Each host_service specified by a service
number and called within the real-time application can collect data for
up to 8 independent host PC applications. For example, if
ControlDeskNext Generation and rtplib already used the specified
service <service_number> for real-time data acquisition, it has only
one service entry free, i.e., for MLIB. This error message occurs, for
example, from an attempt to start data acquisition with the service
<service_number> with an additional MLIB instance.

MLIB 121

Cannot install an acquisition service. An entry of the structure

_TraceServiceStruct cannot be found in the MAP file belonging to
the currently running application.

Probably the data capture service was not correctly initialized. Check if
the function init_trace() is called in your application.
MLIB 122

RESERVED

MLIB 123

Trigger variable does not belong to selected MP subsystem.

Refer to Using MLIB/MTRACE for Multiprocessor Systems on page 23.

MLIB 124

RESERVED

MLIB 125

Cant find address of variable <name>. Check if the global

variable <name> is spelled correctly. Inspect the linker map file of the
currently loaded real-time application where all global variables appear
with a preceding underscore.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

119

Troubleshooting

MLIB 126

Cant find variable <name> in <filename> If <filename> points

to a TRC file, inspect the TRC file or start the TrcView utility to obtain
permissible variable names. Check If <name> is spelled correctly.
If <filename> points to a MAP file, check if <name> is spelled correctly.
Inspect the linker map file of the currently loaded real-time
application, where all global variables appear with a preceding
underscore.

MLIB 127

Time stamping is not supported. Probably the application was

build by the older version on the real-time library.
Time
stamping on PPC boards is supported by the following RTLIB software:
n

RTLIB 1103 version 1.3 or higher

RTLIB 1401 version 1.3 or higher

RTLIB 1005 version 1.1 or higher

RTLIB 1006 version 1.0 or higher

To use the time stamping feature with MLIB/MTRACE, you have to

rebuild your applications with the appropriate version of the RTLIB
library.
MLIB 128

MLIB 129

Your application has been created with an older version of the

real-time library which does not support the distributed tracing
in multiprocessing mode. Please recompile your application and
try again to start this function. The distributed tracing is
supported by:
n

RTLIB 1005 version 1.1 or higher

RTLIB 1005 version 1.0 or higher

Processor ID mismatch.
Each member processor of a
multiprocessor system based on DS1005 or DS1006 has its own
unique processor index (ID). this error message occurs if the processor
indices that are read directly from the real-time application do not
match the indices specified in the SDF file (Keyword: ServiceID).
If the real-time application was generated by RTI-MP, these indices will
be written to the SDF file automatically. If the real-time application was
handcoded, you must set the indices, and they must start with 0 in a
real-time application and with 1 in a SDF file.

MLIB 130

120

No free entry in the reservation table.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB 131

Troubleshooting

Trigger variable does not belong to the selected service group

<service_group>! A service group specifies the processors and the
services with which data acquisition can be carried out. Only variables
which reside on processors belonging to a specified service group can
be traced and selected as trigger variables.
Example

An example of calling up this error message:

A DS1005-based multiprocessor system consists of 3 member

processors Alpha, Beta and Gamma. The ServiceGroupAB is defined as
Processor Alpha, service 1 and Processor Gamma, service 1.
An attempt to set a trace variable residing on processor Beta as a
trigger variable will cause this error message.

MLIB 132

Call mlib(Get,ServiceGroup,<service_group>) to get

information about the CPUs and the services combined in the
service group.

Call the function Set to add a processor with the desired trigger
variable to your service group.

Trace variable does not belong to the selected service group

<service_group>! A service group specifies the processors and the
services with which data acquisition can be carried out. Only variables
which reside on processors belonging to a specified service group can
be traced and selected as a trigger variables.
Example

An example of calling up this error message:

A DS1005-based multiprocessor system consists of 3 member

processors Alpha, Beta and Gamma. The ServiceGroupAB is defined as
Processor Alpha, service 1 and Processor Gamma, service 1.
If a trace variable resides on processor Beta, an attempt to set this
trace variable for data acquisition carried out by service group
ServiceGroupAB will cause this error message.

MLIB133

Call mlib(Get,ServiceGroup,<service_group>) to get

information about the CPUs and the services combined in the
service group.

Call the Set function to add a processor with the desired trace
variable to your service group.

Error while parsing the SDF file <file_name>. <Message>

Parsing of SDF file failed. The reason is given in the <message> string.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

121

Troubleshooting

MLIB134

Error while parsing the TRC file <file_name>. <Message>

Parsing of TRC file failed. The reason is given in the <message> string.

MLIB135

Error while reading file <file_name>. Probably file was not

created by MLIB/MTRACE The LoadFile function can only be used
with the MAT file generated by MLIB/MTRACE in StreamToDisk mode.

MLIB136

It is not possible to find common time axis for variables traced

on the single processors. Probably the data has been collected
with different sampling periods or asynchronously. Use
structure output format instead of matrix.
On
multiprocessor systems based on DS1005 or DS1006, variables are
traced separately on each processor. The time stamps at which the
data was collected are also traced. If the output format is set to
matrix (matrix with time in MP mode), the data series from all
processors must have the same time stamps. It is only possible if data
acquisition is carried out with the same sampling period on each
processor and if the absolute times on each processor are
synchronized with each other.
Use the Set function property OutputFormat to change the format of
the traced data from matrix to structure

MLIB137

The time stamps have not been traced during the data
acquisition. Due to it is not possible to find a time correlation
between data being collected on different processors. The
reason can be a wrong call to the host_service macro in your
real-time application.
The time stamps can be collected only if
the host_service is called with the pointer to the time stamp structure
as a second parameter:
host_service(service_no, ts)
If your application was generated by RTI/RTI-MP, a call to the macro is
made automatically. If your application is handcoded, make sure that
the macro is called correctly.

MLIB 138

The TRC file <file_name> was generated/written in the old

syntax! It does not contain information about which service
belongs to which service group (denoted by service name).
This error message may occur if the ShowServices function is called.
To show service groups predefined in the TRC file, entries in the form
of

122

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Troubleshooting

sampling_period[<index>]
{
value:
0.0
alias:
<serviceName>
}
are required. They are generated with RTI/RTI-MP version 3.6 and
higher. Regenerate your application with the newer RTI/RTI-MP version
if you want to use this function.
MLIB 139

Time out after <number> sec while waiting for desired number
of samples to be collected in the trace buffer. If necessary you
can increase the time after which the timeout error is reported
by altering the Set function property TimeOutValue.
Altering this property allows you to increase the time after which this
error message is displayed, for example, if your applications run very
slowly.
If the FetchData function is called with count as an input argument
and data acquisition is still in progress, MLIB/MTRACE waits until the
desired number of samples are collected in the trace buffer.
To avoid MLIB/MTRACE hanging when the data is not collected in the
trace buffer (for example, if the application crashed), the MATLAB
Command Window is automatically reactivated after <number>
second(s). In this case a timeout error message is displayed.

MLIB 140

Time out after <number> sec while waiting for number of

samples collected in the trace buffer greater or equal the
StartSample value. If necessary you can increase the time after
which the timeout error is reported by altering the Set
function property TimeOutValue. Altering this property allows
you to increase the time after this error message is displayed, for
example, if your applications run very slowly.
If the FetchData function is called with the StartSample property
while data acquisition is still in progress, MLIB/MTRACE waits until the
desired number of samples are collected in the trace buffer.
To avoid MLIB/MTRACE hanging when the data is not collected in the
trace buffer (for example, if the application crashed), the MATLAB
Command Window is automatically reactivated after <number>
second(s). In this case a timeout error message is displayed.

MLIB 200

Host memory allocation failure.

the host heap.

There is insufficient memory on

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

123

Troubleshooting

MLIB 201

Real-time processor memory allocation failure.

insufficient free memory on the real-time processor.

There is

MLIB 202

Host PC - real-time processor communication error. This error

can occur in an attempt to download data to or upload data from a
processor board that is not connected to a host PC (for example,
AutoBox was turned off).

MLIB 203

Can't open file <file_name>. The most likely reason is that the
file cannot be found or does not exist at all.
1. Check if <file name> is spelled correctly and you are in the
appropriate working folder. All linker map files, associated objects
(PPC, X86) and SDF files (for multiprocessor mode only) must reside
in the same folder.

124

MLIB 204

Cannot find file <filename>.

not in the MATLAB search path.

The file probably does not exist or is

MLIB 205

Cannot load file <filename>.

<filename> exists.

Please check if the specified file

MLIB 206

Cannot write data into file <filename>.

which the file resides is probably full.

MLIB 207

Cannot read from file <filename>. Check if the file <filename>

exists and can be found on the MATLAB search path.

MLIB 208

RESERVED

MLIB 209

Cannot create file <filename>. Check if there is enough free

space on your hard disk. Alternatively, check if the file <filename>
already exists and is read-only. In this case it cannot be overwritten.

The destination disk on

Free space on your local disk.

Change the attribute of the existing file <filename>.

Change the name or the location of the file.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Troubleshooting

Invalid Variable Connections When Migrating to MATLAB

R2011a
If you update an existing Simulink model via the slupdate command
when migrating to MATLAB R2011a, the code generated for Lookup
Table blocks and for Lookup Table (2-D) blocks in the model differs
from the code generated with previous MATLAB releases. As a
consequence, the SDF/TRC file entries for Lookup Table blocks and
Lookup Table (2-D) blocks change.
The path information for Lookup Table and Lookup Table (2-D)
variables in an MLIB/MTRACE script becomes invalid when you migrate
to MATLAB R2011a and rebuild an updated model. To solve the
problem, you have to adapt the path information yourself.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

125

126

Troubleshooting

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Appendix

Where to go from here

Information in this section

MLIB/MTRACE Benchmarks

127

Impact of MLIB/MTRACE on Execution Time

130

MLIB/MTRACE Benchmarks
As described in Multiple MLIB/MTRACE Data Captures in One RealTime Application on page 76, the time needed for processor board
access from host applications mainly depends on the type of
connection; for example, whether a PC/AT bus interface or a TCP/IP
network is used. It is impossible to provide general benchmarks for
network connections. Network performance depends on many factors
such as the type of Ethernet cards and cabling, the implementation of
the TCP/IP software, the TCP/IP packet length, and the actual traffic. In
typical configurations, network access is 3-10 times slower than PC/AT
bus access.
The following two graphs show the performance of MLIB/MTRACE
with a Pentium II 400 MHz host connected to a DS1005/400 MHz
processor board via a bus interface.

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

127

Appendix

The first graph shows the number of 64-bit words transferred per
second (N) with mlib('Read', var) versus the number of words to be
read (n = var.length):

The following graph shows the number of transferred 64-bit words

transferred per second (N) via mlib(Write, var, Data, fval) versus the
number of words to be written (n = var.length):

128

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Appendix

The following graph shows the execution time of the MLIB/MTRACE

FetchData function with a Pentium II 400 MHz PC connected to a
DS1005/400 MHz processor board via a bus interface. Nearly identical
results are achieved for other dSPACE boards.

Measurement details

For further information, refer to Impact of MLIB/MTRACE on Execution

Time on page 130.

Execution times

Execution times are available for:

Related topics

MLIB/MTRACE with DS1103 on page 130

MLIB/MTRACE with DS1401 on page 130

MLIB/MTRACE with DS1005 on page 131

n

Execution times for DS1104 and DS1006 are not

available at all.

Execution times for DS1103, DS1401 and DS1005

have not been measured with the latest board
revisions. The processor clock rates of the boards used
are documented.

Basics
Impact of MLIB/MTRACE on Execution Time on page 130
References
MLIB/MTRACE with DS1103 on page 130

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

129

Appendix

Impact of MLIB/MTRACE on Execution Time

In addition to the execution time necessary to calculate model
equations and perform I/O routines of a real-time application, the
service macro host_service, which must be inserted into the interrupt
service routine to let MLIB/MTRACE work (see Getting Started
on page 17), requires additional execution time. The amount of
execution time needed depends on many parameters, such as the
number and kind of variables to trace, the chosen trigger options, and
the memory location used for program data and captured data.
Insufficient execution time will result in processor overflow when the
data acquisition service is performed.
The capture of directly accessed variables requires different execution
times than the capture of variables referenced by pointers. In the
following tables, the symbol nptr indicates the number of variables
accessed by pointers, and nvar is used for the number of the directly
accessed variables.
MLIB/MTRACE with
DS1103

MLIB/MTRACE with
DS1401

130

The execution times for the host_service code were measured on a

333 MHz DS1103 processor board. The program code and data were
loaded into local memory. The traced data is written to the trace buffer
located in the global memory. The following table shows the
measurement results.
Trigger State

Execution Time of host_service in sec

pre-trigger

ca. 3.2 + nvar 0.061 + nptr 0.069

post-trigger

ca. 3.2 + nvar 0.061 + nptr 0.069

trigger OFF

ca. 3.2 + nvar 0.061 + nptr 0.069

The execution times for the host_service code were measured on a

200 MHz DS1401 processor board. The program code and data were
loaded into local memory. The following table shows the
measurement results.
Trigger State

Execution Time of host_service in sec

pre-trigger

ca. 4.9 + nvar 0.2 + nptr 0.2

post-trigger

ca. 5.3 + nvar 0.2 + nptr 0.2

trigger OFF

ca. 4.7 + nvar 0.2 + nptr 0.2

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

MLIB/MTRACE with
DS1005

Appendix

The execution times for the host_service code were measured on a

400 MHz DS1005 processor board. The program code and data were
loaded into the cached area of the global memory. The collected data
was written to the trace buffer located in the non cached area of the
global memory. This is the predefined configuration. The following
table shows the measurement results.
Trigger State

Execution Time of host_service in sec

pre-trigger

ca. 2.3 + nvar 0.083 + nptr 0.083

post-trigger

ca. 2.3 + nvar 0.083 + nptr 0.083

trigger off

ca. 2.3 + nvar 0.083 + nptr 0.083

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

131

132

Appendix

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

Index

Index

A
accessing matrices
MLIB/MTRACE 78
adc_XXXX.m 96
address evaluation functions 33
administrative functions 33
appendix
MLIB/MTRACE 127

B
benchmarks
MLIB/MTRACE 127
built-in Simulink blocks 79

C
CaptureState 34
chkerr_XXXX.m 96
clear command (MLIB) 18
cont_XXXX.m 96

D
data acquisition
real-time 127
data acquisition functions 33
default service
MLIB/TRACE 35
default service group 35
demo files
MLIB 15
descriptor 32
device driver
register 17

E
error messages
MLIB/MTRACE 109
exectime_XXXX.m 96
execution time 130

F
FetchData 36
function reference
MLIB/MTRACE 33
functions supporting MLIB/MTRACEgenerated MAT-files 33

G
Get 40
GetApplInfo 42
GetBoardInfo 44
GetMapVar 45

getting started
MLIB/MTRACE 17
GetTrcVar 47

H
help
MLIB/MTRACE 33
host_service
MLIB 18

I
installation
MLIB 13
process for MLIB 14
interrupt functions 33
introduction 11
Intrpt 51
IsApplRunning 52

L
limitations
accessing variables 108
number of applications 108
simultaneous platform access 108
LoadFile 53
LoadVarFile 55
LockProgram 56
lookup-table 78

M
master_cmd_server 19
MAT file 127
matrices
accessing with MLIB/MTRACE 78
MLIB/MTRACE
appendix 127
benchmarks 127
default service 35
error messages 109
function reference 33
functions 31
getting started 17
help 33
multiple data captures 76
naming conventions 32
programming 75
programming examples 88
troubleshooting 109
with DS1005 131
with DS1103 130
with DS1401 130

mlib_1dlookup 80
Init 80
ReadInput 81
ReadOutput 81
WriteInput 82
WriteOutput 82
mlib_2dlookup 83
Init 83
ReadCol 84
ReadRow 84
ReadTable 84
WriteCol 84
WriteRow 84
WriteTable 85
mlib_matrix 86
Init 86
Read 87
Write 87
mlibini.m 17
multiple data captures
MLIB/MTRACE 76
Multiprocessing 57
multiprocessor systems
MLIB/MTRACE 23

N
network 127

P
programming
MLIB/MTRACE 75
programming examples
MLIB/MTRACE 88

R
Read 57
read functions 33
RTI 93

S
SearchMap 59
SearchTrc 60
SelectBoard 61
service group
default 35
definition 25
Set 62
SetVarProperty 67
S-functions 79
Simulink 93
StartCapture 69
StopCapture 71

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011

133

Index

system requirements
MLIB 13

T
table look-up 78
TCP/IP network 127
time stamp 18
trace buffer 18
TraceView utility 127
TRC file 48
generated by RTI 103
trigger_xxxx.m 96
TriggerState 72
troubleshooting
MLIB/MTRACE 109

U
UnlockProgram 73

W
Write 73
write functions 33

134

MLIB/MTRACE MATLAB-dSPACE Interface Libraries

May 2011