Proceedings of the Technical Sessions, 18 (2002) 1-6 1

Institute of Physics – Sri Lanka

Performance of parallel port based remote data acquisition

systems in real-time data processing

P.C.C.P.W. Goonetilleke and D.U.J. Sonnadara

Department of Physics, University of Colombo, Colombo 3

ABSTRACT
A parallel port based remote data acquisition and process control system was built without on board
RAM and programmable timer to test the effect on the performance when processing real-time data
on a network environment. The controlling software was developed using Delphi and Assembler
languages to achieve high speeds. The developed software use TCP/IP protocol for network
operations.

It was found that the overall performance depends primarily on the parallel port. For different types
of motherboards, the speed of the data transfer varied between 460 Kbps and 890 Kbps. A
minimum deviation was observed when there were no processor-bounded applications running in
the system concurrently. The port speed did not depend on the processor type or the processor
speeds.

For the network transfer tests, client-server architecture was used. In order to increase the
throughput, data compression (Zip method) was used. A data stream of 4.5 Kb (sampled data
packet of 1500 byte) could be reduced to a 427±27 bytes with this technique. The compression and
de-compression time was negligible compared to the computer/network speeds. The client was able
to communicate with the server at an approximate speed of 1.3 Kbps on a network with fiber optic
backbone, which has a connection speed of 10 Mbps.

1. INTRODUCTION

Today, computers or micro-based system have become extremely powerful and versatile,
they are heavily used in many industrial applications and research activities around the
world. However, unlike most other countries the hardware capabilities of these systems are
not fully utilized in Sri Lanka due to the lack of knowledge especially in the areas of data
acquisition and process control.

To acquire data from experiments, researches have to use plug-in cards or especially
designed instrument systems that are extremely expensive according to the local financial
standards. Thus, most Universities and research organizations in Sri Lanka do not posses

Performance of parallel port based remote data acquisition systems in real-time data processing
Proceedings of the Technical Sessions, 18 (2002) 1-6 2
Institute of Physics – Sri Lanka

automated systems to control their experiments (unless it is a donation). To make the

matters worse, most Universities do not provide training in handling microprocessor based
systems or exposed students to such methodologies. This situation could be changed if the
University teachers take the initiative to conduct research in the development of low cost
data acquisition and process control systems that can be used in local industrial applications
and research activities. Today this may be a matter of combining already existing
technology and knowledge in the correct manner to produce the desired results.

Usually, most general-purpose I/O cards have its own multiple digital input/output ports,
multi channel analog input/outputs digital gain controllers, programmable reference
voltages, internal memory, ADC, DAC etc. Developing such cards with such capabilities is
not difficult for lower speeds (less than 10 MHz) that are required in most local
laboratories. However, the task becomes difficult for higher speeds (above 10MHz). Today,
data acquisition cards that are suitable for the local environment can be developed locally at
a reasonable cost for many applications.

The simplest form of all DAQ systems is the parallel-port based system, which is much
easier and cheaper to handle.

Today, data acquisition systems can be made to share by several users in a computer
network by designing the control software. Although it is not difficult to develop software
to achieve this objective, developing software without losing the speed is a difficult task.
To our knowledge, no literature is available that show the effectiveness of PC parallel port
based devices when operated over a network. The aim of this work is to measure the
performance of such devices when operating in a network environment.

2. DATA ACQUISITION HARDWARE

The parallel port of an IBM or compatible PC may be the easiest port available in a modern
computer that can be used to build low speed data acquisition systems. Especially, for the
parallel port, the software development is quite easy, the hardware requirements are very
few and it never conflicts with other type of devices connected to the computer.

The new extended parallel port is the most commonly used port in PC systems. Today this
port is capable of handling input/output data with speeds around 900 kbps (which vary with
the PC main board type).

Performance of parallel port based remote data acquisition systems in real-time data processing
Proceedings of the Technical Sessions, 18 (2002) 1-6 3
Institute of Physics – Sri Lanka

A block diagram of the DAQ system used in this work together with the other supporting
hardware are shown in Figure 1. The system consists of a computer to acquire, store and
display signals, a network to provide access to collected data to outside users and a DAQ
hardware interfaced to the computer through the parallel port.

Figure 1: Device Connection block diagram.

The main data acquisition system, a parallel port based external plug unit was built without
an on board RAM and a programmable timer to test the system performance when
processing real-time data. The controlling software was developed using Delphi and
Assembler languages to achieve high speeds which is essential in this work. The control
system software uses TCP/IP protocol for the network operations.

Figure 2 shows the block diagram of the developed DAQ system. All input and output
subsystems are connected to an internal data bus which connects to the external data bus
through a buffer. PC port is connected to the device buffers first to minimize the current
drawn from the PC parallel port. All the ports in this device (two digital output ports, one
digital input port, two analog output ports and two analog input ports) can be controlled by
the software. Switching between the ports is not transparent to the external connections
because the hardware system which is built into the device, keeps the last written values in
their buffer, until it is over written by another value at another time.

Performance of parallel port based remote data acquisition systems in real-time data processing
Proceedings of the Technical Sessions, 18 (2002) 1-6 4
Institute of Physics – Sri Lanka

Figure 2 : Device block diagram.

3. PERFORMANCE

3.1 Parallel port

The maximum device operational speed should directly proportional to the speed of the
parallel port. However the port speed may vary even in the same PC system especially
when other applications are running simultaneously. Figure 3 shows the port speed
variation with time for three different PC systems.
Intel 400 AMD 800 MHz Intel 833 MHz
900

800

700
Port reading sp eed (kb ps)

600

500

400

300

200

100

0
1 8 15 22 29 36 43 50 57 64 71 78 85 92 9 9 1 0 6 1 1 3 1 2 0 1 2 7 13 4

T im e (s )

Figure 3: Variation of port speed when other applications run simultaneously

Performance of parallel port based remote data acquisition systems in real-time data processing
Proceedings of the Technical Sessions, 18 (2002) 1-6 5
Institute of Physics – Sri Lanka

The measurements shows that the maximum port speed was not the same for all systems
and it did not depend on the processor speed (100MHz / 1000MHz or 2000MHz) or the
processor type (AMD / Intel). It was found that the port speed mainly depended on the main
board type (Main board hardware). Table 1 shows the values of measured speeds for
different main board types.

Table 1: Speed variation with motherboard types

Parameters Port speed (kbps)

Process Speed Primary Mother
OS Data Control Status
or (MHz) RAM (MB) board
AMD
799.498 130.3 MSI Win 98 SE 510 510 511
Duron
AMD
1133.14 261.6 MSI Win 98 SE 514 514 514
Athlon
Intel 730.892 63,7 Dell Win 98 SE 835 639 831
Intel 730.833 63,7 Dell Win 98 SE 833 640 827
Intel 334,041 32.2 Acer Win 97 EP 595 593 592
Intel 450.222 63.7 Dell Win 98 SE 759 573 750
Intel 797,249 63.6 Dell Win 98 SE 895 688 891
Intel 334.038 64.9 Acer Win 97 EP 594 593 593
AMD
700 64.9 GigaByte Win 98 SE 830 825 847
Duron
AMD
1000.12 130.3 GigaByte Win 98 SE 888 863 889
Athlon
AMD K5 133 32.8 UMC Win 95 840 832 861

3.2 Data transfer over the network

The computing environment of the Internet / Intranet is based on a client/server model. A

client/server network has two roles for computers: clients and servers. The server is a
computer that responds to requests for its services. The client is a computer that requests a
service from a server. In a client/server network, the server can be considered as the host
and the client can be considered as the guest.

For the DAQ device, server software and client software were developed separately and the
DAQ device was directly connected to the server. The device was controlled by the server
software while handling requests from many clients simultaneously.

Performance of parallel port based remote data acquisition systems in real-time data processing
Proceedings of the Technical Sessions, 18 (2002) 1-6 6
Institute of Physics – Sri Lanka

When a DAQ device is connected to a network, its speed mainly depends on the speed of
the network since network transfer speeds are always slower than the data acquisition
speed. The software developed for this work supports networks with TCP/IP protocol.

To increase the data transferring speed, data compression technique was used at the server
end (Zip compression). In order to do this, a digitized data with 1500 samples were taken
and sent as one packet to the client on request. The initial packet size was approximately
4.5 Kbytes. When this was compressed, it reduced to approximately 420 bytes. This
technique increased the data transferring speed drastically.

The device was able to transfer data at a speed of 1.3 Kbytes per second within the
University of Colombo network, which is controlled by a (slow) proxy server. Although the
expected values were much higher, due to the network traffic the device was not able to
achieve faster speeds than 1.3 Kbytes/per second. This could be reduced further if accessed
over the Internet due to the poor local telecommunication infrastructure.

4. DISCUSSION

A parallel port based remote data acquisition and process control system was built to study
the effect on the performance when processing real-time data on network environment. The
controlling software was developed using Delphi and Assembler languages to achieve
higher speeds.

It was found that the overall performance depends mainly on the parallel port speed which
is mainly depends on the main board type. The port speed did not depend on the processor
type or the processor speeds. For the network transfer speed tests, client-server architecture
was used. The client was able to communicate with the server at an approximate speed of
1.3 Kbps on a network with fiber optic backbone at the University of Colombo.

5. REFERENCES

1. http://www.torry.ru
2. http://www.epanorama.net
3. http://www.beyondlogic.org

Performance of parallel port based remote data acquisition systems in real-time data processing