Module

3
Embedded Systems I/O
Version 2 EE IIT, Kharagpur 1
 Lesson
19
Analog Interfacing
Version 2 EE IIT, Kharagpur 2
Instructional Objectives
After going through this lesson the student would be able to

• Know the interfacing of analog signals to microcontrollers/microprocessors

• Generating Analog Signals
• Designing AD and DA interfaces
• Various Methods of acquiring and generating analog data

Pre-Requisite
Digital Electronics, Microprocessors

19(I) Introduction
Fig.19.1 shows a typical sensor network. You will find a number of sensors and actuators
connected to a common bus to share information and derive a collective decision. This is a
complex embedded system. Digital camera falls under such a system. Only the analog signals are
shown here. Last lesson discussed in detail about the AD and DA conversion methods. This
chapter shall discuss the inbuilt AD-DA converter and standalone converters and their
interfacing.

Fig. 19.1 The Analog Interfacing Network

Version 2 EE IIT, Kharagpur 3

 Fig. 19.2 The Analog-Digital-Analog signal path with real time processing

Different Stages of Fig.19.2

Stage-1 Signal Amplification and Conditioning; Stage-2 Anti-aliasing Filter; Stage-3 Sample and
Hold;
Stage-4 Analog to Digital Converter; Stage-5 Digital Processing and Data manipulation in a
Processor;
Stage-6 Processed Digital Values are temporarily stored in a latch before D-A conversion;
Stage-7 Digital to Analog Conversion; Stage-8 Removal of Glitches and Spikes; Stage-8 Final
Low pass filtering

19(II) Embedded AD Converters in Intel 80196

Fig.19.3 shows the block diagram of the AD converter inbuilt to 80196 embedded processor. The
details of the subsystems are given as follows:
Analog Inputs
EPA or PTS
Command
VREF ANGND
Analog Mux
Control
Successive Logic
Sample Approximation
and Hold A/D
Converter
Status

AD_RESULT AD_COMMAND AD_TIME AD_TEST

† Multiplexed with port inputs

Fig. 19.3 The block diagram of the Internal AD converter

Analog Inputs: There are 12 input channels which are multiplexed with the Port P0 and Port P1
of the processor.

Version 2 EE IIT, Kharagpur 4

ANGND: It is the analog ground which is separately connected to the circuit from where analog
voltage is brought inside the processor.

Vref: It is reference voltage which decides the range of the input voltage. By making it negative
bipolar inputs can be used.

EPA: Event Processor Array

Control applications often require high-speed event control. For example, the controller may
need to periodically generate pulse-width modulated outputs or an interrupt. In another
application, the controller may monitor an input signal to determine the status of an external
device. The event processor array (EPA) was designed to reduce the CPU overhead associated
with these types of event control. This chapter describes the EPA and its timers and explains how
to configure and program them. The EPA can control AD converter such as generating timing
pulses, start conversion signals etc.

PTS: Peripheral Transaction Server

The microcontroller’s interrupt-handling system has two components: the programmable
interrupt controller and the peripheral transaction server (PTS). The programmable interrupt
controller has a hardware priority scheme that can be modified by the software. Interrupts that go
through the interrupt controller are serviced by interrupt service routines that you provide. The
upper and lower interrupt vectors in special-purpose memory contain the interrupt service
routines’ addresses. The peripheral transaction server (PTS), a microcoded hardware interrupt
processor, provides high-speed, low-overhead interrupt handling; it does not modify the stack or
the Processor Status Word. The PTS supports seven microcoded routines that enable it to
complete specific tasks in lesser time than an equivalent interrupt service routine can. It can
transfer bytes or words, either individually or in blocks, between any memory locations; manage
multiple analog-to-digital (A/D) conversions; and transmit and receive serial data in either
asynchronous or synchronous mode.

Analog Mux: Analog Multiplexer

It selects a particular analog channel for conversion. Only after completing conversion of one
channel it switches to subsequent channels.

The associated Registers

AD_COMMAND register
This register selects the A/D channel, controls whether the A/D conversion starts immediately or
is triggered by the EPA, and selects the operating mode.

AD_RESULT
For an A/D conversion, the high byte contains the eight MSBs from the conversion, while the
low byte contains the two LSBs from a 10- bit conversion (undefined for an 8-bit conversion),
indicates which A/D channel was used, and indicates whether the channel is idle. For a

Version 2 EE IIT, Kharagpur 5

threshold-detection, calculate the value for the successive approximation register and write that
value to the high byte of AD_RESULT. Clear the low byte or leave it in its default state.

AD_TEST A/D Conversion Test

This register specifies adjustments for zero-offset errors.

AD_TIME A/D Conversion Time

This register defines the sample window time and the conversion time for each bit.

INT_MASK Interrupt Mask

The AD bit in this register enables or disables the A/D interrupt. Set the AD bit to enable the
interrupt request.

INT_PEND Interrupt Pending

The AD bit in this register, when set, indicates that an A/D interrupt request is pending.

A/D Converter Operation

An A/D conversion converts an analog input voltage to a digital value, stores the result in
the AD_RESULT register, and sets the A/D interrupt pending bit. An 8-bit conversion provides
20 mV resolution, while a 10-bit conversion provides 5 mV resolution. An 8-bit conversion takes
less time than a 10-bit conversion because it has two fewer bits to resolve and the comparator
requires less settling time for 20 mV resolution than for 5 mV resolution. Either the voltage on
an analog input channel or a test voltage can be converted. Converting the test inputs is used to
calculate the zero-offset error, and the zero-offset adjustment is used to compensate for it. This
feature can reduce or eliminate off-chip compensation hardware. Typically, the test voltages are
converted to adjust for the zero-offset error before performing conversions on an input channel.
The AD_TEST register is used to program for zero-offset adjustment. A threshold-detection
compares an input voltage to a programmed reference voltage and sets the A/D interrupt pending
bit when the input voltage crosses over or under the reference voltage. A conversion can be
started by a write to the AD_COMMAND register or it can be initiated by the EPA, which can
provide equally spaced samples or synchronization with external events.

Once the A/D converter receives the command to start a conversion, a delay time elapses
before sampling begins. During this sample delay, the hardware clears the successive
approximation register and selects the designated multiplexer channel. After the sample delay,
the device connects the multiplexer output to the sample capacitor for the specified sample time.
After this sample window closes, it disconnects the multiplexer output from the sample capacitor
so that changes on the input pin will not alter the stored charge while the conversion is in
progress. The device then zeros the comparator and begins the conversion. The A/D converter
uses a successive approximation algorithm to perform the analog-to-digital conversion. The
converter hardware consists of a 256-resistor ladder, a comparator, coupling capacitors, and a 10-
bit successive approximation register (SAR) with logic that guides the process. The resistive
ladder provides 20 mV steps (VREF = 5.12 volts), while capacitive coupling creates 5 mV steps
within the 20 mV ladder voltages. Therefore, 1024 internal reference voltage levels are available
for comparison against the analog input to generate a 10-bit conversion result. In 8- bit
conversion mode, only the resistive ladder is used, providing 256 internal reference voltage
levels. The successive approximation conversion compares a sequence of reference voltages to

Version 2 EE IIT, Kharagpur 6

the analog input, performing a binary search for the reference voltage that most closely matches
the input. The ½ full scale reference voltage is the first tested. This corresponds to a 10-bit result
where the most-significant bit is zero and all other bits are ones (0111111111). If the analog
input was less than the test voltage, bit 10 of the SAR is left at zero, and a new test voltage of ¼
full scale (0011111111) is tried. If the analog input was greater than the test voltage, bit 9 of
SAR is set. Bit 8 is then cleared for the next test (0101111111). This binary search continues
until 10 (or 8) tests have occurred, at which time the valid conversion result resides in the
AD_RESULT register where it can be read by software. The result is equal to the ratio of the
input voltage divided by the analog supply voltage. If the ratio is 1.00, the result will be all ones.
The following A/D converter parameters are programmable:
• conversion input — input channel
• zero-offset adjustment — no adjustment, plus 2.5 mV, minus 2.5 mV, or minus 5.0 mV
• conversion times — sample window time and conversion time for each bit
• operating mode — 8- or 10-bit conversion or 8-bit high or low threshold detection
• conversion trigger — immediate or EPA starts

19(III) The External AD Converters (AD0809)

START CLOCK

8-BIT A/D
CONTROL & END OF
TIMING CONVERSION
(INTERRUPT)
8
CHANNELS
MULTIPLE-
8 ANALOG
XING
INPUTS S.A.R
ANALOG
SWITCHES
COMPARATOR TRI-
STATE
OUTPUT 8-BIT
LATCH OUTPUTS
BUFFER

SWITCH TREE

3-BIT ADDRESS ADDRESS

LATCH
AND
ADDRESS DECODER
LATCH ENABLE 256R REGISTOR
LADDER

VCC GND REF(+) REF(-) OUTPUT

ANABLE

Fig. 19.4 The internal architecture of 0809 AD converter

Version 2 EE IIT, Kharagpur 7

 IN3 1 28 IN2
IN4 2 27 IN1
IN5 3 26 IN0
IN6 4 25 ADD A
IN7 5 24 ADD B
START 6 23 ADD C
EOC 7 22 ALE
-5
2 8 21 2-1MSB
OUTPUT ENABLE 9 20 2-2
CLOCK 10 19 2-3
VCC 11 18 2-4
VREF (+) 12 17 2-8LSB
GND 13 16 VREF (-)
-7
2 14 15 2-6

Fig. 19.5 The signals of 0809 AD converter

Functional Description

Multiplexer
The device contains an 8-channel single-ended analog signal multiplexer. A particular input
channel is selected by using the address decoder. Table 1 shows the input states for the address
lines to select any channel. The address is latched into the decoder on the low-to-high transition
of the address latch enable signal.

TABLE 1
SELECTED ANALOG ADDRESS LINE
CHANNEL C B A
IN0 L L L
IN1 L L H
IN2 L H L
IN3 L H H
IN4 H L L
IN5 H L H
IN6 H H L
IN7 H H H

The Converter
This 8-bit converter is partitioned into 3 major sections: the 256R ladder network, the successive
approximation register, and the comparator. The converter’s digital outputs are positive true. The

Version 2 EE IIT, Kharagpur 8

256R ladder network approach (Figure 1) was chosen over the conventional R/2R ladder because
of its inherent monotonicity, which guarantees no missing digital codes. Monotonicity is
particularly important in closed loop feedback control systems. A non-monotonic relationship
can cause oscillations that will be catastrophic for the system. Additionally, the 256R network
does not cause load variations on the reference voltage.

CONTROLS FROM S.A.R.

REF(+)

1½ R

256 R TO
COMPARATOR
INPUT
R

½R

REF(-)

Fig. 19.6 The 256R ladder network

The bottom resistor and the top resistor of the ladder network in Fig.19.6 are not the same value
as the remainder of the network. The difference in these resistors causes the output characteristic
to be symmetrical with the zero and full-scale points of the transfer curve. The first output
transition occur when the analog signal has reached +1⁄2 LSB and succeeding output transitions
occur every 1 LSB later up to full-scale. The successive approximation register (SAR) performs
8-iterations to approximate the input voltage. For any SAR type converter, n-iterations are
required for an n-bit converter. Fig.19.7 shows a typical example of a 3-bit converter. The A/D
converter’s successive approximation register (SAR) is reset on the positive edge of the start
conversion (SC) pulse. The conversion is begun on the falling edge of the start conversion pulse.
A conversion in process will be interrupted by receipt of a new start conversion pulse.
Continuous conversion may be accomplished by tying the end-of-conversion (EOC) output to the
SC input. If used in this mode, an external start conversion pulse should be applied after power
up. End-of-conversion will go low between 0 and 8 clock pulses after the rising edge of start
conversion. The most important section of the A/D converter is the comparator. It is this section
which is responsible for the ultimate accuracy of the entire converter.

Version 2 EE IIT, Kharagpur 9

 111 +1/2 LSB INFINITE R
FULL-SCALE 111 TOTAL PERFECT CO
110 IDEAL CURVE ERROR = 1/2 LSB UNADJUSTED
A/D OUTPUT CODE

A/D OUTPUT CODE

110 ERROR IDEAL 3-BIT CODE
101
101
100
100 -1 LSB
011 ABSOLUTE
NONLINEARITY = 1/2 LSB 011
ACCURACY
010
NONLINEARITY = -1/2 LSB 010 -1/2 LSB
001 001 QUANTIZATION
ZERO ERROR = -1/4 LSB ERROR
000 VIN 000
0/8 1/8 2/8 3/8 4/8 5/8 6/8 7/8 VIN
0/8 1/8 2/8 3/8 4/8 5/8 6/8 7/8
VIN AS FRACTION OF FULL-SCALE VIN AS FRACTION OF FULL-SCALE

Fig. 19.7 The 3-bit AD Converter Resolution

Interface to a typical Processor

Fig.19.8 shows the layout for interface to a processor with 16-address lines(AD0-AD15), read
and write lines and 8-data lines (DB0-DB7). The address lines are divided into two groups. AD0-
AD2 are used to select the analog channel. The ALE signal of the ADC is used to latch the
address on the lines A0-A2 for keeping a particular channel selected till the end of conversion.
The other group (AD3-AD15) are decoded and combined with Read and Write signals to
generate the START, ALE and OE (output enable) signals. A write operation starts the ADC.
The EOC signal can be used to initiate an interrupt driven data transfer. The interrupt service
subroutine can read the data through DB0-DB7 and initiate the next conversion by subsequent
write operation. Fig.19.9 shows the timing diagram with system clock (not the ADC clock).

READ

INTERRUPT
500 kHz CLK 0E
ADDRESS 5.000V VREF (+) E0C INTERRUPT
DECODE
(AD4 – AD15)* 0.000V VREF (-)
2-1 DB7 MSB
START 2 -2
DB6
ALE 2-3 DB5
WRITE
2-4 DB4
AD0 A 2-5 DB3
ADC0808
AD1 B ADC0809 2 -6
DB2
-7
AD2 C 2 DB1
2-8 DB0 LSB
5V SUPPLY

VCC In7 VIN 8

GND

GROUND 0-5V ANALOG

INPUT RANGE

In0 VIN 1

Fig. 19.8 Interface to a typical processor

Version 2 EE IIT, Kharagpur 10
The timing Diagram (Fig.19.9)
The address latch enable signal and the start conversion are almost made high at the same time as
per the connections in Fig.19.8. The analog input should be stable across the hold capacitor for
the conversion time(tc). The digital outputs remain tri-stated till the output is enabled externally
by the Output Enable(OE) signal. The comparator input changes by the SAR counter and switch
tree through the ladder network till the output almost matches the voltage ate the selected analog
input channel.
Important Specifications
8- time-multiplexed analog channels
Resolution 8 Bits
Supply 5 VDC
Average Power consumption 15 mW
Conversion Time 100 μs

19(IV) The DA Converter DAC0808

The DAC0808 is an 8-bit monolithic digital-to-analog converter (DAC). Fig.19.9 shows the
architecture and pin diagram of such a chip.

MSB LSB
A1 A2 A3 A4 A5 A6 A7 A8

RANGE
CURRENT SWITCHES I0
CONTROL

R-2R LADDER BIAS CIRCUIT GND

VREF (+)
NPN CURRENT VCC
SOURCE PAIR
VREF (-)
REFERENCE COMPEN
CURRENT AMP

VEE

Version 2 EE IIT, Kharagpur 11

 NC (NOTE 2) 1 16
COMPENSATION

GND 2 15
VREF(-)

VEE 3 14
VREF(+)

I0 4 13
VCC
DAC0808
MSB A1 5 12
A8 LSB
6 11
A2 A7

A3 7 10
A6

A4 8 9
A5

Fig. 19.9 The DAC 0808 Signals

The pins are labeled A1 through A8, but note that A1 is the Most Significant Bit, and A8 is the
Least Significant Bit (the opposite of the normal convention). The D/A converter has an output
current, instead of an output voltage. An op-amp converts the current to a voltage. The output
current from pin 4 ranges between 0 (when the inputs are all 0) to Imax*255/256 when all the
inputs are 1. The current, Imax, is determined by the current into pin 14 (which is at 0 volts).
Since we are using 8 bits, the maximum value is Imax*255/256. The output of the D/A converter
takes some time to settle. Therefore there should be a small delay before sending the next data to
the DA. However this delay is very small compared to the conversion time of an AD Converter,
therefore, does not matter in most real time signal processing platforms. Fig.19.10 shows a
typical interface.

Version 2 EE IIT, Kharagpur 12

 VCC = 5V

13
5 14 5.000k
MSBA1 10.000V = VREF
6
A2 5k
7 15
A3 5.000k
8 2
DIGITAL A4
9 DAC0808
INPUTS A5
10 4
A6 -
11
A7
12 16 LF351 V0
LSB A8 OUTPUT
3 0.1 μF +

VEE = -15V

Fig. 19.10 Typical connection of DAC0808

LF351 is an operational amplifier used as current to proportional voltage converter. The 8-digital
inputs at A8-A1 is converted into proportional current at pin no.4 of the DAC. The reference
voltages(10V) are supplied at pin 14 and 15(grounded through resistance). A capacitor is
connected across the Compensation pin 16 and the negative supply to bypass high frequency
noise.
Important Specifications
±0.19% Error
Settling time: 150 ns
Slew rate: 8 mA/μs
Power supply voltage range: ±4.5V to ±18V
Power consumption: 33 mW @ ±5V

19(V) Conclusion
In this lesson you learnt about the following
The internal AD converters of 80196 family of processor
The external microprocessor compatible AD0809 converter
A typical 8-bit DA Converter

Both the ADCs use successive approximation technique. Flash ADCs are complex and therefore
generate difficult VLSI circuits unsuitable for coexistence on the same chip. Sigma-Delta need
very high sampling rate.

Version 2 EE IIT, Kharagpur 13

Question Answers
Q.1. What are the possible errors in a system as shown in Fig. 19.2?

Ans:

Stage-1 Signal Amplification and Conditioning This can also amplify the noise.

Stage-2 Anti-aliasing Filter Some useful information such as transients in the real systems
cannot be captured.

Stage-3 Sample and Hold The leakage and electromagnetic interference due to switching

Stage-4 Analog to Digital Converter Quantization error due to finite bit length

Stage-5 Digital Processing and Data manipulation in a Processor: Numerical round up errors due
to finite word length and the delay caused by the algorithm.

Stage-6 Processed Digital Values are temporarily stored in a latch before D-A conversion: Error
in reconstruction due to zero-order approximation

Q.2 Why it is necessary to separate the digital ground from analog ground in a typical ADC?

Ans: Digital circuit noise can get to analogue signal path if separate grounding systems are not
used for digital and analogue parts. Digital grounds are invariably noisier than analog grounds
because of the switching noise generated in digital chips when they change state. For large
current transients, PCB trace inductances causes voltage drops between various ground points on
the board (ground bounce). Ground bounce translates into varying voltage level bounce on signal
lines. For digital lines this isn't a problem unless one crosses a logic threshold. For analog it's just
plain noise to be added to the signals.

Version 2 EE IIT, Kharagpur 14