Analog Devices an273 Application Notes

AN-273

a

APPLICATION NOTE

One Technology Way • P.O. Box 9106 • Norwood, MA 02062-9106 • 781/329-4700 • World Wide Web Site: http://www.analog.com

Use of the AD590 Temperature Transducer

in a Remote Sensing Application

by Paul Klonowski

INTRODUCTION

The AD590 is a two-terminal integrated circuit tempera­ture transducer that produces an output proportional to
absolute temperature. For supply voltages between +4 V
and +30 V the device acts as a high impedance, constant

current source supplying 1 µA/K. Laser trimming of the

chip’s thin-film resistors is used to calibrate the device

to an output of 298.2 µA at 298.2K (+25°C).

A typical application for the AD590 is a remote tempera­ture-to-current transducer. Figure 1 shows a thermom-

eter circuit that measures temperature from –55°C to
+100°C and whose output voltage is 100 mV/°C. Since

the AD590 measures absolute temperature (its nominal

output is 1 µA/K), the output must be offset by 273.2 µA

in order to read out in degrees Celsius. The output cur-

rent of the AD590 flows through a 1 kΩ resistance, devel-

oping a voltage of 1 mV/K. The output of the AD580 2.5 V
reference is divided down by resistors to provide a

273.2 mV offset, which is subtracted from the voltage

across the 1 kΩ resistor by an AD524 instrumentation

amplifier. The amplifier provides a gain of 100, so that

the output range corresponding to –55°C to +100°C is
–5.5 V to +10 V (100 mV/°C). An operational amplifier can

substitute for the instrumentation amplifier, although
care must be taken when designing with the
op amp since the gain at the two input terminals will be
different.

THE PROBLEM

A question often asked of Analog Devices Applications
Engineers by customers using the AD590 in a remote
temperature-to-current application is, “How long can I
make the cable and how can I eliminate any noise that
the cable picks up?” Experiments were performed in an
effort to provide some guidelines for answering this
question using the circuit of Figure 1 with a 1000'
shielded, though initially ungrounded, twisted pair cable
(Belden 9461, style 2092). In order to duplicate actual
conditions the experiments were performed in an indus­trial environment.

TYPES OF NOISE

There are three basic types of noise inherent in a
data-acquisition system. The first type is

noise

: noise received with the original signal and indis-

tinguishable from it. The second type is

transmitted

intrinsic noise:

noise generated within the devices used in a circuit, e.g.,
resistors, op amps, etc. Included in this category are
Johnson, shot and popcorn noise. The third type is

induced noise

: noise picked up from the outside world

and coupled into the circuit. This application note dis­cusses methods of reducing induced noise, which is the
only form of noise that can be influenced by choices of
wiring and shielding.

AD590

I

T

I

T

I

T

+15V

+

7V

–

0.1% LOW

TCR RESISTOR

1mV/K

<

Figure 1. Thermometer Circuit

1kV

AD580

909kV

200V

1kV

G = 100

RG

AD524

2

INSTRUMENTATION AMP

GAIN = 100, 100mV/8C

2kV

E

O

AN-273

NOISE FACTORS

Three elements are involved in any noise problem. The
first is the source of the noise. Possible noise sources
include AM radio signals, logic signals, magnetic fields
and power line transients. The second element is the
coupling medium. That is, how is the noise source en­tering the circuit? Possible coupling mediums include a
common circuit impedance (Figure 2), stray capacitance
(Figure 3) and mutual inductance (Figure 4). A brief de­scription of each follows.

Common impedance noise is developed by an imped­ance common to several circuits. This might occur, as
shown in Figure 2, when a pulse output source and an
op amp’s reference terminal are both connected to a
“ground” point having tangible impedance to the
power supply terminal. The noisy return current of Cir­cuit 1 develops a voltage, V

, across the common im-

NOISE

pedance Z which will appear as a noise signal to Circuit

2. Possible solutions to this problem include proper cir­cuits for distributing power and the use of isolation
transformers and optical isolators.

Capacitively-coupled noise is produced by stray capaci­tance which couples the voltage changing noise source
into high impedance circuits, as shown in Figure 3. The
nature of the impedance Z determines the shape of the
response. Methods of reducing capacitively-coupled
noise include reducing the noise source, properly imple­menting shields and reducing the stray capacitance.

Magnetically coupled noise is produced by mutual in­ductance and can occur, for example, in an incorrectly
shielded cable as shown in Figure 7. Figure 4 is a simple,
model of this incorrectly shielded cable, where L
sents the inductance of the shield, L

represents the in-

C

repre-

S

ductance of one of the center conductors, and L
represents the mutual inductance between the two. The
noise current l(t) flows through L

and establishes a

S

magnetic flux; this time-varying flux also surrounds L
and produces a voltage V
rate of change of the current I(t) flowing through L

proportional to the time

NOISE(t)

. This

S

voltage can be expressed as

dI(t)

V

NOISE(t)

=

LM

dt

The third element involved in any noise problem is the
receiver, or the circuit susceptible to the noise. It is im­portant to understand the role of each of the three ele­ments (the noise source, the coupling medium, and the
receiver) in order to solve the noise problem*. In this
experiment it was determined that the noise sources
were 60 Hz pickup and AM radio signals, the coupling
medium was stray capacitance and the receiver was the
AD524.

L

M

L

I

(t)

L

S

V

C

NOISE(t)

M

C

CIRCUIT 2

= I

V

NOISE

NOISE

POWER SUPPLY COMMON

V

S1

CIRCUIT 1

+

I

Z

COMMON

IMPEDANCE

NOISE

Figure 2. Common Impedance Noise

C

S

V

NOISE

SOURCE

Z

Figure 3. Stray Capacitance Noise

Z

V

NOISE

Figure 4. Mutual Inductance Noise

INITIAL NOISE EFFECT

The photograph in Figure 5 shows the output of the cir­cuit in Figure 1 with the shield ungrounded and with the

+

V

S2

remote AD590 at 30°C. Ideally, the output should be a
3 V (100 mV/°C) dc signal. However, a 60 Hz signal result-

ing from an electric field has been capacitively coupled
into the circuit via the stray capacitance of the cable and
then amplified by a gain of 100. Note, however, that the
60 Hz signal is offset by a dc signal; when the output
voltage of the AD524 was measured with a dc voltmeter
the value read was 3.0 V. This is because the voltmeter
read the value of the dc signal due to the AD590 along
with the average value of the 60 Hz sine wave noise sig­nal. The average value of a sine wave is zero. In essence,
notwithstanding the interfering signal the average value
was correct. The accuracy of all the measurements were
verified through the use of an RTD measurement sys­tem; the AD590 under test was physically attached to
the RTD.

*An excellent article dealing with this subject is “Understanding Inter-

ference-Type Noise” by Alan Rich, found in Section 20 of the Analog
Devices Databook.

–2–