Temperature Sensors Chapter7

8/6/2019 Temperature Sensors Chapter7

1/39

TE M P E R A T U R E S EN SO RS

7.1

SECTION 7

TEMP ER ATURE SENSORS

W a l t K es t er , J a m es B r y a n t , W a l t J u n g

INTRODUCTION

Measu rem ent of temp era tu re is critical in m odern electr onic devices, especially

expensive lapt op computer s an d other p ort able devices with den sely packed circuits

which dissipate consider able power in t he form of heat . Knowledge of system

tempera tu re can a lso be used to contr ol batter y cha rging as well as prevent da mage

to expensive microprocessors.

Compact high power port able equipmen t often ha s fan cooling to maint ain jun ction

temperatures at proper levels. In order to conserve battery life, the fan should only

operat e when necessar y. Accur at e cont rol of the fan requ ires a kn owledge of critical

temperatures from the appropriate temperature sensor.

n Monitoring

u Portable Equipment

u CPU Temperature

u Battery Temperature

u Ambient Temperature

n Compensation

u Oscillator Drift in Cellular Phones

u Thermocouple Cold-Junction Compensation

n Control

u Battery Charging

u Process Control

APPLICATIONS OF TEMPERATURE SENSORS

Figure 7.1

Accura te temperatu re measurements are r equired in man y other measurement

systems such as process control and instrumentation applications. In most cases,

becau se of low-level nonlinea r outpu ts, th e sensor out put mu st be properly

conditioned a nd am plified before furth er processing can occur .

Except for IC sensors, all temperatu re sensors ha ve nonlinear tra nsfer functions. In

th e pa st, complex an alog conditioning circuits wer e designed t o corr ect for th e sens or

nonlinear ity. These circuits often r equired ma nu al calibra tion and pr ecision

resist ors to achieve the desired a ccur acy. Today, however, sensor outpu ts m ay be


2/39


7.2

digitized directly by high resolution ADCs. Linear ization a nd calibra tion is then

perform ed digitally, ther eby redu cing cost an d complexity.

Resistance Temperature Devices (RTDs) are accurate, but require excitation current

an d ar e generally used in bridge circuits. Thermistors ha ve the m ost sensitivity but

ar e th e most non-linear . However, th ey are popular in portable applications such as

measu rement of batt ery temperat ur e and other critical temperat ures in a system.

Modern semiconductor temperature sensors offer high accuracy and high linearity

over a n opera ting ra nge of about 55C to +150C. Inter na l amplifiers can scale the

outpu t t o convenient va lues, such a s 10mV/C. They ar e a lso useful in cold-junction-

compen sat ion circuits for wide tem pera tu re r an ge ther mocouples. Semicondu ctor

tempera tur e sensors can be integrated into multi-function ICs which perform a

num ber of other ha rdwar e monitoring functions.

Figure 7.2 lists the m ost popular types of temperatu re tr ansdu cers an d th eir

characteristics.

TYPES OF TEMPERATURE SENSORS

THERMOCOUPLE RTD THERMISTOR SEMICONDUCTOR

Widest Range:

184C to +2300C

Range:

200C to +850C

Range:

0C to +100C

Range:

55C to +150C

High Accuracy and

Repeatability

Fair Linearity Poor Linearity Linearity: 1C

Accuracy: 1C

Needs Cold JunctionCompensation

RequiresExcitation

RequiresExcitation

Requires Excitation

Low-Voltage Output Low Cost High Sensitivity 10mV/K, 20mV/K,

or 1A/K TypicalOutput

Figure 7.2

THE R MOC OUPL E P RINCIP LES AND COLD-J UNCTION

COMPENSATION

Ther mocouples are sm all, rugged, relatively inexpensive, an d opera te over t he

widest ra nge of all temper at ur e sensors. They are especially useful for m ak ing

measu remen ts a t extrem ely high tempera tur es (up t o +2300C) in h ostile

environmen ts. They pr oduce only millivolts of outpu t, however, and requ ire

precision amplification for further processing. They also require cold-junction-

compen sat ion (CJ C) techn iques which will be discussed short ly. They ar e more

linear tha n m an y other sensors, and their n on-linearity ha s been well cha ra cterized.

Some comm on t her mocouples a re sh own in F igure 7.3. The most comm on m eta ls

used a re Ir on, Pla tinu m, Rhodium, Rhenium , Tungsten , Copper, Alum el (composed


3/39


7.3

of Nickel and Aluminum), Chromel (composed of Nickel and Chromium) and

Const an ta n (composed of Copper an d Nickel).

COMMON THERMOCOUPLES

JUNCTION MATERIALS

TYPICAL

USEFUL

RANGE (C)

NOMINAL

SENSITIVITY

(V/C)

ANSI

DESIGNATION

Platinum (6%)/ Rhodium-

Platinum (30%)/Rhodium

38 to 1800 7.7 B

Tungsten (5%)/Rhenium -

Tungsten (26%)/Rhenium

0 to 2300 16 C

Chromel - Constantan 0 to 982 76 E

Iron - Constantan 0 to 760 55 J

Chromel - Alumel 184 to 1260 39 K

Platinum (13%)/Rhodium-

Platinum

0 to 1593 11.7 R

Platinum (10%)/Rhodium-

Platinum

0 to 1538 10.4 S

Copper-Constantan 184 to 400 45 T

Figure 7.3

Figur e 7.4 shows the volta ge-tem pera tu re curves of th ree comm only used

th erm ocouples, referr ed to a 0C fixed-temper at ur e reference jun ction. Of the

th erm ocouples shown, Type J th erm ocouples ar e the m ost sensitive, producing thelargest output voltage for a given tempera tu re change. On the other ha nd, Type S

ther mocouples ar e th e least sensitive. These chara cteristics a re very importa nt to

consider wh en designing signal conditioning circuitry in t ha t t he t her mocouples'

relat ively low outpu t signals requ ire low-noise, low-drift, h igh-gain am plifiers.

To un derst an d th erm ocouple beha vior, it is necessary to consider th e non-linear ities

in their response to temperature differences. Figure 7.4 shows the relationships

between sensing junction t empera tur e and voltage out put for a nu mber of

th erm ocouple t ypes (in a ll cases, t he r eferen ce cold junction is maintained at 0C). It

is evident t hat the r esponses ar e not quite linear, but th e nat ur e of the n on-linearity

is not so obvious .

Figur e 7.5 sh ows h ow t he Seebeck coefficient (the change of outpu t volta ge with

change of sensor junction temp era tu re - i.e., th e first derivat ive of out put with

respect t o temperatu re) var ies with sensor junction t emperat ure (we are st ill

considerin g the case wher e the r eferen ce junction is ma inta ined at 0C).

When selecting a ther mocouple for mak ing measur ements over a par ticular ra nge of

tem pera tu re, we sh ould choose a th erm ocouple wh ose Seebeck coefficient varies a s

little as possible over that range.


4/39


7.4

THERMOCOUPLE OUTPUT VOLTAGES FORTYPE J, K, AND S THERMOCOUPLES

-250 0 250 500 750 1000 1250 1500 1750

-10

0

10

20

30

40

50

60

THERMOCOUPLEOUTPUTVOLTA

GE(mV)

TEMPERATURE (C)

TYPE J

TYPE K

TYPE S

Figure 7.4

THERMOCOUPLE SEEBECK COEFFICIENTVERSUS TEMPERATURE

-250 0 250 500 750 1000 1250 1500 1750

0

10

20

30

40

50

60

70

SEEBECKCOEFFICIENT-V/C

TEMPERATURE (C)

TYPE J

TYPE K

TYPE S

Figure 7.5


5/39


7.5

For exam ple, a Type J th erm ocouple ha s a Seebeck coefficient which var ies by less

th an 1V/C between 200 an d 500C, which ma kes it ideal for measu remen ts in th is

range.

Pr esenting these dat a on th ermocouples serves two purposes: First, F igure 7.4

illustra tes th e ra nge and sensitivity of the thr ee therm ocouple types so tha t t he

system designer can, a t a glance, determ ine tha t a Type S th ermocouple has t he

widest u seful tempera tur e ra nge, but a Type J th ermocouple is more sensitive.

Second, the Seebeck coefficients provide a quick guide to a thermocouple's linearity.

Using F igure 7.5, the syst em designer can choose a Type K ther mocouple for it s

linear Seebeck coefficient over th e r an ge of 400C t o 800C or a Type S over th e

ra nge of 900C to 1700C. The beh avior of a t her mocoup le's Seebeck coefficient isimport ant in applicat ions where variations of tempera tu re ra ther tha n absolute

ma gnitude ar e importa nt . These data a lso indicat e what perform ance is required of

th e a ssociated signal conditioning circuitr y.

To use t her mocouples successfully we must un derst an d th eir basic principles.

Consider th e diagrams in Figure 7.6.

THERMOCOUPLE BASICS

T1

Metal A

Metal B

ThermoelectricEMF

RMetal A Metal A

R = Total Circuit ResistanceI = (V1 V2) / R

V1 T1 V2T2

V1 V2

Metal B

Metal A Metal A

V1

V1

T1

T1

T2

T2

V2

V2

VMetal AMetal A

Copper Copper

Metal BMetal B

T3 T4

V = V1 V2, If T3 = T4

A. THERMOELECTRIC VOLTAGE

B. THERMOCOUPLE

C. THERMOCOUPLE MEASUREMENT

D. THERMOCOUPLE MEASUREMENT

I

V1

Figure 7.6

If we join two dissimilar metals at any temperature above absolute zero, there will

be a potentia l difference between th em (th eir "ther moelectric e.m.f." or "conta ct

potent ial") which is a function of the t emper at ur e of th e junction (Figure 7.6A). If we

join t he t wo wires a t t wo places, two jun ctions ar e form ed (Figure 7.6B). If the t wo

junctions a re a t different tempera tur es, there will be a net e.m.f. in th e circuit, and a

curr ent will flow determ ined by the e.m.f. and t he t ota l resistan ce in th e circuit

(Figure 7.6B). If we break one of the wires, t he volta ge across th e brea k will be


6/39


7.6

equa l to the n et t her moelectric e.m.f. of the circuit, a nd if we measu re t his voltage,

we can use it to calculate the temperature difference between the two junctions

(Figure 7.6C). We mu st always rem ember that a th erm ocouple m easures the

temperature difference between tw o jun ctions, n ot the absolute temperatu re at one

junction . We can only measure t he temper atu re at the m easur ing junction if we

kn ow th e tem pera tu re of the other junction (often called the "reference" junction or

th e "cold" junction).

But it is not so easy to measu re t he volta ge genera ted by a ther mocouple. Suppose

tha t we a tt ach a voltmeter to the circuit in Figure 7.6C (Figure 7.6D). The wires

at tached to the voltmeter will form fur ther ther mojunctions wher e they a re

at ta ched. If both these a dditional junctions a re a t t he sam e tempera tur e (it does not

matt er what temperatur e), then t he "Law of Intermediate Metals" states t hat they

will ma ke no net cont ribut ion t o th e total e.m.f. of th e system. If they ar e at

different temperatures, they will introduce errors. Since every pair of dissim ilar

m etals in contact generates a therm oelectric e.m .f. (including copper/solder ,

kovar /copper [kovar is the a lloy used for IC leadfra mes] and a luminu m/kovar [at th e

bond ins ide the I C]), it is obvious t ha t in pra ctical circuits t he pr oblem is even more

complex, and it is necessary to ta ke extreme care t o ensur e th at all the junctionpairs in t he circuitry ar ound a t herm ocouple, except th e measu rement an d reference

junctions th emselves, ar e at th e same tempera tu re.

Ther mocouples gener at e a voltage, albeit a very sma ll one, and do not require

excitat ion. As shown in Figur e 7.6D, however, t wo junctions (T1, the m easu rem ent

jun ction a nd T2, the r eferen ce jun ction) ar e involved. If T2 = T1, th en V2 = V1, and

th e outpu t volta ge V = 0. Ther mocouple out put voltages a re often defined with a

reference jun ction t emper at ur e of 0C (hen ce th e term coldor ice poin tjunction), so

th e th erm ocouple provides an outpu t volta ge of 0V at 0C. To ma inta in system

accuracy, the reference junction must therefore be at a well-defined temperature

(but not n ecessar ily 0C). A conceptu ally simple appr oach t o this n eed is shown in

Figur e 7.7. Alth ough an ice/wat er ba th is relat ively easy to define, it is quiteinconvenient t o mainta in.

Today a n ice-point reference, and its inconvenient ice/wat er ba th , is gener ally

replaced by electr onics. A tem pera tu re sen sor of an other sort (often a semicondu ctor

sensor, sometimes a ther mistor) measur es the tempera tur e of the cold junction a nd

is used t o inject a voltage int o the t her mocouple circuit which compen sat es for t he

difference between the actual cold junction temperature and its ideal value (usually

0C) as shown in F igure 7.8. Ideally, th e compen sat ion volta ge should be an exa ct

ma tch for t he differen ce voltage requ ired, which is why th e diagra m gives th e

volta ge a s f(T2) (a function of T2) rath er t han KT2, where K is a simple consta nt. In

pra ctice, since the cold junction is r ar ely more t ha n a few tens of degrees from 0C,an d genera lly var ies by little more th an 10C, a linea r a pproximat ion (V=KT2) to

th e more complex rea lity is sufficiently accur at e an d is wha t is often used. (The

expression for t he outpu t voltage of a t her mocouple with its m easu ring junction at

TC an d its referen ce at 0C is a polynomial of th e form V = K1T + K2T2 + K3T

3 +

..., but the values of the coefficients K2, K3, etc. are very sma ll for most comm on

types of thermocouple. References 8 and 9 give the values of these coefficients for a

wide ra nge of ther mocouples.)


7/39


7.7

CLASSICAL COLD-JUNCTION COMPENSATION USING ANICE-POINT (0C) REFERENCE JUNCTION

METAL A METAL A

METAL B

ICEBATH

0C

V(0C)

T1 V1

V1 V(0C)

T2

Figure 7.7

USING A TEMPERATURE SENSORFOR COLD-JUNCTION COMPENSATION

TEMPERATURECOMPENSATION

CIRCUIT

TEMPSENSOR

T2V(T2)T1 V(T1)

V(OUT)

V(COMP)

SAMETEMP

METAL A

METAL B

METAL A

COPPERCOPPER

ISOTHERMAL BLOCKV(COMP) = f(T2)

V(OUT) = V(T1) V(T2) + V(COMP)

IF V(COMP) = V(T2) V(0C), THEN

V(OUT) = V(T1) V(0C)

Figure 7.8


8/39


7.8

When electronic cold-junction compensation is used, it is common practice to

eliminat e th e additional t herm ocouple wire a nd t ermina te t he t herm ocouple leads in

the isothermal block in the arrangement shown in Figure 7.9. The Metal A-Copper

an d th e Metal B-Copper junctions, if at th e same t emperat ur e, are equivalent t o the

Meta l A-Meta l B th erm ocouple jun ction in Figur e 7.8.

TERMINATING THERMOCOUPLE LEADSDIRECTLY TO AN ISOTHERMAL BLOCK

TEMPERATURECOMPENSATION

CIRCUITTEMPSENSOR

METAL A

METAL B

COPPER

COPPER

COPPER

V(OUT) = V1 V(0C)

T1 V1

T2

T2

ISOTHERMAL BLOCK

Figure 7.9

The circuit in F igure 7.10 conditions t he out put of a Type K t her mocouple, while

providing cold-junction compen sat ion, for tem pera tu res between 0C a nd 250C. The

circuit opera tes from single +3.3V to +12V supplies an d ha s been designed t o

produce an out put voltage t ra nsfer char acter istic of 10mV/C.

A Type K t he rm ocouple exhibit s a Seebeck coefficient of appr oxima tely 41V/C;

ther efore, at the cold junction, th e TMP35 voltage outpu t sen sor with a t emperat ur e

coefficient of 10mV/C is u sed wit h R1 and R2 t o int rodu ce an opposing cold-jun ction

tem pera tu re coefficient of 41V/C. This prevent s t he isoth erm al, cold-junction

conn ection between th e circuit's print ed circuit board t ra ces an d th e th erm ocouple's

wires from introducing an er ror in th e measu red tem perat ure. This compensat ion

work s extrem ely well for circuit am bient t emper at ur es in th e ra nge of 20C to 50C.Over a 250C measur ement t emperat ur e ran ge, the t herm ocouple produces an

outpu t voltage chan ge of 10.151mV. Since th e r equired circuit's outpu t full-scale

volta ge cha nge is 2.5V, th e gain of th e circuit is set to 246.3. Choosing R4 equa l to

4.99k sets R5 equ al t o 1.22M. Since th e closest 1% value for R5 is 1.21M, a50k potentiometer is used with R5 for fine tr im of th e full-scale outpu t volta ge.Alth ough th e OP193 is a single-supply op am p, its out put sta ge is not r ail-to-ra il,

an d will only go down to about 0.1V above ground. F or th is rea son, R3 is added t o

th e circuit t o supply an out put offset volta ge of about 0.1V for a nomina l supply

volta ge of 5V. This offset (10C) mu st be subt ra cted when m ak ing measu rem ent s


9/39


7.9

referen ced to th e OP193 out put . R3 also provides an open th erm ocouple detection,

forcing the output voltage to greater than 3V should the thermocouple open.

Resistor R7 balan ces th e DC inpu t impeda nce of th e OP193, and t he 0.1F film

capacitor redu ces noise coupling int o its non-invert ing input .

R1*24.9k

USING A TEMPERATURE SENSOR FOR

COLD-JUNCTION COMPENSATION (TMP35)

TMP35

OP193

ISOTHERMALBLOCK

COLDJUNCTION

R6100k

R4*4.99k

R2*102

P150k

R5*1.21M

R3*

1.24M

TYPE KTHERMOCOUPLE

CHROMEL

ALUMEL

+

+

Cu

Cu

3.3V TO 5.5V

VOUT0.1 - 2.6V

* USE 1% RESISTORS

10mV/C

0 C < T < 250 C

0.1F

R7*4.99k 0.1F

FILM

Figure 7.10

The AD594/AD595 is a complete in str um ent at ion a mplifier a nd t her mocouple cold

jun ction compen sat or on a monolithic chip (see F igure 7.11). It combines an ice point

reference with a precalibrat ed a mplifier t o provide a h igh level (10mV/C) out put

directly from t he t her mocouple signal. Pin -stra pping options allow it t o be used as a

linear am plifier-compen sat or or a s a switched outpu t set -point cont roller u sing

either fixed or r emote set-point contr ol. It can be u sed to am plify its compen sat ion

voltage directly, ther eby becoming a sta nd-alone Celsius t ra nsdu cer with 10mV/C

outpu t. In su ch a pplicat ions it is very importan t t ha t t he IC chip is at t he sam e

tem pera tu re a s th e cold junction of th e th erm ocouple, which is us ua lly achieved by

keeping th e two in close proximity a nd isolat ed from a ny hea t sour ces.

The AD594/AD595 includes a th erm ocouple failure a lar m t ha t in dicat es if one orboth th erm ocouple leads open. The alar m outpu t h as a flexible form at which

includes TTL drive capability. The device can be powered from a single-ended supply

(which m ay be as low as +5V), but by including a n egative supply, temper at ur es

below 0C can be mea sur ed. To minimize self-heat ing, an un loaded AD594/AD595

will operat e with a supp ly cur ren t of 160A, but is also capa ble of delivering 5mA

to a load.

The AD594 is precalibrated by laser wafer trimming to match the characteristics of

type J (iron/const an ta n) ther mocouples, an d th e AD595 is laser t rimm ed for t ype K


10/39


7.10

(chromel/alumel). The temperature transducer voltages and gain control resistors

ar e available at the pa ckage pins so that the circuit can be r ecalibrated for other

th erm ocouple types by th e addition of resistors. These t erm inals a lso allow more

precise calibra tion for both t her mocouple an d th erm omet er a pplicat ions. The

AD594/AD595 is available in t wo performa nce gra des. The C a nd t he A versions

ha ve calibra tion a ccur acies of 1C a nd 3C, respectively. Both a re designed t o be

used with cold junctions between 0 to +50C. The circuit sh own in F igure 7.11 will

provide a dir ect outp ut from a type J th erm ocouple (AD594) or a type K

th erm ocouple (AD595) capa ble of measu rin g 0 t o +300C.

AD594/AD595 MONOLITHIC THERMOCOUPLE AMPLIFIERSWITH COLD-JUNCTION COMPENSATION

ICEPOINTCOMP

+

OVERLOADDETECT

VOUT10mV/C

+5V

BROKENTHERMOCOUPLE

ALARM

4.7k

G

+

TC

+TC+

+ATHERMOCOUPLE

G

AD594/AD595

TYPE J: AD594TYPE K: AD595

0.1F

Figure 7.11

The AD596/AD597 ar e m onolithic set-point cont rollers which h ave been optimized

for use at elevated temperatures as are found in oven control applications. The

device cold-junction compen sat es a nd am plifies a type J /K th erm ocouple t o derive an

inter na l signal pr oport iona l to tem pera tu re. They can be configured to provide a

voltage output (10mV/C) directly from type J/K thermocouple signals. The device ispackaged in a 10-pin metal can an d is trimmed to operat e over an a mbient ra nge

from +25C to +100C. The AD596 will amplify thermocouple signals covering the

entire 200C to +760C temperature range recommended for type J thermocouples

while th e AD597 can accomm odate 200C to +1250C type K inpu ts. Th ey have a

calibration a ccur acy of 4C at an ambient temper at ure of 60C an d an am bient

temperature stability specification of 0.05C/C from +25C to +100C.

None of the thermocouple amplifiers previously described compensate for

th erm ocouple n on-linear ity, th ey only provide conditioning a nd voltage ga in. High


11/39


7.11

resolution ADCs su ch as th e AD77XX fam ily can be used to digitize th e

th erm ocouple outpu t dir ectly, allowing a m icrocontr oller t o perform t he t ra nsfer

function linear ization as shown in F igure 7.12. The two multiplexed input s to the

ADC a re u sed t o digitize th e th erm ocouple volta ge an d t he cold-jun ction

tempera tur e sensor output s directly. The input P GA gain is progra mma ble from 1

to 128, an d th e ADC resolut ion is between 16 an d 22 bits (depending upon t he

particular ADC selected). The microcontroller performs both the cold-junction

compensat ion a nd t he linearization a rithm etic.

AD77XX ADC USED WITHTMP35 TEMPERATURE SENSOR FOR CJC

MUX

TMP35

ADC

OUTPUTREGISTER

CONTROL

REGISTER

SERIALINTERFACE

PGA

3V OR 5V(DEPENDING ON ADC)

THERMOCOUPLE

AD77XX SERIES

(16-22 BITS)

TO MICROCONTROLLER

G=1 TO 128

0.1F

AIN1+

AIN1

AIN2

AIN2+

Figure 7.12

R ESISTANCE TE MPE R AT UR E DE T E C T OR S (RTDS)

The Resistance Temperatu re Detector, or th e RTD, is a sensor whose resistan ce

cha nges with tem perat ure. Typically built of a platinum (Pt ) wire wr apped a round aceram ic bobbin, the RTD exhibits beha vior wh ich is more accur at e an d more linear

over wide tempera tur e ranges tha n a th ermocouple. Figure 7.13 illustr at es the

tem pera tu re coefficient of a 100 RTD and the Seebeck coefficient of a Type Sth erm ocouple. Over th e ent ire ra nge (appr oximat ely 200C to +850C), the RTD is

a more linear device. Hence, linea rizing an RTD is less complex.


12/39


7.12

RESISTANCE TEMPERATURE DETECTORs (RTD)

n Platinum (Pt) the Most Common

n 100,, 1000 Standard Valuesn Typical TC = 0.385% / C,

0.385 / / C for 100 Pt RTD

n Good Linearity - Better than Thermocouple,

Easily Compensated

0 400 8000.275

0.300

0.325

0.350

0.375

0.400

5.50

6.50

7.50

8.50

9.50

10.5

11.5

TYPE STHERMOCOUPLE

100 Pt RTDRTDRESISTANCE

TC, / C

TYPE STHERMOCOUPLE

SEEBECKCOEFFICIENT,

V / C

TEMPERATURE - C

Figure 7.13

Unlike a thermocouple, however, an RTD is a passive sensor and requires current

excitat ion t o produce an out put voltage. The RTD's low tem pera tu re coefficient of

0.385%/C r equires similar high-perform an ce signa l conditioning circuitr y to th at

used by a t her mocouple; however, the volta ge drop across a n RTD is mu ch larger

than a thermocouple output voltage. A system designer may opt for large valueRTDs with higher out put , but lar ge-valued RTDs exhibit slow response tim es.

Fu rth ermore, although the cost of RTDs is higher th an tha t of ther mocouples, they

use copper leads, a nd th erm oelectr ic effects from t erm inat ing junctions do not affect

th eir accura cy. And finally, becau se th eir resist an ce is a fun ction of the absolute

tem pera tu re, RTDs require n o cold-jun ction compen sat ion.

Caut ion m ust be exercised using curr ent excitation because the curr ent t hr ough th e

RTD cau ses heat ing. This self-heat ing chan ges the tem perat ur e of the RTD and

appear s as a mea sur ement err or. Hence, car eful att ention mu st be paid to the

design of th e signal conditioning circuitr y so th at self-heat ing is kept below 0.5C.

Man ufactur ers specify self-heatin g errors for va rious RTD values a nd sizes in st ill

an d in m oving air. To reduce the err or due t o self-heat ing, the minimum cur rent

should be used for t he r equired system resolution, and th e largest RTD value chosen

tha t r esults in acceptable response time.

Another effect that can produce measurement error is voltage drop in RTD lead

wires. This is especially critical with low-value 2-wire RTDs becau se t he

temperature coefficient and the absolute value of the RTD resistance are both small.

If the RTD is located a long dista nce from t he signa l conditioning circuitr y, then th e

lead resistan ce can be ohms or t ens of ohms, a nd a small am ount of lead r esistan ce

can contr ibute a significant error to the t emperat ur e measur ement. To illustr ate


13/39


7.13

th is point, let us assu me th at a 100 plat inum RTD with 30-gauge copper lea ds islocated about 100 feet from a cont roller's display console. The r esista nce of 30-gauge

copper wire is 0.105/ft, and t he t wo leads of th e RTD will contr ibute a tota l 21 toth e net work which is shown in F igure 7.14. This add itiona l resista nce will produce a

55C error in the measurement! The leads' temperature coefficient can contribute an

additional, and possibly significant, error to the measurement. To eliminate the

effect of the lead r esistan ce, a 4-wire t echn ique is used.

A 100 Pt RTD WITH 100 FEETOF 30-GAUGE LEAD WIRES

R = 10.5

R = 10.5

COPPER

COPPER

100

Pt RTD

RESISTANCE TC OF COPPER = 0.40%/C @ 20C

RESISTANCE TC OF Pt RTD = 0.385%/ C @ 20C

Figure 7.14

In Figur e 7.15, a 4-wire, or Kelvin, connection is m ade t o the RTD. A consta nt

cur rent is applied though th e FORCE leads of the RTD, and the voltage across th e

RTD itself is measured remotely via the SENSE leads. The measuring device can be

a DVM or an instrumentation amplifier, and high accuracy can be achieved provided

tha t t he m easur ing device exhibits high input impedance and/or low input bias

cur rent . Since th e SENSE leads do not car ry a ppreciable current , this t echnique is

insensit ive to lead wire length . Sour ces of err ors ar e th e sta bility of the const an t

cur rent sour ce and the inpu t impedance and/or bias curr ents in t he am plifier or

DVM.

RTDs ar e genera lly configured in a four -resistor br idge circuit. The bridge outpu t is

am plified by an instr um ent at ion a mplifier for fur th er pr ocessing. However, high

resolution measu rem ent ADCs such as t he AD77XX series allow th e RTD out put to

be digitized directly. In t his ma nn er, linear ization can be perform ed digita lly,

th ereby easing th e an alog circuit r equirements.


14/39


7.14

FOUR-WIRE OR KELVIN CONNECTION TO Pt RTDFOR ACCURATE MEASUREMENTS

I

FORCELEAD

FORCELEAD

RLEAD

RLEAD

100Pt RTD

SENSELEAD

SENSELEAD

TO HIGH - ZIN-AMP OR ADC

Figure 7.15

Figure 7.16 shows a 100 Pt RTD driven with a 400A excitation current source.The out put is digitized by one of the AD77XX series ADCs. Note th at th e RTD

excitat ion cur ren t sour ce also genera tes t he 2.5V referen ce voltage for t he ADC via

th e 6.25k resistor. Variat ions in t he excitat ion cur ren t do not affect t he circuitaccuracy, since both the input voltage and the reference voltage vary ratiometrically

with the excitation current. However, the 6.25k resistor m ust have a lowtemperature coefficient to avoid errors in the measurement. The high resolution of

th e ADC an d th e input PGA (gain of 1 to 128) eliminat es th e need for a dditional

conditioning circuits.

The ADT70 is a complete P t RTD signa l conditioner which provides a n out put

voltage of 5mV/C when using a 1k RTD (see Figure 7.17). The Pt RTD and t he1k reference resistor ar e both excited with 1mA ma tched cur ren t sources. Thisallows tempera tur e measu remen ts t o be ma de over a ra nge of approximately 50C

to +800C.

The ADT70 cont ain s th e two matched curr ent sources, a pr ecision r ail-to-ra il outpu t

instrumentation amplifier, a 2.5V reference, and an uncommitted rail-to-rail output

op am p. The ADT71 is th e sam e as t he ADT70 except t he int ern al volta ge referen ce

is omitt ed. A shu tdown function is included for ba tt ery powered equipmen t t ha t

redu ces th e quiescent curr ent from 3mA to 10A. The gain or full-scale ra nge for t hePt RTD and ADT701 system is set by a pr ecision exter na l resistor conn ected to the

instru menta tion amplifier. The uncommitt ed op amp m ay be used for scaling the

inter na l voltage r eferen ce, providing a "Pt RTD open" signa l or "over tem pera tu re"

war ning, providing a hea ter switching signa l, or other extern al conditioning

deter mined by th e u ser. Th e ADT70 is specified for operat ion from 40C t o +125C

an d is available in 20-pin DIP a nd SOIC pa ckages.


15/39


7.15

INTERFACING A Pt RTD TO A HIGH RESOLUTION ADC

ADC

OUTPUTREGISTER

CONTROLREGISTER

SERIALINTERFACE

PGA

3V OR 5V(DEPENDING ON ADC)

AD77XX SERIES

(16-22 BITS)

TO MICROCONTROLLER

G=1 TO 128

400A

100Pt RTD

+

AIN1+

AIN1

MUX

+VREF

VREF

RREF6.25k

Figure 7.16

CONDITIONING THE PLATINUM RTD USING THE ADT70

2.5VREFERENCE

SHUTDOWN

1k Pt

RTD

1k REFRES

INSTAMP

RG = 50k

MATCHED1mA SOURCES

+5V

-1V TO -5V

OUT = 5mV/ C

ADT70

GNDREF

Note: Some Pins Omittedfor Clarity

+

+

0.1F

Figure 7.17


16/39


7.16

THE R MI ST OR S

Similar in function to the RTD, thermistors are low-cost temperature-sensitive

resistors a nd a re const ru cted of solid semicondu ctor m at erials which exhibit a

positive or n egative temp era tu re coefficient. Alth ough positive temper at ur e

coefficient devices ar e ava ilable, th e most comm only used t her mistors a re t hose witha negative tempera tur e coefficient. Figure 7.18 shows t he r esistance-tempera tur e

char acter istic of a comm only used NTC (Negat ive Tempera tu re Coefficient )

ther mistor. The ther mistor is highly non-linear an d, of the thr ee tempera tur e

sensors discussed, is th e most sensitive.

RESISTANCE CHARACTERISTICS OF A10k NTC THERMISTOR

0

10

20

30

40

0 20 40 60 80 100

THERMISTORRESISTANCE

k

TEMPERATURE - C

Nominal Value @ 25 C

ALPHA THERMISTOR, INCORPORATEDRESISTANCE/TEMPERATURE CURVE 'A'10 k THERMISTOR, #13A1002-C3

Figure 7.18

The t her mistor's high s ensitivity (typically, 44,000ppm/C at 25C, as sh own inFigur e 7.19), allows it to detect minut e variat ions in t emper at ur e which could not be

observed with a n RTD or t her mocouple. This high sensitivity is a distin ct advan ta ge

over the RTD in that 4-wire Kelvin connections to the thermistor are not needed to

compen sat e for lead wire err ors. To illust ra te t his point , suppose a 10k NTCth erm istor, with a typical 25C temp era tu re coefficient of 44,000ppm/C, were

substitut ed for t he 100 Pt RTD in t he example given earlier, then a total lead wireresist an ce of 21 would generate less tha n 0.05C error in th e measu rement . This isroughly a factor of 500 improvement in err or over a n RTD.


17/39


7.17

TEMPERATURE COEFFICIENT OF10k NTC THERMISTOR

-20000

-30000

-40000

-50000

-60000

0 20 40 60 80 100

THERMISTORTEMPERATURECOEFFICIENT

ppm/ C

TEMPERATURE - C

ALPHA THERMISTOR, INCORPORATEDRESISTANCE/TEMPERATURE CURVE 'A'10 k THERMISTOR, #13A1002-C3

Figure 7.19

However, the thermistor's high sensitivity to temperature does not come without a

price. As was shown in F igure 7.18, the t emper at ur e coefficient of th erm istors does

not decrease linear ly with increasing t emperat ur e as it does with RTDs; therefore,

linear ization is required for a ll but t he n ar rowest of tempera tur e ra nges. Therm istor

applications are limited to a few hun dred degrees at best because t hey are more

susceptible to dama ge at h igh t emperat ur es. Compar ed to therm ocouples and RTDs,

thermistors are fragile in construction and require careful mounting procedures to

prevent crushing or bond separation. Although a thermistor's response time is short

due to its small size, its small thermal mass makes it very sensitive to self-heating

errors.

Thermistors are very inexpensive, highly sensitive temperature sensors. However,

we ha ve shown tha t a th ermistor's tempera tur e coefficient var ies from 44,000

ppm/C at 25C to 29,000ppm/C at 100C. Not only is t his n on-linear ity th e

largest sour ce of err or in a t emperat ure m easur ement, it a lso limits useful

applicat ions t o very na rr ow tempera tur e ra nges if linear ization techniques are not

used.

It is possible to use a th ermistor over a wide temper at ure ra nge only if the system

designer can tolerat e a lower sen sitivity to achieve impr oved linea rity. One a pproach

to linearizing a thermistor is simply shunting it with a fixed resistor. Paralleling the

th erm istor with a fixed resistor increases t he linear ity significantly. As shown in

Figure 7.20, the parallel combination exhibits a more linear variation with

temperature compared to the thermistor itself. Also, the sensitivity of the

combina tion still is high compa red t o a t her mocouple or RTD. The pr imar y


18/39


7.18

disadvanta ge to this technique is tha t linear ization can only be achieved within a

nar row ran ge.

LINEARIZATION OF NTC THERMISTORUSING A 5.17k SHUNT RESISTOR

0

10

20

30

40

0 20 40 60 80 100

RESISTANCEk

TEMPERATURE - C

THERMISTOR

PARALLEL COMBINATION

Figure 7.20

The va lue of th e fixed resistor can be calculated from t he following equa tion:

R =RT RT RT RT RT

RT RT RT

2 1 3 2 1 3

1 3 2 2

+ +

( ),

where RT1 is the t herm istor r esista nce at T1, the lowest tempera tur e in the

measu remen t ra nge, RT3 is the therm istor r esista nce at T3, the highest

tempera tur e in th e ran ge, and RT2 is the th ermistor resistance at T2, the midpoint,

T2 = (T1 +T3)/2.

For a typical 10k NTC thermistor, RT1 = 32,650 at 0C, RT2 = 6,532 at 35C,an d RT3 = 1,752 at 70C. This resu lts in a value of 5.17k for R. The accur acyneeded in t he signal conditioning circuitry depends on t he linear ity of the net work .

For t he exa mple given above, the n etwork s hows a n on-linear ity of 2.3C/ + 2.0 C.

The outpu t of th e network can be a pplied to an ADC to perform furt her linear ization

as sh own in Figure 7.21. Note th at the outpu t of the t herm istor n etwork h as a slope

of appr oximat ely 10mV/C, which implies a 12-bit ADC ha s m ore th an sufficient

resolution.


19/39


7.19

LINEARIZED THERMISTOR AMPLIFIER

10k NTCTHERMISTOR

5.17kLINEARIZATION

RESISTOR

226A

LINEARITY 2C, 0C TO +70C

VOUT 0.994V @ T = 0C

VOUT

0.294V @ T =70C

VOUT/T 10mV/C AMPLIFIEROR ADC

Figure 7.21

S EMICONDUCTOR TE MPE R AT UR E S E NSOR S

Modern semiconductor temperature sensors offer high accuracy and high linearity

over a n opera ting r an ge of about 55C to +150C. Inter na l amplifiers can scale the

outpu t to convenient values, su ch as 10mV/C. They a re a lso useful in cold-junction-

compensation circuits for wide temperature range thermocouples.

All semiconductor temperature sensors make use of the relationship between a

bipolar junction t ra nsistor's (BJT) base-emitter voltage t o its collector cur ren t:

VBEkT

q

Ic

Is=

ln

where k is Boltzmann 's consta nt, T is th e absolute tempera tu re, q is the char ge of

an electr on, an d Is is a cur rent related to the geometr y and the t emperat ur e of the

tr an sistors. (The equa tion a ssumes a voltage of at least a few hun dred mV on the

collector, and ignores Early effects.)

If we take N transistors identical to the first (see Figure 7.22) and allow the total

current Ic to be shared equa lly among them, we find tha t t he new base-emitt er

volta ge is given by th e equa tion

VNkT

q

Ic

N Is=

ln


20/39


7.20

BASIC RELATIONSHIPS FOR SEMICONDUCTORTEMPERATURE SENSORS

IC IC

VBE VN

VBE VBE VNkT

qN== == ln( )

VBEkT

q

ICIS

==

ln VN

kT

q

ICN IS

==

ln

INDEPENDENT OF IC, IS

ONE TRANSISTORN TRANSISTORS

Figure 7.22

Neith er of th ese circuits is of much use by itself becau se of the str ongly temper at ur e

dependent current Is, but if we have equal cur rent s in one BJ T and N similar BJ Ts

then the expression for t he difference between the two base-emitter voltages is

proport iona l to absolute t emperat ure and does not conta in Is.

VBE VBE VNkT

q

Ic

Is

kT

q

Ic

N Is= =

ln ln

VBE VBE VNkT

q

Ic

Is

Ic

N Is= =

ln ln

VBE VBE VNkT

q

I c

IsIc

N Is

kT

qN= =

=ln ln( )

The circuit shown in F igur e 7.23 implements t he a bove equation a nd is known as

the "Brokaw Cell" (see Reference 10). The voltage VBE = VBE VN appear s acrossresistor R2. The emitter curr ent in Q2 is ther efore VBE/R2. The op a mp's ser voloop an d th e resistors, R, force the sa me curr ent to flow thr ough Q1. The Q1 an d Q2

curr ents a re equa l and a re su mmed a nd flow into resistor R1. The corr esponding

voltage developed across R1 is proport iona l to absolut e tem pera tu re (PTAT) an d

given by:


21/39


7.21

( )VPTAT

VBE VN

R

R

R

kT

qN=

=

2R1

22

1

2ln( ) .

CLASSIC BANDGAP TEMPERATURE SENSOR

"BROKAW CELL"R R

+I2 I1

Q2NA

Q1

A

R2

R1

VN VBE

(Q1)

VBANDGAP = 1.205V

+VIN

VPTAT = 2R1

R2kTq

ln(N)

VBE VBE VNkT

q N== == ln( )

Figure 7.23

The ba ndga p cell reference voltage, VBANDGAP , appears a t th e base of Q1 and isthe sum of VBE (Q1) and VPTAT. VBE(Q1) is complementary to absolute

tempera tur e (CTAT), and su mming it with VPTAT causes th e bandgap voltage to be

const an t with r espect to tem pera tu re (assu ming proper choice of R1/R2 ra tio an d N

to ma ke t he ba ndga p voltage equa l to1.205V). This circuit is t he ba sic band-gap

tempera tur e sensor, and is widely used in semiconductor tempera tur e sensors.

C u r r e n t a n d Vo lt a g e Ou t p u t T e m p e r a t u r e S e n s o r s

The concepts used in t he ban dgap temper atu re sensor discussion above can be used

as t he basis for a variety of IC temperatu re sensors to genera te either curr ent or

voltage outputs. The AD592 and TMP17 (see Figure 7.24) are curr ent outpu t

sensors wh ich h ave scale factors of 1A/K. The sens ors do not r equire exter na l

calibra tion an d ar e available in severa l accur acy gra des. The AD592 is ava ilable in

th ree a ccur acy gra des. The highest grade version (AD592CN) has a ma ximum error

@ 25C of 0.5C and 1.0C er ror from 25C t o +105C. Linea rit y er ror is 0.35C.

The TMP17 is available in two accur acy gra des. The highest gr ade version

(TMP17F ) ha s a ma ximum err or @ 25C of 2.5C a nd 3.5C er ror from 40C t o

+105C. Typical linea rity err or is 0.5C. The AD592 is a vailable in a TO-92 pa ckage

an d th e TMP17 in a n SO-8 package.


22/39


7.22

CURRENT OUTPUT SENSORS: AD592, TMP17

n 1A/K Scale Factor

n Nominal Output Current @ +25C: 298.2A

n Operation from 4V to 30V

n 0.5C Max Error @ 25C, 1.0C Error Over Temp,

0.1C Typical Nonlinearity (AD592CN)

n 2.5C Max Error @ 25C, 3.5C Error Over Temp,

0.5C Typical Nonlinearity (TMP17F)n AD592 Specified from 25C to +105C

n TMP17 Specified from 40C to +105C

V+

V

AD592: TO-92 PACKAGE

TMP17: SO-8 PACKAGE

Figure 7.24

RATIOMETRIC VOLTAGE OUTPUT SENSORS

R(T)

I(VS)

AD22103

VS = +3.3V

REFERENCE

INPUT

ADC

+

GND

VOUT

VOUTVS

VV

mV

CTA== ++

3 3

0 2528

..

0.1F

Figure 7.25


23/39


7.23

In some cases, it is desira ble for the outpu t of a t emperat ure sensor to be rat iometr ic

with its supply voltage. The AD22103 (see Figure 7.25) has an output that is

ra tiometr ic with its su pply voltage (nominally 3.3V) according to th e equa tion:

VOU TVS

V

VmV

C

TA= +

3 3

0 2528

.

. .

The circuit sh own in F igure 7.25 uses the AD22103 power su pply as th e reference to

th e ADC, ther eby eliminat ing th e need for a precision volta ge reference. The

AD22103 is specified over a ra nge of 0C to +100C an d ha s an accur acy better th an

2.5C and a linearity better than 0.5C.

The TMP 35/TMP36/TMP37 a re low voltage (2.7V to 5.5V) SOT-23 (5-pin), SO-8, or

TO-92 packaged volta ge out put tem pera tu re sen sors with a 10mV/C (TMP35/36) or

20mV/C (TMP37) scale factor (see Figure 7.26). Supply current is below 50A,

providing very low self-heatin g (less t ha n 0.1C in still air). A shu tdown featu re is

provided which redu ces th e cur ren t t o 0.5A.

The TMP35 provides a 250mV outpu t a t +25C an d rea ds temper at ure from +10C

to +125C. The TMP36 is specified from 40C to +125C. and provides a 750mV

out put at 25C. Both the TMP35 an d TMP36 have an out put scale factor of

+10mV/C. The TMP 37 is int ended for applications over th e r an ge +5C to +100C,

an d pr ovides an outpu t scale factor of 20mV/C. The TMP37 pr ovides a 500mV

out put at +25C.

ABSOLUTE VOLTAGE OUTPUT SENSORSWITH SHUTDOWN

n VOUT:

u TMP35, 250mV @ 25C, 10mV/C (+10C to +125C)

uTMP36, 750mV @ 25C, 10mV/C (40C to +125C)

u TMP37, 500mV @ 25C, 20mV/C ( +5C to +100C)

n 2C Error Over Temp (Typical), 0.5C Non-Linearity (Typical)n Specified 40C to +125C

n 50A Quiescent Current, 0.5A in Shutdown Mode

TMP35TMP36

TMP37

+VS = 2.7V TO 5.5V

VOUT

SHUTDOWN

SOT-23-5

ALSOSO-8

OR TO-92

0.1F

Figure 7.26


24/39


7.24

The ADT45/ADT50 are voltage output temperature sensors packaged in a SOT-23-3

packa ge designed for an operat ing volta ge of 2.7V to 12V (see F igure 7.27). The

devices are specified over the range of 40C to +125C. The output scale factor for

both devices is 10mV/C. Typical accuracies are 1C at +25C an d 2C over t he 40C to +125C ra nge. The ADT45 provides a 250mV out put at +25C an d is

specified for tem pera tu re from 0C to +100C. The ADT50 provides a 750mV out put

at +25C a nd is specified for tem pera tu re from 40C to +125C.

ADT45/ADT50 ABSOLUTE VOLTAGE OUTPUT SENSORS

n VOUT:

u ADT45, 250mV @ 25C, 10mV/C Scale Factor

u ADT50, 750mV @ 25C, 10mV/C Scale Factor

n 2C Error Over Temp (Typical), 0.5C Non-Linearity (Typical)n Specified 40C to +125C

n 60A Quiescent Current

ADT45

ADT50

+VS = 2.7V TO 12V

VOUT

0.1F

SOT-23

Figure 7.27

If the ADT45/ADT50 sensors ar e th erma lly att ached an d pr otected, they can be

used in any temperature measurement application where the maximum

temperature range of the medium is between 40C to +125C. Properly cementedor glued to th e sur face of th e medium , th ese sensors will be with in 0.01C of th e

sur face temper at ur e. Caut ion sh ould be exercised, as an y wiring to the device can

act as h eat pipes, intr oducing errors if th e sur round ing air-surface int erface is not

isoth erm al. Avoiding th is condition is ea sily achieved by dabbing th e leads of the

sensor a nd t he h ookup wires with a bea d of th erm ally condu ctive epoxy. This will

ensur e tha t t he ADT45/ADT50 die temperat ure is not a ffected by the su rr oun ding

air temperat ure.


25/39


7.25

In the SOT-23-3 package, the thermal resistance junction-to-case, J C, is 180C/W.The t herm al r esistance case-to-am bient, CA, is th e differen ce between J A andJ C, and is deter mined by th e cha ra cteristics of th e th erm al conn ection. With n o airflow an d th e device soldered on a PC boar d, J A is 300C/W. The temperaturesensor's power dissipation, P

D, is the p roduct of th e tota l voltage a cross th e device

an d its total supp ly curr ent (including any curr ent delivered to the load). The r ise in

die temperat ur e above the medium's ambient tempera tu re is given by:

TJ = PD (J C + CA) + TA.

Thu s, th e die temper at ur e rise of an un loaded ADT45/ADT50 (SOT-23-3 packa ge)

soldered on a boar d in st ill air at 25C an d dr iven from a +5V supply (quiescent

curr ent = 60A, PD = 300W) is less th an 0.09C. In order t o prevent furt her

tempera tur e rise, it is importa nt to minimize the load curr ent, always keeping it less

than 100A.

The tr an sient r esponse of th e ADT45/ADT50 sensors to a step cha nge intempera tur e is determined by th e therma l resistan ces and the th erma l mass of the

die and th e case. The therm al mass of th e case varies with t he measu rement

medium since it includes anyth ing tha t is in direct conta ct with the package. In all

pra ctical cases, the ther mal m ass of the case is th e limiting factor in the ther mal

response tim e of th e sensor an d can be represen ted by a single-pole RC time

constant. Thermal mass is often considered the thermal equivalent of electrical

capacitance.

The th erma l time consta nt of a tempera tur e sensor is defined to be the t ime

requ ired for t he sen sor to reach 63.2% of the fina l value for a st ep chan ge in th e

temper at ur e. Figure 7.28 shows th e th erma l time consta nt of the ADT45/ADT50

series of sensors wit h th e SOT-23-3 pa ckage soldered t o 0.338" x 0.307" copper PC

board as a function of air flow velocity. Note the rapid drop from 32 seconds to 12

seconds as th e a ir velocity increa ses from 0 (still air) to 100 LFPM. As a p oint of

referen ce, the t her ma l time consta nt of th e ADT45/ADT50 series in a stir red oil bat h

is less th an 1 second, which verifies that the major pa rt of the th erma l time consta nt

is determined by th e case.

The power supp ly pin of th ese sensors sh ould be bypassed t o groun d with a 0.1F

ceram ic capacitor h aving very short leads (preferably surface mount ) an d located as

close to the power supply pin a s possible. Since th ese temper at ur e sensors opera te

on very little supp ly cur ren t an d could be exposed to very h ostile electr ical

environmen ts, it is import an t t o minimize the effects of EMI/RFI on t hese devices.The effect of RFI on these t emperat ure sensors is man ifested as abnorma l DC shifts

in th e outpu t volta ge due to rectificat ion of th e high frequen cy noise by th e inter na l

IC junctions. In t hose cases wh ere th e devices ar e opera ted in t he pr esence of high

frequency radia ted or condu cted noise, a lar ge value ta nt alum electr olytic capacitor

(>2.2F) placed across th e 0.1F ceram ic ma y offer a dditional noise immu nity.


26/39


7.26

THERMAL RESPONSE IN FORCED AIR FOR SOT-23-3

0 100 200 300 400 500 600 700

0

5

10

15

20

25

30

35

AIR VELOCITY - LFPM

TIMECONSTANT-SECONDS

SOT-23-3 SOLDERED TO 0.338" x 0.307" Cu PCBV+ = 2.7V TO 5VNO LOAD

Figure 7.28

D ig it a l O u t p u t T e m p e r a t u r e S e n s or s

Temperat ur e sensors which have digital outpu ts ha ve a num ber of advan ta ges over

th ose with an alog outpu ts, especially in rem ote a pplicat ions. Opto-isolat ors can a lso

be used to provide galvanic isolat ion between th e rem ote sen sor an d th emea sur emen t system . A volta ge-to-frequency converter driven by a voltage outpu t

temperature sensor accomplishes this function, however, more sophisticated ICs are

now available which a re more efficient a nd offer severa l perform an ce adva nt ages.

The TMP03/TMP04 digita l out put sensor family includes a volta ge reference,

VPTAT genera tor, sigma-delta ADC, and a clock sour ce (see Figur e 7.29). The

sensor outpu t is digitized by a first-order sigma-delta modulat or, also known as th e

"char ge balan ce" type an alog-to-digital convert er. Th is convert er ut ilizes time-

domain oversam pling and a h igh accur acy compa ra tor t o deliver 12 bits of effective

accur acy in an extrem ely compa ct circuit.

The output of the sigma-delta modulator is encoded using a proprietary technique

which results in a serial digital output signal with a mar k-space rat io forma t (see

Figur e 7.30) th at is easily decoded by an y microprocessor into either degrees

centigrade or degrees Fa hr enheit, an d r eadily tra nsmitted over a single wire. Most

importa nt ly, th is encoding meth od avoids ma jor er ror sources comm on to oth er

modulat ion t echniques, as it is clock-indepen dent . The nominal outpu t frequen cy is

35Hz a t + 25C, an d t he device operat es with a fixed high-level pulse width (T1) of

10ms.


27/39


7.27

DIGITAL OUTPUT SENSORS: TMP03/04

REFERENCEVOLTAGE

TEMP

SENSORVPTAT

SIGMA-DELTAADC

CLOCK(1MHz)

OUTPUT

(TMP04)

OUTPUT(TMP03)

TMP03/TMP04

+VS = 4.5 TO 7V

GND

Figure 7.29

TMP03/TMP04 OUTPUT FORMAT

n T1 Nominal Pulse Width = 10ms

n 1.5C Error Over Temp, 0.5C Non-Linearity (Typical)n Specified 40C to +100C

n Nominal T1/T2 @ 0C = 60%

n Nominal Frequency @ +25C = 35Hz

n 6.5mW Power Consumption @ 5V

n TO-92, SO-8, or TSSOP Packages

T1 T2

TEMPERATURE CT

T( ) ==

235400 1

2

TEMPERATURE FT

T( ) ==

455720 1

2

Figure 7.30


28/39


7.28

The TMP03/TMP04 outpu t is a st ream of digita l pulses, and t he temper at ure

inform ation is conta ined in th e ma rk-space ra tio per the equations:

Tempera ture CT

T( ) =

235400 1

2

Tempera ture FT

T( ) =

455

720 1

2.

Popular microcont rollers, such a s th e 80C51 an d 68HC11, have on-chip timers

which can ea sily decode th e ma rk -space rat io of the TMP 03/TMP04. A typical

inter face to the 80C51 is shown in F igure 7.31. Two timers, labeled Tim er 0 and

Tim er 1 ar e 16 bits in lengt h. Th e 80C51's system clock, divided by twelve, provides

th e sour ce for t he t imers. The syst em clock is norma lly derived from a crystal

oscillator, so timing measurements are quite accurate. Since the sensor's output is

ratiometric, the actual clock frequency is not important. This feature is important

because the microcont roller's clock frequency is often defined by some extern al

timing constr aint, such as th e serial baud r ate.

INTERFACING TMP04 TO A MICROCONTROLLER

CPU

TIMER

CONTROL

OSCILLATOR 12

TIMER 0

TIMER 1

80C51 MICROCONTROLLER

TMP04 OUT

V+

GND

+5V

NOTE: ADDITIONALPINS OMITTEDFOR CLARITY

XTAL

P1.0

0.1F

Figure 7.31

Softwa re for t he sen sor inter face is str aight forwa rd. The microcont roller simply

monitors I/O port P1.0, an d sta rt s Tim er 0 on t he r ising edge of th e sensor out put .

The m icrocontroller cont inues to monitor P 1.0, stopping Tim er 0 and s tar t ing Timer

1 when t he sensor output goes low. When the outpu t r etur ns h igh, th e sensor's T1

an d T2 times are conta ined in r egisters Tim er 0 and Tim er 1, respectively. Further

software routines can then apply the conversion factor shown in the equations above

and calculate th e temperatur e.


29/39


7.29

The TMP03/TMP04 are ideal for monitoring th e th erma l environment within

electr onic equipmen t. For exam ple, th e sur face mount ed package will accur at ely

reflect t he t her ma l conditions which affect near by int egrat ed circuits. The TO-92

package, on t he other ha nd, can be mount ed above th e sur face of the boar d to

measu re th e tempera tur e of the air flowing over the board.

The TMP03 an d TMP04 measu re an d convert th e tempera tur e at th e surface of

their own semiconductor chip. When they ar e used to measu re t he t emperat ure of a

near by heat source, the ther ma l impedan ce between th e heat source and th e sensor

must be considered. Often, a ther mocouple or other tempera tur e sensor is used to

measu re th e tempera tur e of the sour ce, while the TMP03/TMP04 temperat ur e is

monitored by measu ring T1 and T2. Once the th erma l impedance is determ ined, the

tempera tu re of the heat sour ce can be inferred from th e TMP03/TMP04 out put.

One exam ple of using the TMP 04 to monitor a high power dissipation

microprocessor or oth er IC is sh own in F igure 7.32. The TMP04, in a sur face mount

packa ge, is mount ed directly benea th th e microprocessor's pin grid a rr ay (PGA)

packa ge. In a typical a pplicat ion, th e TMP04's out put would be conn ected t o an

ASIC where t he m ar k-space rat io would be measur ed. The TMP04 pulse outpu tprovides a significan t a dvan ta ge in th is applicat ion because it pr oduces a linear

tempera tur e output, while needing only one I/O pin an d without r equiring an ADC.

MONITORING HIGH POWER MICROPROCESSOROR DSP WITH TMP04

FAST MICROPROCESSOR, DSP, ETC.,

IN PGA PACKAGE

PGA SOCKET

PC BOARD

TMP04 IN SURFACEMOUNT PACKAGE

Figure 7.32

T h e r m o s t a t i c S w it c h e s a n d S e t p o i n t C o n t r o l le r s

Temperat ure sensors u sed in conjunction with compar ators can a ct a s th ermostatic

switches. ICs such a s th e ADT05 accomplish t his function at low cost a nd a llow a

single exter na l resistor to progra m th e setpoint to 2C accur acy over a r an ge of

40C to +150C (see Figur e 7.33). The device assert s a n open collector out put when

th e ambient temperat ur e exceeds the user-progra mmed setpoint tempera tur e. The

ADT05 has appr oximat ely 4C of hyster esis which pr events r apid t her ma l on/off

cycling. The ADT05 is designed to operate on a single supply voltage from +2.7V to


30/39


7.30

+7.0V facilita ting operat ion in ba tt ery powered a pplicat ions a s well as indu str ial

cont rol syst ems . Becau se of low power dissipa tion (200W @ 3.3V), self-hea tin g

err ors a re min imized, an d bat ter y life is maximized. An optiona l int ern al 200kpull-up r esistor is included t o facilitat e driving light loads su ch as CMOS inpu ts.

The setpoint resistor is determined by th e equat ion:

RSETM C

TSET C Ck=

+

39

281690 3

( ) .

. .

The setp oint r esistor should be conn ected dir ectly between th e RSET pin (Pin 4) an d

th e GND pin (Pin 5). If a groun d plane is used, th e resistor ma y be conn ected

directly to this plan e at t he closest a vailable point.

The setpoint resistor can be of nearly any resistor type, but its initial tolerance and

thermal drift will affect the accuracy of the programmed switching temperature. For

most a pplicat ions, a 1% meta l-film r esistor will provide th e best tr adeoff between

cost an d a ccur acy. Once RSET ha s been calculated, it ma y be found tha t t he

calculated value does not agree with readily available standard resistors of thechosen t olera nce. In order t o achieve a va lue as close as possible to the calculated

value, a compoun d resist or can be const ru cted by conn ecting t wo resistors in series

or parallel.

ADT05 THERMOSTATIC SWITCH

n 2C Setpoint Accuracyn 4C Preset Hysteresis

n Specified Operating Range: 40C to + 150C

n Power Dissipation: 200W @ 3.3V

SET-POINT

TEMPSENSOR

200k

RSET

+VS = 2.7V TO 7V

OUT

RPULL-UP

ADT05

SOT-23-5

0.1F

Figure 7.33


31/39


7.31

The TMP01 is a dua l setpoint t emperat ur e contr oller which also generates a PTAT

outpu t volta ge (see Figur e 7.34 and 7.35). It a lso genera tes a cont rol signa l from one

of two out put s when th e device is either above or below a specific temper at ur e

ra nge. Both t he high/low tempera tur e trip points an d hysteresis band ar e

determined by user-selected external resistors.

TMP01 PROGRAMMABLE SETPOINT CONTROLLER

VPTAT

+

TEMPERATURE

SENSOR ANDVOLTAGE

REFERENCE

+

HYSTERESISGENERATOR

OVER

UNDER

V+2.5VVREF

SETHIGH

SETLOW

R1

R2

R3

GND

WINDOWCOMPARATOR

TMP01

Figure 7.34

The TMP01 consist s of a ba ndga p voltage r eferen ce combined with a pa ir of ma tched

compar at ors. The r eference provides both a consta nt 2.5V outpu t an d a PTAT

outpu t volta ge which h as a precise temper at ur e coefficient of 5mV/K an d is 1.49V

(nomina l) at +25C. The compa ra tors compa re VPTAT with t he exter na lly set

tempera tur e tr ip points a nd genera te a n open-collector output signa l when one of

their respective thresholds has been exceeded.

Hysteresis is also progra mmed by th e external r esistor cha in an d is determined by

the total curr ent dr awn out of the 2.5V reference. This cur rent is mirrored and used

to generate a hysteresis offset voltage of the appropriate polarity after a comparator

ha s been tripped. The compar ators ar e conn ected in par allel, which gua ra ntees th at

ther e is no hysteresis overlap a nd eliminat es erra tic tran sitions between a djacent

tr ip zones.


32/39


7.32

The TMP01 utilizes laser trimmed thin-film resistors to maintain a typical

tempera tur e accur acy of1C over th e ra ted t emper at ur e ra nge. The open-collectoroutpu ts a re capa ble of sink ing 20mA, ena bling the TMP01 t o drive cont rol relays

directly. Opera ting from a +5V supply, quiescent curr ent is only 500A ma ximum.

TMP01 SETPOINT CONTROLLER KEY FEATURESn VC: 4.5 to 13.2V

n Temperature Output: VPTAT, +5mV/K

n Nominal 1.49V Output @ 25C

n 1C Typical Accuracy Over Temperature

n Specified Operating Range: 55C to + 125C

n Resistor-Programmable Hysteresis

n Resistor-Programmable Setpoints

n Precision 2.5V 8mV Referencen 400A Quiescent Current, 1A in Shutdown

n Packages: 8-Pin Dip, 8-Pin SOIC, 8-Pin TO-99

n Other Setpoint Controllers:

u Dual Setpoint Controllers: ADT22/ADT23

(3V Versions of TMP01 with Internal Hysteresis)

u Quad Setpoint Controller: ADT14

Figure 7.35

The ADT22/23-series are similar t o the TMP01 but h ave intern al hysteresis an d ar e

designed to opera te on a 3V supply. A quad (ADT14) setpoint cont roller is a lso

available.

AD C s Wi t h O n -C h i p T e m p e r a t u r e S e n s o r s

The AD7816/7817/7818-series digital tem pera tu re sen sors ha ve on-boar d

temperature sensors whose outputs are digitized by a 10-bit 9s conversion time

switched capacitor SAR ADC. The seria l inter face is compa tible with t he In tel 8051,

Motorola SPI an d QSPI, an d Nat iona l Semicondu ctor's MICROWIRE

protocol. The device fam ily offer s a var iety of inpu t options for fur th er flexibility.

The AD7416/7417/7418 are similar but ha ve stan dar d ser ial interfaces. Fu nctiona l

block diagra ms of the AD7816, AD7817, and AD7818 ar e sh own in Figur es 7.36, 37,

an d 38, and k ey specificat ions in F igure 7.39


33/39


7.33

AD7816 10-BIT DIGITAL TEMPERATURE SENSORWITH SERIAL INTERFACE

2.5V

REF

10-BITCHARGE

REDISTRIBUTION

SAR ADC

TEMP

SENSOR

OVER TEMPREGISTER

A > B

CLOCK

+VDD = 2.7V TO 5.5V

OTI

SCLK

DIN/OUT

AGND

RD/WR

CONVST

MUX

REFIN

CONTROLREGISTER

OUTPUT

REGISTER

AD7816

Figure 7.36

AD7817 10-BIT MUXED INPUT ADC WITH TEMP SENSOR

2.5VREF

10-BIT

CHARGE

REDISTRIBUTIONSAR ADC

TEMPSENSOR

OVER TEMPREGISTER

CONTROLREGISTER

A > B

CLOCK

+VDD

= 2.7V TO 5.5V

OTI

SCLK

DOUT

AGND

RD/WR

CONVST

MUX

REFIN

DGND BUSY

VIN1

VIN2

VIN3

VIN4

CS

OUTPUTREGISTER

DIN

AD7817

Figure 7.37


34/39


7.34

AD7818 SINGLE INPUT 10-BIT ADC WITH TEMP SENSOR

2.5VREF

10-BITCHARGE

REDISTRIBUTIONSAR ADC

TEMPSENSOR

OVER TEMPREGISTER

A > B

CLOCK

+VDD = 2.7V TO 5.5V

OTI

SCLK

AGND CONVST

MUX

CONTROLREGISTER

OUTPUTREGISTERVIN1

DIN/OUT

RD/WR

AD7818

Figure 7.38

AD7816/7817/7818 - SERIES TEMP SENSOR10-BIT ADCs WITH SERIAL INTERFACE

n 10-Bit ADC with 9s Conversion Time

n Flexible Serial Interface (Intel 8051, Motorola SPI and QSPI,National MICROWIRE)

n On-Chip Temperature Sensor: 55C to +125C

n Temperature Accuracy: 2C from 40C to +85C

n On-Chip Voltage Reference: 2.5V 1%

n +2.7V to +5.5V Power Supply

n 4W Power Dissipation at 10Hz Sampling Rate

n Auto Power Down after Conversion

n Over-Temp Interrupt Output

n Four Single-Ended Analog Input Channels: AD7817

n One Single-Ended Analog Input Channel: AD7818

n AD7416/7417/7418: Similar, but have I2C Compatible Interface

Figure 7.39


35/39


7.35

M I C R OPR OC E SSOR TE MPE R AT UR E MONITORING

Today's comput ers r equire tha t h ar dware a s well as softwar e operat e properly, in

spite of the ma ny th ings tha t can cause a system cra sh or lockup. The pur pose of

ha rdwar e monitoring is to monitor the critical items in a comput ing system a nd t ake

corrective action should problems occur.

Microprocessor supply voltage and temperature are two critical parameters. If the

supply voltage dr ops below a specified minim um level, fur th er opera tions should be

ha lted un til the volta ge retu rn s to accepta ble levels. In some cases, it is desirable to

reset th e microprocessor un der "brownout " conditions. It is also comm on pra ctice to

reset th e microprocessor on power-up or power-down. Switching to a ba tt ery backu p

ma y be requir ed if th e sup ply voltage is low.

Un der low voltage conditions it is ma nda tory to inh ibit t he m icroprocessor from

writing to extern al CMOS memory by inhibiting the Ch ip Ena ble signal to th e

externa l memory.

Man y microprocessors can be pr ogra mm ed to periodically out put a "watchdog"

signal. Monitoring th is signa l gives an indicat ion t ha t t he pr ocessor a nd its softwa re

ar e fun ctioning properly and t ha t t he pr ocessor is not stu ck in a n endless loop.

The need for h ar dware m onitoring has r esulted in a num ber of ICs, traditiona lly

called "microprocessor supervisory products," which perform some or all of the above

functions. These devices range from simple manual reset generators (with

debouncing) to complete microcontroller-based monitoring sub-systems with on-chip

tem pera tu re sensors an d ADCs. Ana log Devices' ADM-family of products is

specifically to perform the various microprocessor supervisory functions required in

different systems.

CPU tempera tur e is critically importan t in the Pent ium II microprocessors. For t his

reason, all new Pentium II devices have an on-chip substra te PN P t ra nsistor which

is designed to monitor the actual chip temperature. The collector of the substrate

PNP is connected to the subst ra te, and th e base and emitter a re brought out on two

separa te pins of the P entium II.

The ADM1021 Microprocessor Temperature Monitor is specifically designed to

process th ese outpu ts a nd convert the voltage into a digita l word representing t he

chip temper at ur e. The simplified an alog signal pr ocessing portion of th e ADM1021

is shown in Figur e 7.40.

The technique used to measur e the tem perat ur e is identical to the "VBE " principlepreviously discussed. Two differen t curr ent s (I and N I)ar e applied to the sen sing

tr an sistor, an d th e voltage measu red for ea ch. In the ADM1021, the nominal

curr ent s ar e I = 6A, (N = 17), NI = 102A. The cha nge in th e base-emitt er volta ge,

VBE , is a P TAT volta ge and given by the equ at ion:

VBEkT

qN= ln( ) .


36/39


7.36

Figure 7.40 shows the external sensor as a su bstra te t ra nsistor, provided for

tem pera tu re m onitoring in t he m icroprocessor, but it could equa lly well be a discret e

tr an sistor. If a discrete t ra nsist or is us ed, the collector sh ould be conn ected t o th e

base an d not grounded. To prevent ground noise interfering with th e measu rement ,

th e more negat ive ter mina l of th e sensor is not referenced to ground, but is biased

above ground by an in ter na l diode. If th e sensor is opera ting in a n oisy environmen t,

C ma y be optionally added a s a n oise filter. Its va lue is typically 2200pF, but should

be no more tha n 3000pF.

ADM1021 MICROPROCESSOR TEMPERATURE MONITORINPUT SIGNAL CONDITIONING CIRCUITS

65kHz

LOWPASSFILTER

OSCILLATOR

CHOPPER

AMPLIFIER

AND RECTIFIER

TO ADC

GAIN=G

I N I

VOUT

VOUT = G kTq ln N

PREMOTESENSING

TRANSISTOR

SPNP

IBIAS

BIAS

DIODE

C

VDD = +3V TO +5.5V

kT

q ln NVBE =

D+

D

Figure 7.40

To measu re VBE , the sensing tra nsistor is switched between operating current s ofI a nd N I. The r esulting waveform is passed th rough a 65kHz lowpass filter t orem ove noise, then to a chopper-sta bilized amplifier which per form s t he fun ction of

am plificat ion a nd synchronous r ectificat ion. The resu lting DC voltage is pr oport iona l

to VBE an d is digitized by an 8-bit ADC. To fur th er r educe th e effects of noise,digita l filtering is perform ed by avera ging the r esults of 16 mea sur ement cycles.


37/39


7.37

In addition, th e ADM1021 conta ins an on-chip temper atu re sensor, an d its signal

conditioning and m easur ement is performed in t he sam e man ner.

One LSB of th e ADC corr esponds t o 1C, so the ADC can t heoretically measu re from

128C to +127C, alth ough th e pra ctical lowest value is limited t o 65C du e to

device maximum ra tings. The r esults of the local an d remote temper at ure

measu rement s are stored in th e local and r emote tempera tur e value registers, and

ar e compa red with limits pr ogra mm ed into the local and r emote high an d low limit

registers as sh own in F igure 7.41. An ALERT outpu t signals when t he on-chip or

remote tempera tu re is out of ra nge. This outpu t can be used as an interr upt, or as

an SMBus alert .

The limit register s can be pr ogra mm ed, and th e device contr olled and configur ed, via

th e serial System Man agement Bus (SMBus). The conten ts of any register can also

be rea d back by th e SMBus. Contr ol an d configur at ion fun ctions consist of:

switching the device between normal operation and standby mode, masking or

enabling the ALERT outpu t, an d selecting the conversion r at e which can be setfrom 0.0625Hz to 8Hz.

STATUS

REGISTER

ADM1021 SIMPLIFIED BLOCK DIAGRAM

ADDRESS POINTER

REGISTER

ONE-SHOT

REGISTER

CONVERSION RATEREGISTER

LOCAL TEMPERATURE

LOW LIMIT REGISTER

LOCAL TEMPERATUREHIGH LIMIT REGISTER

REMOTE TEMPERATURE

LOW LIMIT REGISTER

REMOTE TEMPERATURE

HIGH LIMIT REGISTER

CONFIGURATION

REGISTER

INTERRUPT

MASKING

SMBUS INTERFACE

LOCAL TEMPERATURE

LOW LIMIT COMPARATOR

LOCAL TEMPERATUREHIGH LIMIT COMPARATOR

REMOTE TEMPERATURE

LOW LIMIT COMPARATOR

REMOTE TEMPERATURE

HIGH LIMIT COMPARATOR

LOCAL TEMPERATURE

VALUE REGISTER

REMOTE TEMPERATURE

VALUE REGISTER

SIGNAL CONDITIONING

AND ANALOG MUX

8-BIT

ADC

TEMP

SENSOR

D+

D

TEST VDD NC GND GND NC NC TEST SDATA SCLK ADD0 ADD1

STBY

ALERT

RUN/STANDBYBUSY

EXTERNAL DIODE OPEN CIRCUIT

Figure 7.41


38/39


7.38

ADM1021 KEY SPECIFICATIONS

n On-Chip and Remote Temperature Sensing

n 1C Accuracy for On-Chip Sensor

n 3C Accuracy for Remote Sensor

n Programmable Over / Under Temperature Limits

n 2-Wire SMBus Serial Interface

n 70A Max Operating Current

n 3A Standby Current

n +3V to +5.5V Supplies

n 16-Pin QSOP Package

Figure 7.42


39/39


R E F E R E N C E S

1. Ra m on Pa lla s-Ar en y a n d J oh n G. Webs ter , S e n s or s a n d S i gn a l

C o n d i t i o n i n g, John Wiley, New York, 1991.

2. Da n Shein gold, E ditor , T r a n s d u c e r I n t e r f a c in g H a n d b o o k , Ana log

Devices, Inc., 1980.

3. Walt Kester , Editor , 1992 Amp l i fi e r App l ica t ions Gu ide , Section 2, 3,

Ana log Devices, Inc., 1992.

4. Walt Kester , Editor , Sys t em App l ica t i on s Gu i d e , Section 1, 6, Analog

Devices, Inc., 1993.

5. J im Williams, T hermocouple Measurem ent, L in e a r T e c h n o l og y

App l ica t ion Not e 28, Linear Techn ology Corpora tion.

6. Dan Sheingold, N o n l in e a r C i r c u i t s H a n d b o o k , Ana log Devices, Inc.

7. J ames Wong, Tem perature Measurements Gain from Adv ances in High-

precision Op A m ps , El ec t ron i c Des ign , 15 May 1986.

8. OME GA Tem perature Measurement Ha nd book, Omega Instr umen ts, Inc.

9. H a n d b o o k o f C h e m i st r y a n d P h y s ic s , Chemical Rubber Co.

10. Paul Brokaw,A S im ple Three-T erm inal IC Ban dgap V oltage Reference,

I E E E J o u r n a l of S o li d S t a t e C i r c u i t s , Vol. SC-9, December , 1974.

Documents

Temperature Sensors Chapter7