Practical Temperature Measurements - Unicampelnatan/ee641/Temperatura.pdf · Temperature Measurements 001. Agenda Background, history Mechanical sensors Hewlett-Packard Classroom

Practical

Hewlett-Packard Classroom Series

PracticalTemperature

Measurements

001

Agenda

�Background, history�Mechanical sensors


�Mechanical sensors�Electrical sensors

� Optical Pyrometer

� RTD� Thermistor, IC

A1

� Thermocouple�Summary & Examples

What is Temperature?

�A scalar quantity that determines the direction of heat flow between two bodies


direction of heat flow between two bodies�A statistical measurement�A difficult measurement�A mostly empirical measurement

002

How is heat transferred?

�Conduction� Metal coffee cup


�Convection

�Radiation

003

The Dewar

�Glass is a poor conductor


�Glass is a poor conductor�Gap reduces conduction�Metallization reflects radiation�Vacuum reduces convection

004

Thermal Mass

�Don't let the measuring


Sensor

�Don't let the measuring device change the temperature of what you're measuring.

�Response time = � f{Thermal mass}� f{Measuring device}

005

Sensor

� f{Measuring device}

Temperature errors

�What is YOUR normal temperature?�Thermometer accuracy, resolution�Contact time


�Contact time�Thermal mass of thermometer, tongue�Human error in reading

006

97.6 98.6 99.6 36.5 37 37.5

History of temperature sensors

�1600 ad

�1700 ad12 96


ad ad

�Galileo: First temp. sensor � Early

12

10

96�Fahrenheit

� Instrument Maker

� 12*8=96 points

� Hg:

007

temp. sensor� pressure-

sensitive� not repeatable

� Early thermometers� Not

repeatable� No good way

to calibrate

� Hg: Repeatable

� One standard scale

The 1700's: Standardization

�1700 ad �1800 ad

0 100


�Celsius:�Common,

�Thomson effect� Absolute zero

0

100 0

100

008

�Common, repeatable calibration reference points

�"Centigrade" scale

1821: It was a very good year

�1800 ad �1900 ad


�The Seebeck effect

�Pt 100 @ O deg.C

�Davy: The RTD

�

009

�Pt 100 @ O deg.C�

The 1900's: Electronic sensors

�1900 ad �2000 ad


�Thermistor�1 uA/K

�IC sensor

�IPTS 1968 �IPTS 1990

010

�IPTS 1968

�"Degree Kelvin">> "kelvins"�"Centigrade">> " Celsius"

�IPTS 1990

Temperature scales

-273.15

Absolute zero

�Celsius 0 100

Freezing point H O2

Boiling point H O 2


-273.15

0

-459.67

0

Celsius

�Kelvin

�Fahrenheit

�Rankine

0

273.15

32

427.67

100

373.15

212

671.67

011

0 427.67 671.67

�"Standard" is "better": � Reliable reference

points� Easy to understand

IPTS '90: More calibration points

– 273.16: TP H2O

1234.93: FP – 1337.33: FP

Au– 234.3156: TP

Hg

– 1357.77: FP Cu


– 83.8058: TP Ar54.3584: TP

Large gap

– 1234.93: FP AgAu

– 692.677: FP Zn

Hg

– 505.078: FP

– 933.473: FP Al

012

Ar– 54.3584: TP

O2– 24.5561: TP Ne– 20.3: BP H2– 17 Liq/vapor H2 – 13.81 TP H2

– 429.7485: FP In

– 302.9146: MP Ga

– 505.078: FP Sn

– 3 to 5: Vapor He

Agenda






A2


Bimetal thermometer

�Two dissimilar metals, tightly

�Forces due to thermal �Result


metals, tightly bonded

thermal expansion

�Result

�Bimetallic thermometer� Poor accuracy� Hysteresis

�Thermal expansion causes big

0100 300

200

400

013

�Thermal expansion causes big problems in other designs:� IC bonds� Mechanical interference

Liquid thermometer; Paints100


0

�Liquid-filled thermometer� Accurate over a small range

�Thermally-sensitive paints� Irreversible change� Low resolution� Useful in hard-to-measure areas

014

� Accurate over a small range� Accuracy & resolution= f(length)� Range limited by liquid� Fragile� Large thermal mass� Slow

Agenda






A3

� Thermistor, IC� Thermocouple

�Summary & Examples

Optical Pyrometer


�Infrared Radiation-sensitive�Photodiode or photoresistor�Accuracy= f{emissivity}�Useful @ very high temperatures�Non-contacting

015

�Non-contacting�Very expensive�Not very accurate

Agenda






A4



Resistance Temperature Detector

Most accurate & stable


�Most accurate & stable�Good to 800 degrees Celsius�Resistance= f{Absolute T}�Self-heating a problem�Low resistance�Nonlinear

016

�Nonlinear

RTD Equation

�R= 100 Ohms @ O C�Callendar-Van Deusen Equation:


�R=Ro(1+aT) - Ro(ad(.01T)(.01T-1))� Ro=100 @ O C� a= 0.00385 / - C � d= 1.49

Callendar-Van Deusen Equation:

R

300

For T>OC:

for Pt

017

�0 200 400 600 800T

300200100

Nonlinearity

Measuring an RTD: 2 -wire method

100Rlead

V-

+I ref= 5 mAPt

�Rx

Rlead


��R= Iref*(Rx + 2* Rlead)

� Error= 2 /.385= more than 5 degrees C for 1 ohm Rlead!

�Self-heating:� For 0.5 V signal, I= 5mA; P=.5*.005=2.5 mwatts� @ 1 mW/deg C, Error = 2.5 deg C!

-

018

� @ 1 mW/deg C, Error = 2.5 deg C!

�Moral: Minimize Iref; Use 4-wire method� If you must use 2-wire, NULL out the lead resistance

The 4-Wire technique

100Rlead=1 V

-

+I ref= 5 mA�

�

Rx


� R= Iref * Rx� Error not a function of R in source or sense

leads� No error due to changes in lead R

-�

019

� Twice as much wire� Twice as many scanner channels� Usually slower than 2-wire

Offset compensation

100 V+

I ref (switched)

Voffset

�


�Eliminates thermal voltages� Measure V without I applied� Measure V I appliedWith

V-

I ref (switched)�

020

� Measure V I applied

R=V

I

With

Bridge method

100�1000�


V

�High resolution (DMM stays on most sensitive range)

�Nonlinear output�Bridge resistors too

100�1000�

021

�Bridge resistors too close to heat source

3-Wire bridge

V

1000 100

Rlead 1

3-Wire


��

V

100

1000

�Keeps bridge away from heat source�Break DMM lead (dashed line); connect to RTD through 3rd "sense" wire

Rlead 2Sense wire

3-Wire PRTD�

�

022

RTD through 3rd "sense" wire�If Rlead 1= Rlead 2, sense wire makes error small

�Series resistance of sense wire causes no error

Agenda

�Background, history





A5



Electrical sensors: Thermistor

5k V-

+I= 0.1 mARlead=1� �

Rlead=1�


�Hi-Z; Sensitive: 5 k @ 25C; R = 4%/deg C

-

�2-Wire method: R= I * (Rthmr + 2*Rlead)

�

�

�Limited range

023

� Lead R Error= 2 /400= 0.005 degrees C�Low thermal mass: High self-heating�Very nonlinear

�

I.C. Sensor

I= 1 uA/K


�

+-

V

I= 1 uA/K

5V �100

960= 1mV/K

�High output�Very linear�Accurate @ room ambient

�Limited range�

024

�Limited range�Cheap

Summary: Absolute T devices

RTD Thermistor I.C.


�Expensive

RTD

�Most accurate �Most stable�Fairly linear

Thermistor

�High output�Fast�2-wire meas.

�Very nonlinear

I.C.

�High output�Most linear�Inexpensive

�Limited variety

025

�Expensive�Slow�Needs I source�Self-heating�4-wire meas.

�Very nonlinear�Limited range�Needs I source�Self-heating�Fragile

�Limited variety�Limited range�Needs V source�Self-heating

Agenda






Thermocouple

A6


ThermocouplesThe Gradient Theory

TxTa


Tx

V

Tx

�The WIRE is the sensor, not the junction

�The Seebeck coefficient (e) is a

026

V= e(T) dT

Ta

coefficient (e) is a function of temperature

Making a thermocouple

Tx

Ta

B


�Two wires make a thermocouple

�Voltage output is nonzero if metals are

Tx Ta

TxV

TaA

B

027

nonzero if metals are not the sameV= e dT

Ta

+ e dT

Tx

A B

Gradient theory also says...

Tx

Ta

A


�If wires are the same type, or if there is one wire, and both ends are at the same temperature,

Tx Ta

TxV

TaA

A

028

same temperature, output= Zero.V= e dT

Ta

Tx

+ e dT = 0

Tx

A A

Now try to measure it:

�Theoretically,Vab= f{Tx-Tab}

Con

aTx

Fe

b


�But, try to measure it with a DMM:

Tx

Fe

V

Cu

=

Conb

Fe

Tx

Cu

V

�Result: 3 unequal junctions, all at unknown temperatures

029

ConCu=

Cu Con

Solution: Reference Thermocouple�Problems: a) 3 different thermocouples,

b) 3 unknown temperatures�Solutions: a) Add an opposing thermocouple


�Solutions: a) Add an opposing thermocoupleb) Use a known reference temp.

Cu

V Con

Fe

Tx

Isothermal block

Cu

V Con

Fe

Tx Add

030

V

Cu Fe

Tref= 0 C

Con

o

V

Cu Fe

Tref

The Classical Method

Cu Fe

Tx�If both Cu junctions are at


V

Cu Fe

Tref= 0 C

Con

Tx

o

�If both Cu junctions are at same T, the two "batteries" cancel

�Tref is an ice bath (sometimes an electronic ice bath)

�All T/C tables are

031

�All T/C tables are referenced to an ice bath

�V= f{Tx-Tref}

�Question: How can we eliminate the ice bath?

Eliminating the ice bath

Cu Fe�Don't force Tref to icepoint, just measure it


Tref

Cu

V

Cu Fe

Con

Fe

Tx

measure it�Compensate for Tref mathematically:V=f{ Tx - Tref }

�If we know Tref , we can

TiceTice

032

Cu Fe �If we know Tref , we can

compute Tx.Tice

Eliminating the second T/C

�Extend the isothermal blockCu Fe


�Extend the isothermal block�If isothermal, V1-V2=0

2V

Cu Fe

Con

Tx

1

Cu

V Con

Fe

Tx

2Tref

Tref

033

1V

Cu

Con2

1

Tref

The Algorithm for one T/C

�Measure Tref: RTD, IC or thermistor�Tref ==> Vref @ O C for Type

Cu Fe

Tx

Trefo


�Tref ==> Vref @ O C for Type J(Fe-C)

�Know V, Know Vref: Compute Vx�Solve for using Vx

TxV

Cu

ConTref

VxCompute

034

0 Tref

VxVref

Tx

ComputeVx=V+Vref V

o

LinearizationV

Small sectors


�Polynomial: T=a +a V +a V +a V +.... a V�Nested (faster): T=a +V(a +V(a +V(a

2

1 2 30 1 2 3

39

9

0

0 Tref Txo T

Small sectors

035

�Nested (faster): T=a +V(a +V(a +V(a +.......)))))))))

�Small sectors (faster): T=T +bV+cV �Lookup table: Fastest, most memory

1 2 32

00

Common ThermocouplesmV

60

EE

K

E

Platinum T/Cs


0 500 1000 2000 deg C

20

40

60

R

NKJ

ST

Platinum T/CsBase Metal T/Cs

036

0 500 1000 2000

�All have Seebeck coefficients in MICROvolts/deg.C

Common Thermocouples

SeebeckCoeff: uV/CType Metals


JKTSEN

Fe-ConNi-CrCu-ConPt/Rh-PtNi/Cr-ConNi/Cr/Si-Ni/Si

504038105939

�Microvolt output is a tough measurement

�Type "N" is fairly new.. more rugged and higher temp. than type K, but still cheap

037

N Ni/Cr/Si-Ni/Si 39 K, but still cheap

Extension Wires �Possible problemhere


Large extension wiresSmall diametermeasurementwires

�Extension wires are cheaper, more rugged,

038

�Extension wires are cheaper, more rugged, but not exactly the same characteristic curve as the T/C.

�Keep extension/TC junction near room temperature

�Where is most of the signal generated in this circuit?

Noise: DMM Glossary

DMM

Normal Modeac NOISE

HI


DMMInputResistance

Normal Modedc SIGNAL

DMM

HI

HI

LO

�Normal Mode: In series with input�Common Mode: Both HI and LOterminals driven equally

039

DMMInputResistance

Common Modeac NOISE

LO

terminals driven equally

Generating noise

DMMHI

ElectrostaticNoise

MagneticNoise


Normal Mode

�Large surface area, high Rlead: Max. static coupling

�Large loop area: Max. magnetic coupling

DMMInputResistance dc SIGNAL

DMM

HI

HI

LO

�Large R lead, small R leak:

040

DMMInputResistance LO

Common Modeac source

R lead

R leak

Common Mode Current

�Large R lead, small R leak: Max.common mode noise

Eliminating noise

DMHI

ElectrostaticNoise

MagneticNoise


Normal Modedc SIGNAL

�Filter, shielding, small loop area(Caution: filter slows down the measurement)

�Make R leak close to

DMMInputR

DMM

HI

LO

041

�Make R leak close to MInput R L

OCommon Modeac source

R leak

Common Mode Current

- +

Magnetic Noise

�Magnetic coupling

DMMInduced I


DMMInputResistance

Induced I

042

�Minimize area�Twist leads�Move away from strong fields

Reducing Magnetic Noise

�Equal and opposite induced currents

DMM


DMMInputResistance

�Even with twisted pair:� Minimize area� Move away from strong fields

043

� Move away from strong fields

Electrostatic noiseAC Noisesource

Stray capacitances


DMMInputResistance

Stray resistances

Stray capacitances

Inoise

044

�Stray capacitance causes I noise�DMM resistance to ground is important

Reducing Electrostatic Coupling AC Noise source


DMMInputResistance

Shield shunts stray current

HI

LO

Rleak

045

�Shield shunts stray current�For noise coupled to the tip, Rleak is still important

Rleak

A scanning system for T/Cs

�One thermistor,


OHMsConv.

�One thermistor,multiple T/C channels

�Noise reduction�CPU linearizes T/C�DMM must be

very high quality

Conv.

046

HI

LO

Floating Circuitry Grounded Circuitry

Isolators

uP uPI/O

(HP-IB,RS-232) To

ComputerROMLookup

IntegratingA/D

Errors in the system

Thermal emf

Ref. Thermistor cal, linearity

T/C Calibration& Wire errors

Ref. Block Thermal gradient


OHMsConv.

Linearization algorithm

ReferenceThermistorOhmsmeasurement

Ref. Thermistor cal, linearity

Extension wirejunction error

Conv.

047

HI

LO

Floating Circuitry Grounded Circuitry

Isolators

uP uPI/O

(HP-IB,RS-232)

ROMLookup

IntegratingA/D

DMM offset, linearity, thermal emf, noise

Physical errors

�Shorts, shunt impedance �Sensor accuracy


Shorts, shunt impedance�Galvanic action�Decalibration

Sensor accuracy�Thermal contact�Thermal shunting

048

Physical Errors


�Water droplets cause galvanic action; huge offsets

�Hot spot causes shunt Z, meter shows the WRONG temperature

049

�Exceeding the T/C's range can cause permanent offset

�Real T/C's have absolute accuracy of 1 deg C @ 25C: Calibrate often and take care

Physical error: Thermal contact


Surface probe

�Make sure thermal mass is much smallerthan that of object being measured

050

than that of object being measured

Physical errors: Decalibration

300 C350 C

975 C


1000 C200 C300 C 975 C

100 C

This section Don't exceed Tmax of T/C

051

This section produces theENTIRE signal

�Don't exceed Tmax of T/C�Temp. cycling causes work-hardening,decalibration

�Replace the GRADIENT section

Agenda






A7



The basic 4 temperature sensors

ThermocoupleRTD Thermistor I.C.


Thermocouple

�Wide variety�Cheap�Wide T. range�No self-heating

�Hard to measure�Relative T. only

�Expensive�Slow�Needs I source

�Most accurate �Most stable�Fairly linear

�High output�Fast�2-wire meas.

�Very nonlinear�Limited range�Needs I source

�High output�Most linear�Cheap

�Limited variety�Limited range�Needs V

052

�Relative T. only�Nonlinear�Special

connectors

�Needs I source�Self-heating�4-wire meas.

�Needs I source�Self-heating�Fragile

�Needs V source

�Self-heating

Absolute temperature sensors

Summary

�Innovation by itself is not enough...


�Innovation by itself is not enough...you must develop standards

�Temperature is a very difficult, mostly empirical measurement

�Careful attention to detail is required

053

Examples

Measurement Sensor


�Photochemical process control:

�Flower petal:

�Molten glass:

�RTD (most accurate)

�Thermistor(lowest thermal mass)

�Optical pyrometer (hi temp, no contact)

�RTD (if <800C); or T/C

054

�Induction furnace:

�100 degree Heat aging oven:

�RTD (if <800C); or T/C(Beware magnetic I noise)

�Any of the 4 sensors

Documents

Practical Temperature Measurements - Unicampelnatan/ee641/Temperatura.pdf · Temperature Measurements 001. Agenda Background, history Mechanical sensors Hewlett-Packard Classroom