Agilent Technologies Classroom Series
Practical
Temperature
Measurements
001
Agenda
Background, history
Mechanical sensors
Electrical sensors
 Optical
Pyrometer
 RTD
 Thermistor, IC
 Thermocouple
Summary & Examples
A1
What is Temperature?
A scalar quantity that determines the
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
Gap reduces conduction
Metallization reflects radiation
Vacuum reduces convection
004
Thermal Mass
Don't let the measuring
Sensor
device change the
temperature of what you're
measuring.
Response time =
 f{Thermal mass}
 f{Measuring device}
Sensor
005
Temperature errors
What is YOUR normal temperature?
Thermometer accuracy, resolution
Contact time
Thermal mass of thermometer, tongue
Human error in reading
97.6 98.6 99.6
36.5 37 37.5
006
History of temperature sensors
1600
ad
1700
12
ad
96
Fahrenheit
1
 Instrument
0
Galileo: First
temp. sensor
 pressure-
sensitive
 not repeatable
 Early
thermometers
 Not
repeatable
 No good way
to calibrate
Maker
 12*8=96
points
 Hg:
Repeatable
 One
standard
scale
007
The 1700's: Standardization
1700 ad
1800 ad
0
100
Thomson effect
 Absolute zero
100
Celsius:
Common,
repeatable
calibration
reference points
0
"Centigrade"
scale
008
1821: It was a very good year
1800 ad
1900 ad
The Seebeck
effect
Davy: The RTD
Pt 100
d @ O deg.C
009
The 1900's: Electronic sensors
1900 ad
Thermistor
2000 ad
1 uA/K
IC sensor
IPTS 1968 IPTS
1990
"Degree Kelvin">> "kelvins"
"Centigrade">> " Celsius"
010
Temperature scales
Freezing
point H2O
Absolute
zero
Celsius
0
100
Kelvin
273.15
373.15
32
212
-273.15
0
Fahrenheit
-459.67
0
Boiling point
H O2
Rankine
427.67
671.67
"Standard" is "better":

Reliable reference
points
 Easy to understand
011
IPTS '90: More calibration points
–
–
273.16: TP
H2O
234.3156: TP
Hg
Large gap
–
–
–
–
–
–
–
–
–
–
–
–
83.8058: TP
Ar
54.3584: TP
O2
24.5561: TP
20.3:
BP
Ne
17 Liq/vapor
H2
13.81
TP
H2
3 to 5: Vapor
H2
He
–
–
1357.77: FP
Cu
1337.33: FP
Au
1234.93: FP
Ag
933.473: FP
Al
692.677: FP
Zn
505.078: FP
Sn
429.7485: FP
In
–
302.9146: MP
Ga
012
Agenda
Background, history
Mechanical sensors
Electrical sensors
 Optical
Pyrometer
 RTD
 Thermistor, IC
 Thermocouple
Summary & Examples
A2
Bimetal thermometer
Two dissimilar Forces due to
thermal
expansion
metals, tightly
bonded
0
Result
100 200 300 Bimetallic thermometer
400
 Poor accuracy
 Hysteresis
Thermal expansion causes big
problems in other designs:
 IC bonds
 Mechanical interference
013
Liquid thermometer; Paints
100
0
Thermally-sensitive paints
Liquid-filled thermometer






 Irreversible change
 Low resolution
 Useful in hard-to-measure area
Accurate over a small range
Accuracy & resolution= f(length)
Range limited by liquid
Fragile
Large thermal mass
Slow
014
Agenda
Background, history
Mechanical sensors
Electrical sensors
 Optical
Pyrometer
 RTD
 Thermistor, IC
 Thermocouple
Summary & Examples
A3
Optical Pyrometer
Infrared Radiation-sensitive
Photodiode or photoresistor
Accuracy= f{emissivity}
Useful @ very high temperatures
Non-contacting
Very expensive
Not very accurate
015
Agenda
Background, history
Mechanical sensors
Electrical sensors
 Optical
Pyrometer
 RTD
 Thermistor, IC
 Thermocouple
Summary & Examples
A4
Resistance Temperature Detector
Most accurate & stable
Good to 800 degrees Celsius
Resistance= f{Absolute T}
Self-heating a problem
Low resistance
Nonlinear
016
RTD Equation
R= 100 Ohms @ O C
Callendar-Van Deusen Equation:
For T>OC:
R=Ro(1+aT) - Ro(ad(.01T)(.01T-1))



R
Ro=100 @ O C
a= 0.00385
/ - C for Pt
d= 1.49
300
200
100
0
Nonlinearity
200
400
600
800
T
017
Measuring an RTD: 2-wire method
Rx
100d
Pt
Rlead
+
Rlead
-
V
I ref= 5 mA
R= Iref*(Rx + 2* Rlead)

Error= 2d /.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!
Moral: Minimize Iref; Use 4-wire method
 If you must use 2-wire, NULL out the lead resistance
018
The 4-Wire technique
Rx
100d
+
Rlead=1d
-
V
I ref= 5 mA
 R= Iref * Rx
 Error not a function of R in source or sense
leads
 No error due to changes in lead R



Twice as much wire
Twice as many scanner channels
Usually slower than 2-wire
019
Offset compensation
Voffset
+
100d
-
V
I ref (switched)
Eliminates thermal voltages
 Measure V without I applied
 Measure V With I applied
R=
V
I
020
Bridge method
100 d
d
1000
V
d
1000
100d
High resolution (DMM
stays on most sensitive
range)
Nonlinear output
Bridge resistors too
close to heat source
021
3-Wire bridge
100d
1000d
Rlead 1
V
1000d
Sense wire
3-Wire
PRTD
Rlead 2
Keeps bridge away from heat source
Break DMM lead (dashed line); connect to
100d
RTD through 3rd "sense" wire
If Rlead 1= Rlead 2, sense wire makes
error small
Series resistance of sense wire causes no
error
022
Agenda
Background, history
Mechanical sensors
Electrical sensors
 Optical
Pyrometer
 RTD
 Thermistor, IC
 Thermocouple
Summary & Examples
A5
Electrical sensors: Thermistor
Rlead=1
5k d
d
Rlead=1d
+
-
Hi-Z; Sensitive: 5 k
V
d
I= 0.1 mA
@ 25C;
R = 4%/deg C
Limited range
2-Wire method: R= I * (Rthmr + 2*Rlead)

Lead R Error= 2
d /400= 0.005 degrees C
Low thermal mass: High self-heating
Very nonlinear
023
I.C. Sensor
I= 1 uA/K
+
5V -
100 d
V = 1mV/K
960d
High output
Very linear
Accurate @
room ambient
Limited range
Cheap
024
Summary: Absolute T devices
RTD
Thermistor
I.C.
Most accurate
Most stable
Fairly linear
High output
Fast
2-wire meas.
High output
Most linear
Inexpensive
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
025
Agenda
Background, history
Mechanical sensors
Electrical sensors
 Optical
Pyrometer
 RTD
 Thermistor, IC
 Thermocouple
Summary & Examples
A6
Thermocouples
The Gradient Theory
Ta
Tx
The WIRE is the
V=
V
sensor, not the
junction
Tx
The Seebeck
e(T) dT
Ta
coefficient (e) is a
function of
temperature
026
Making a thermocouple
Ta
Tx
B
V
Two wires make a
A
Ta
thermocouple
V=
e
Ta
A
Voltage output is
Ta
Tx
dT +
e
B
dT
nonzero if metals are
not the same
Tx
027
Gradient theory also says...
Ta
Tx
A
V
If wires are the
A
Ta
Ta
Tx
e dT +
A
V=
Ta
e
A
dT = 0
same type, or if there
is one wire, and
both ends are at the
same temperature,
output= Zero.
Tx
028
Now try to measure it:
a
Fe
b
Tx Theoretically,
Vab= f{Tx-Tab}
Con
But, try to measure it with a DMM:
Cu
Tx
V
Cu
Cu
Fe
Con
=
Fe
V
Tx
Con
Cu
Result: 3 unequal junctions, all at unknown
temperatures
029
Solution: Reference Thermocouple
Problems: a) 3 different thermocouples,
b) 3 unknown temperatures
Solutions: a) Add an opposing thermocouple
b) Use a known reference temp. Isothermal
block
Fe
Cu
Cu Fe
Tx
Add
Tx
Con
V
V Con
Tref
Tref
= 0 oC
Cu
Fe
Cu Fe
030
The Classical Method
Cu Fe
Tx
V
Con
Tref
o
=0C
Cu Fe
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
referenced to an ice bath
V= f{Tx-Tref}
Question: How can we eliminate the ice bath?
031
Eliminating the ice bath
Don't force Tref to icepoint, just
Cu
V
Fe
Con
Cu
Fe
measure it
Compensate for Tref
Tx
mathematically:
V=f{ Tx
- Tref
}
Tice
Tref
Tice
If we know Tref
, we can
Tice
compute Tx.
032
Eliminating the second T/C
Cu
Fe
Tx
V
Extend the isothermal block
If isothermal, V1-V2=0
2
Con
Cu
Fe
Tref
Cu
Fe
1
Tx
V
Tref
2
Con
Cu
1
033
The Algorithm for one T/C
Cu
V
Tref
Cu
Fe
IC or thermistor
Tx Measure Tref: RTD,
o
Tref ==> Vref @ O C for Type J(FeCon C)
Know V, Know Vref: Compute Vx
Solve for Tx using Vx
Compute
Vx=V+Vref
Vx
Vref
0 o Tref
V
Tx
034
V
Linearization
Small sectors
0 o Tref
Tx
T
2
3 V +.... a9 V
Polynomial: T=a
+a
V
+a
V
+a
0 1
2
3
9
Nested (faster): T=a
0 +V(a
1 +V(a
2 +V(a
3
2
+.......)))))))))
0
Small sectors (faster): T=T +bV+cV
Lookup table: Fastest, most memory
035
Common Thermocouples
mV
E
60
Platinum T/Cs
Base Metal T/Cs
K
J N
40
20
T
0
500
RS
1000
2000
deg C
All have Seebeck coefficients in MICROvolts/deg.C
036
Common Thermocouples
Type
J
K
T
S
E
N
Metals
Seebeck
Coeff: uV/C
Fe-Con
Ni-Cr
Cu-Con
Pt/Rh-Pt
Ni/Cr-Con
Ni/Cr/Si-Ni/Si
50
40
38
10
59
39
Microvolt output is a
tough measurement
Type "N" is fairly new..
more rugged and
higher temp. than type
K, but still cheap
037
Extension Wires
Possible problem
here
Large extension wires
Small diameter
measurement
wires
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?
038
Noise: DMM Glossary
Normal Mode
ac NOISE
DMM
Input
Resistance
HI
LO
Normal Mode
dc SIGNAL
Normal Mode: In series with input
Common Mode: Both HI and LO
HI
DMM
Input
Resistance
terminals driven equally
LO
Common Mode
ac NOISE
039
Generating noise
Electrostatic
Noise
DMM
Input
Resistance
Magnetic
Noise
HI
Normal Mode
LO
dc SIGNAL
Large surface area, high Rlead: Max. static
DMM
Input
Resistance
R leak
coupling
HILarge loop area: Max. magnetic coupling
Large R lead, small R leak:
R lead
Max.
LO
common mode noise
Common Mode Current
Common Mode
ac source
040
Eliminating noise
Electrostati
c
Noise
DM
M
Inpu
tR
Magneti
c
Noise
HI
Normal Mode
dc SIGNAL
L
O
Filter, shielding, small loop area
(Caution: filter slows down the
measurement)
HI
DM
M
Input
R
R
leak
Make R leak close to
L
O
-
+
Common Mode
Current
Common
Mode
ac source
041
Magnetic Noise
Magnetic coupling
DMM
Input
Resistance
Induced I
Minimize area
Twist leads
Move away from strong fields
042
Reducing Magnetic Noise
Equal and opposite induced currents
DMM
Input
Resistance
Even with twisted pair:
 Minimize area
 Move away from strong fields
043
Electrostatic noise
AC Noise
source
Stray capacitances
DMM
Input
Resistance
Inoise
Stray resistances
Stray capacitance causes I noise
DMM resistance to ground is important
044
Reducing Electrostatic
AC Noise source
Coupling
DMM
Input
Resistance
HI
LO
Rleak
Shield shunts stray current
For noise coupled to the tip,
Rleak is still important
045
A scanning system for T/Cs
One thermistor,
multiple T/C channels
Noise reduction
CPU linearizes T/C
DMM must be
very high quality
OHMs
Conv.
Isolators
HI
uP
LO
Integrating
A/D
Floating Circuitry
ROM
Lookup
uP
I/O
(HP-IB,
RS-232)
To
Computer
Grounded Circuitry
046
Errors in the system
Ref. Block Thermal gradient
T/C Calibration
& Wire errors
Thermal emf
Ref. Thermistor cal, linearity
OHMs
Conv.
Reference
Thermistor
Ohms
measurement
Linearization
algorithm
Isolators
HI
uP
LO
Integrating
A/D
Floating Circuitry
ROM
Lookup
DMM offset, linearity, thermal emf, noise
uP
Extension wire
junction error
I/O
(HP-IB,
RS-232)
Grounded Circuitry
047
Physical errors
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, mete
shows the WRONG temperatur
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
049
Physical error: Thermal contact
Surface probe
Make sure thermal mass is much smaller
than that of object being measured
050
Physical errors: Decalibration
350 C
300 C
200 C
100 C
975 C
1000 C
This section
Don't exceed Tmax of T/C
produces the Temp. cycling causes work-hardening,
ENTIRE signal decalibration
Replace the GRADIENT section
051
Agenda
Background, history
Mechanical sensors
Electrical sensors
 Optical
Pyrometer
 RTD
 Thermistor, IC
 Thermocouple
Summary & Examples
A7
The basic 4 temperature sensors
RTD Thermistor
Most accurate High output
Fast
Most stable
2-wire meas.
Fairly linear
Expensive
Very nonlinear
Slow
Limited range
Needs I source Needs I source
Self-heating
Self-heating
4-wire meas. Fragile
I.C.
Thermocouple
High output
Most linear
Cheap
Limited variety
Limited range
Needs V
source
Self-heating
Wide variety
Cheap
Wide T. range
No self-heating
Hard to measure
Relative T. only
Nonlinear
Special
connectors
Absolute temperature sensors
052
Summary
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
Photochemical process
Sensor
RTD (most accurate)
control:
Thermistor
Flower petal:
Molten glass:
Induction furnace:
100 degree Heat aging oven:
(lowest thermal mass)
Optical pyrometer
(hi temp, no contact)
RTD (if <800C); or T/C
(Beware magnetic I
noise)
Any of the 4 sensors
054