Transcript Overview of Temperature Measurement

Overview of Temperature
Measurement
Figures are from www.omega.com “Practical Guidelines for Temperature
Measurement” unless otherwise noted
Outline

Thermocouples
– overview, reference junction, proper connections, types, special
limits of error wire, time constants, sheathing, potential problems,
DAQ setup

RTDs
– overview, bridges, calibration, accuracy, response time, potentail
problems


Thermistors
Infrared Thermometry
– fundamentals, emissivity determination, field of view

Other
– Non-electronic measurement, thin-film heat flux gauge


Temperature Controllers
How to Choose
– Standards, cost, accuracy, stability, sensitivity, size, contact/noncontact, temperature range, fluid type
Thermocouples

Seebeck effect
– If two wires of dissimilar metals are joined at both ends and
one end is heated, current will flow.
– If the circuit is broken, there will be an open circuit voltage
across the wires.
– Voltage is a function of temperature and metal types.
– For small DT’s, the relationship with temperature is linear
DV  DT
– For larger DT’s, non-linearities may occur.
Measuring the Thermocouple Voltage

If you attach the thermocouple directly to a voltmeter, you will
have problems.

You have just created another junction! Your displayed voltage
will be proportional to the difference between J1 and J2 (and
hence T1 and T2). Note that this is “Type T” thermocouple.
External Reference Junction

A solution is to put J2 in an ice-bath; then you know
T2, and your output voltage will be proportional to
T1-T2.
Other types of thermocouples

Many thermocouples don’t have one copper wire.
Shown below is a “Type J” thermocouple.

If the two terminals aren’t at the same temperature,
this also creates an error.
Isothermal Block

The block is an electrical insulator but good heat
conductor. This way the voltages for J3 and J4 cancel
out. Thermocouple data acquisition set-ups include
these isothermal blocks.

If we eliminate the ice-bath, then the isothermal block
temperature is our reference temperature
V   T1  Tblock 
Software Compensation



How can we find the temperature of the block? Use a
thermister or RTD.
Once the temperature is known, the voltage
associated with that temperature can be subtracted
off.
Then why use thermocouples at all?
– Thermocouples are cheaper, smaller, more flexible and
rugged, and operate over a wider temperature range.

Most data acquisition systems have software
compensation built in. To use Labview,you’ll need to
know if you have a thermister or RTD.
Hardware Compensation


With hardware compensation, the temperature of the
isothermal block again is measured, and then a
battery is used to cancel out the voltage of the
reference junction.
This is also called an “electronic ice point reference”.
With this reference, you can use a normal voltmeter
instead of a thermocouple reader. You need a
separate ice-point reference for every type of
thermocouple.
Making Thermocouple Beads




Soldering, silver-soldering, butt or spot or beaded gas
welding, crimping, and twisting are all OK.
The third metal introduced doesn’t effect results as
long as the temperature everywhere in the bead is
the same.
Welding should be done carefully so as to not
degrade the metals.
If you consistently will need a lot of thermocouples,
you can buy a thermocouple welder; you stick the two
ends into a hole, hit a button, and the welding is
done.
Time Constant vs. Wire Diameter
Time Constant vs. Wire Diameter, cont.
Thermocouple Types
If you do your own
calibration, you can
usually improve on the
listed uncertainties.
Thermocouple Types, cont.






Type B – very poor below 50ºC; reference junction temperature
not important since voltage output is about the same from 0 to
42 ºC
Type E – good for low temperatures since dV/dT () is high for
low temperatures
Type J – cheap because one wire is iron; high sensitivity but
also high uncertainty (iron impurities cause inaccuracy)
Type T – good accuracy but low max temperature (400 ºC); one
lead is copper, making connections easier; watch for heat being
conducted along the copper wire, changing your surface temp
Type K – popular type since it has decent accuracy and a wide
temperature range; some instability (drift) over time
Type N – most stable over time when exposed to elevated
temperatures for long periods
Sheathing and SLE





“Special Limits of Error” wire can be used to improve accuracy.
Sheathing of wires protects them from the environment (fracture,
oxidation, etc.) and shields them from electrical interference.
The sheath should extend completely through the medium of
interest. Outside the medium of interest it can be reduced.
Sometimes the bead is exposed and only the wire is covered by
the sheath. In harsher environments, the bead is also covered.
This will increase the time constant.
Platinum wires should be sheathed in non-metallic sheaths
since they have a problem with metallic vapor diffusion at high
temperatures.
Sheathing, cont.

From J. Nicholas & D. White, 2001, Traceable Temperatures: An Introduction to
Temperature Measurement and Calibration, 2nd ed. John Wiley & Sons.
Potential Problems

Poor bead construction
– Weld changed material characteristics because the weld
temp. was too high.
– Large solder bead with temperature gradient across it

Decalibration
– If thermocouples are used for very high or cold
temperatures, wire properties can change due to diffusion of
insulation or atmosphere particles into the wire, coldworking, or annealing.
– Inhomogeneities in the wire; these are especially bad in
areas with large temperature gradients; esp. common in iron.
Metallic sleeving can help reduce their effect on the final
temperature reading.
Potential Problems, cont.

Shunt impedence
– As temperature goes up, the resistance of many insulation
types goes down. At high enough temperatures, this creates
a “virtual junction”. This is especially problematic for small
diameter wires.

Galvanic Action
– The dyes in some insulations form an electrolyte in the
water. This creates a galvanic action with a resulting emf
potentially many times that of the thermocouple. Use an
appropriate shield for a wet environment. “T Type”
thermocouples have less of a problem with this.
Potential Problems, cont.

Thermal shunting
– It takes energy to heat the thermocouple, which results in a small
decrease in the surroundings’ temperature. For tiny spaces, this
may be a problem.
– Use small wire (with a small thermal mass) to help alleviate this
problem. Small-diameter wire is more susceptible to decalibration
and shunt impedence problems. Extension wire helps alleviate this
problem. Have short leads on the thermocouple, and connect them
to the same type of extension wire which is larger. Extension wire
has a smaller temperature range than normal wire.

Noise
– Several types of circuit set-ups help reduce line-related noise. You
can set your data acquisition system up with a filter, too.
– Small-diameter wires have more of a problem with noise.
Potential Problems

Conduction along the thermocouple wire
– In areas of large temperature gradient, heat can be
conducted along the thermocouple wire, changing the bead
temperature.
– Small diameter wires conduct less of this heat.
– T-type thermocouples have more of a problem with this than
most other types since one of the leads is made of copper
which has a high thermal conductivity.

Inaccurate ice-point
Data Acquisition Systems for
Thermocouples


Agilent, HP, and National Instruments are probably
the most popular DAQ systems
Example National Instruments DAQ setup for
thermocouples and costs
item
16-bit temperature data acquisition card
analog input module for thermocouples
chassis
terminal block for thermocouples
shielded cable
Total cost:
part number
PCI 6232E
SCXI-1112
SCXI-1000
SCXI-1303
SH68-68-EP
cost
1495
695
695
275
95
3255
Things to Note During System Assembly










Make sure materials are clean, esp. for high temperatures.
Check the temperature range of materials. Materials may degrade
significantly before the highest temperature listed.
Make sure you have a good isothermal junction.
Use enough wire that there are no temperature gradients where it’s
connected to your DAQ system.
If you’re using thermocouple connectors, use the right type for your
wire.
If you’re using a DAQ system, use the right set-up for thermocouples.
Check the ice-point reference.
Provide proper insulation for harsh environments.
Pass a hair-dryer over the wire. The temperature reading should only
change when you pass it over the bead.
Mount a thermocouple only on a surface that is not electrically live
(watch for this when measuring temperatures of electronics).
RTDs (Resistance Temperature
Detectors)





Resistivity of metals is a function of temperature.
Platinum often used since it can be used for a wide temperature
range and has excellent stability. Nickel or nickel alloys are used
as well, but they aren’t as accurate.
In several common configurations, the platinum wire is exposed
directly to air (called a bird-cage element), wound around a
bobbin and then sealed in molten glass, or threaded through a
ceramic cylinder.
Metal film RTDs are new. To make these, a platinum or metalglass slurry film is deposited onto a ceramic substrate. The
substrate is then etched with a laser. These RTDs are very small
but aren’t as stable (and hence accurate).
RTDs are more accurate but also larger and more expensive
than thermocouples.
RTD geometry
From Nicholas & White, Traceable Temperatures.



Sheathing: stainless steel or iconel, glass, alumina, quartz
Metal sheath can cause contamination at high temperatures and
are best below 250ºC.
At very high temperatures, quartz and high-purity alumina are
best to prevent contamination.
Resistance Measurement

Several different bridge circuits are used to determine
the resistance. Bridge circuits help improve the
accuracy of the measurements significantly. Bridge
output voltage is a function of the RTD resistance.
Resistance/Temperature Conversion


Published equations relating bridge voltage to
temperature can be used.
For very accurate results, do your own calibration.
– Several electronic calibrators are available.
– The most accurate calibration that you can do easily yourself
is to use a constant temperature bath and NIST-traceable
thermometers. You then can make your own calibration
curve correlating temperature and voltage.
Accuracy and Response Time

Response time is longer than thermocouples; for a ¼
sheath, response time can easily be 10 s.
Potential Problems




RTDs are more fragile than thermocouples.
An external current must be supplied to the RTD. This current
can heat the RTD, altering the results. For situations with high
heat transfer coefficients, this error is small since the heat is
dissipated to air. For small diameter thermocouples and still air
this error is the largest. Use the largest RTD possible and
smallest external current possible to minimize this error.
Be careful about the way you set up your measurement device.
Attaching it can change the voltage.
When the platinum is connected to copper connectors, a voltage
difference will occur (as in thermocouples). This voltage must be
subtracted off.
Thermistors





Thermistors also measure the change in resistance with
temperature.
Thermistors are very sensitive (up to 100 times more than RTDs
and 1000 times more than thermocouples) and can detect very
small changes in temperature. They are also very fast.
Due to their speed, they are used for precision temperature
control and any time very small temperature differences must be
detected.
They are made of ceramic semiconductor material (metal
oxides).
The change in thermistor resistance with temperature is very
non-linear.
Thermistor Non-Linearity
Resistance/Temperature Conversion




Standard thermistors curves are not provided as
much as with thermocouples or RTDs. You often
need a curve for a specific batch of thermistors.
No 4-wire bridge is required as with an RTD.
DAQ systems can handle the non-linear curve fit
easily.
Thermistors do not do well at high temperatures and
show instability with time (but for the best ones, this
instability is only a few millikelvin per year)
Infrared Thermometry





Infrared thermometers measure the amount of
radiation emitted by an object.
Peak magnitude is often in the infrared region.
Surface emissivity must be known. This can add a lot
of error.
Reflection from other objects can introduce error as
well.
Surface whose temp you’re measuring must fill the
field of view of your camera.
Benefits of Infrared Thermometry

Can be used for
– Moving objects
– Non-contact applications
where sensors would
affect results or be
difficult to insert or
conditions are hazardous
– Large distances
– Very high temperatures
Field of View

On some infrared thermometers, FOV is adjustable.
Emissivity




To back out temperature, surface emissivity must be
known.
You can look up emissivities, but it’s not easy to get
an accurate number, esp. if surface condition is
uncertain (for example, degree of oxidation).
Highly reflective surfaces introduce a lot of error.
Narrow-band spectral filtering results in a more
accurate emissivity value.
Ways to Determine Emissivity
1.
2.
3.
4.
5.
Measure the temperature with a thermocouple and an infrared
thermometer. Back out the emissivity. This method works well if
emissivity doesn’t change much with temperature or you’re not
dealing with a large temperature range.
For temperatures below 500°F, place an object covered with
masking tape (which has e=0.95) in the same atmosphere.
Both objects will be at the same temperature. Back out the
unknown emissivity of the surface.
Drill a long hole in the object. The hole acts like a blackbody
with e=1.0. Measure the temperature of the hole, and find the
surface emissivity that gives the same temperature.
Coat all or part of the surface with dull black paint which has
e=1.0.
For a standard material with known surface condition, look up
e.
Spectral Effects



Use a filter to eliminate longer-wavelength atmospheric radiation
(since your surface will often have a much higher temperature
than the atmosphere).
If you know the range of temperatures that you’ll be measuring,
you can filter out both smaller and larger wavelength radiation.
Filtering out small wavelengths eliminates the effects of flames
or other hot spots.
If you’re measuring through glass-type surfaces, make sure that
the glass is transparent for the wavelengths you care about.
Otherwise the temperature you read will be a sort of average of
your desired surface and glass temperatures.
Price and Accuracy


Prices range from $500 (for a cheap handheld) to
$6000 (for a highly accurate computer-controlled
model).
Accuracy is often in the 0.5-1% of full range.
Uncertainties of 10°F are common, but at
temperatures of several hundred degrees, this is
small.
Non-Electronic Temperature Gages




Crayons – You can buy crayons with specified melting
temperatures. Mark the surface, and when the mark melts, you
know the temperature at that time.
Lacquers – Special lacquers are available that change from dull
to glossy and transparent at a specified temperature. This is a
type of phase change.
Pellets – These change phase like crayons and lacquers but are
larger. If the heating time is long, oxidation may obscure crayon
marks. Pellets are also used as thermal fuses; they can be
placed so that when they melt, they release a circuit breaker.
Temperature sensitive labels – These are nice because you can
peel them off when finished and place them in a log book.
Non-Electronic Temperature Gages,
cont.


Liquid crystals – They change color with temperature.
If the calibration is know, color can be determined
very accurately using a digital camera and
appropriate image analysis software. This is used a
fair amount for research.
Naphthalene sublimation (to find h, not T)– Make
samples out of naphthalene and measure their mass
change over a specified time period. Use the heat
and mass transfer analogy to back out h.
Thin-Film Heat Flux Gauge


Temperature difference across a narrow gap of
known material is measured using a thermopile.
A thermopile is a group of thermocouples combined
in series to reduce uncertainty and measure a
temperature difference.
From Nicholas & White, Traceable Temperatures.
Thin-Film Heat Flux Gauge, cont.

Fig pg a-26
Thin-Film Heat Flux Gauge, cont.

Difficulties with these gauges
– The distance between the two sides is very small, so the
temperature difference is small. The uncertainty in the
temperature difference measurement can be large.
– Watch where you place them. If the effective conductivity of
the gauges is different than the conductivity of the material
surrounding it, it will be either easier or harder for heat to
pass through it. Heat will take the path of least resistance, so
if you don’t position the gauge carefully, you may not be
measuring the actual heat flux.
Temperature Controllers

Consider the following when choosing a controller
– Type of temperature sensor (thermocouples and RTDs are
common)
– Number and type of outputs required (for example, turn on a
heater, turn off a cooling system, sound an alarm)
– Type of control algarithm (on/off, proportional, PID)

On/off controllers
–
–
–
–
–
These are the simplest controllers.
On above a certain setpoint, and off below a certain setpoint
On/off differential used to prevent continuous cycling on and off.
This type of controller can’t be used for precise temperature control.
Often used for systems with a large thermal mass (where
temperatures take a long time to change) and for alarms.
Proportional controllers

Proportional controllers
– Power can be varied. For example, in a heating unit the
average power supplied will decrease the closer one gets to
the set point.
– Power is often varied by turning the controller on and off very
quickly rather than using a VFD
– Some proportional controllers use proportional analog
outputs where the output level is varied rather than turning
the controller on and off.
PID



Combines proportional with integral and derivative control.
With proportional control, the temperature usually stabilizes a
certain amount above or below the setpoint. This difference is
called offset.
With integral and derivative control, this offset is compensated
for so that you end up at the setpoint. This provides very
accurate temperature control, even for systems where the temp.
is changing rapidly.
How to Choose a Temperature Control
Device or System

Things to take into account
–
–
–
–
–
–
–
–
–
Standards
Cost
Accuracy
Stability over time (esp. for high temperatures)
Sensitivity
Size
Contact/non-contact
Temperature range
Fluid
International Standards

North America
– NEMA (National Electrical Manufacturers Association), UL
(Underwriters Laboratories), CSA (Canadian Standards
Association
Enclosure Ratings








Type 1 – general purpose indoor enclosure to prevent accidental
contact
Type 2 – indoor use, provides limited protection from dirt and
dripping water
Type 3 – outdoor use to protect against wind-blown dust, sleet,
rain, but no ice formation
Type 3R – outdoor use to protect against falling rain but no ice
formation
Type 4 – add splashing or hose-directed water to 3
Type 4x – add corrosion
Type 6 – add occasional submersion to 4x
etc.
Choice Between RTDs, Thermocouples,
Thermisters








Cost – thermocouples are cheapest by far, followed by RTDs
Accuracy – RTDs or thermisters
Sensitivity – thermisters
Speed - thermisters
Stability at high temperatures – not thermisters
Size – thermocouples and thermisters can be made quite small
Temperature range – thermocouples have the highest range,
followed by RTDs
Ruggedness – thermocouples are best if your system will be
taking a lot of abuse
Simplified Uncertainty Analysis for Lab 1

Random (precision) error
– For temperature measurements, this typically includes
fluctuations in the electronics of the data acquisition units
as well as fluctuations in the quantities measured

Bias (fixed) error
– For temperature measurements, this typically includes the
finite resolution of the A/D card (if one is used), the use of
a curve fit for the thermocouples, reading of calibration
thermometers, and conduction and radiation errors.

Total uncertainty is found using the root mean square of
these two errors
U  random error 2  bias error 2
Random Error

95% confidence interval – 95% of temperature readings
will fall in this range
– =+/- 2 standard deviations
– For your lab, during calibration, take at least 35 data points
(N=35) at one temperature. Then calculate the average and
standard deviation using the equations below.
– Excel can also be used.
1 N
T   Ti
N i 1
 1 N
2


ST  
T

T

i

N

1
i 1


1
2
Bias Error




Conduction and radiation errors should be negligible.
For our lab, we will do a simplified analysis.
Once you have a calibration curve fit, find the
deviation between the curve fit and each data point.
Use the magnitude of the maximum deviation as your
bias error.
In ME 120 you’ll learn a lot more about calculating
uncertainties!