Perils of Replacing Liquid in Glass Thermometers
Post on 01-Dec-2014
79 Views
Preview:
Transcript
Perils of Replacing Liquid-in-Glass Thermometers
Dean Ripple, Greg Strouse, Dawn Cross, and Karen Garrity
NIST Thermometry Group
Overview of thermometer types
Discussion of application issues
Detailed discussion of different thermometer types
Overview of Different Thermometer Types
Thermometer TypesThermometer Types
Standard Platinum Resistance Thermometers (SPRTs) (very accurate, but susceptible to shock)13.8 K to 962 °C
Industrial Platinum Resistance Thermometers (IPRTs)–196 °C to 850 °C
Thermistors–50 °C to 100 °C
Thermocouples–196 ° C to 2100 °C
Liquid-in-Glass Thermometers–150 °C to 400 °C
Digital Thermometers (PRT, thermistor, or thermocouple in disguise)–196 °C to 850 °C
NIST Thermometry Group Calibration Staff
Dean Ripple – Group Leader– dean.ripple@nist.gov, 301 975 4801
Karen Garrity – Thermocouples– karen.garrity@nist.gov, 301 975 4818
Dennis Minor – SPRTs for T > 83 K, Fixed-points, and Bridges– dennis.minor@nist.gov, 301 975 4821
Gregory Strouse – ITS-90 for T > 83 K, SPRTs, Fixed-points, and Bridges– gregory.strouse@nist.gov, 301 975 4803
Wes Tew – ITS-90 for T < 84 K, SPRTs– weston.tew@nist.gov, 301 975 4811
Dawn Cross – Industrial – thermistors, resistance, liquid-in glass– dawn.cross@nist.gov, 301 975 4822
www.nist.gov/thermometry
Considerations in Selecting a Thermometer
Accuracy: Uncertainties range from < 0.1 m°C to >1 °C
Cost of Thermometer: Range from $6 to $6000
Cost of Calibration: from $50 to $12,000
Temperature Range– thermistors 10 °C to 100 °C vs. W-Re thermocouples 0 °C to 2100 °C
Stability and Durability during use– chemical contamination– resistance to high temperatures, moisture, vibrations, and shock
Compatibility with measurement equipment– resistance thermometers, thermocouples easy to integrate to
electronics– liquid-in-glass, digital thermometers much easier for quick visual
inspectionCompatibility with object being measured
– sheath diameter, length chosen for good thermal equilibrium
Thermometer Types:Calibration Ranges and Uncertainties
0.01
0.1
1
10
100
1000
10000
-300 0 300 600 900 1200 1500 1800Temperature / °C
Exp
ande
d U
ncer
tain
ty /
m°C
Pt-Rh alloy thermocouples
Base metal thermocouples
Standard Platinum Resistance Thermometers(SPRTs)
Liquid-in-glass (LiG)
Industrial Platinum Resistance Thermometers (IPRTs), Thermistors
Au/Pt thermocouples
Comparative Thermometer Types:Calibration Methods, Uncertainties, and Costs
60
6 to 50
600
50 to 1000
100 to 1000
3000
3000
Nominal Cost$
1000<0.5Comparison–20 to 400LiG, Hg
10001Comparison0 to 1100Thermocouple,Base Metal
10000.3Comparison0 to 1100Thermocouple,Noble Metal
1200<0.005Comparison–10 to 100Thermistor
1750<0.01Comparison–196 to 500IPRT
1750<0.01Comparison–196 to 500SPRT
7000<0.001Fixed-Point–196 to 660SPRT
Calibration Cost, $
Calibration U (k=2), °C
Calibration Method
Temperature Range, °C
Thermometer Type
General Issues with Replacing Liquid-in-Glass (LiG) Thermometers
Tolerances vs. Calibration Uncertainties
0.001
0.01
0.1
1
10
-200 0 200 400 600Temperature / °C
Tole
ranc
e or
unc
erta
inty
/ °C
Thermistor, accuracy class 1
LiG, total immersion
LiG, partial immersion
PRT, grade A
TC, type K, special
Colored lines: ASTM tolerances (ASTM E1, E1137, E230, and E879).
Dashed lines: NIST calibration uncertainties (k=2)
Tolerances vs. Calibration Uncertainties
Tolerance band: manufacturer’s guarantee that the instrument response will conform to a standard response function to within an error equal to the tolerance.
Calibrated thermometer: may or may not have a response close to the nominal response function for that thermometer type.
Response of individual unit is reported, along with uncertainties of the calibration process.
Individually calibrated thermometers cannot be considered directly interchangeable, unless the readouts or software are adjusted toincorporate the individual response function.
Temperature Non-Uniformity
Total-Immersion Liquid-in-Glass Thermometer: Immersion depth varies with temperature
Partial-Immersion Liquid-in-Glass Thermometer: Immersion depth specified on thermometer
Platinum Resistance Thermometer: Placing thermometer at a fixed depth may introduce a bias, due to temperature variations in apparatus
LiG
PRT
2 °C
Bias of Liquid-in-Glass Thermometers
1. For a partial immersion thermometer, if the stem temperature during use differs significantly from the ASTM E 1 stem temperature specified in Table 4 of E 1 and a correction is not applied, there will be an error (see ASTM E 77).
2. Total-immersion thermometer is used at a fixed, partial immersion, with no correction applied. Extreme care must be taken in selecting an alternative thermometer for these applications, because use of a different thermometer type, while reducing the measurement error, may cause changes in the bias of the standard.
3. If the thermometer is not in good thermal contact with the body being measured, there may be significant errors due to thermal conduction along the thermometer sheath. Temperature reading biased even though the precision is acceptable.
Typical Measurement Uncertainty Budget: Thermocouples
Component Method of evaluation Calibration uncertainty or tolerance
Manufacturer or calibration laboratory, or ASTM E 230 tolerance
Thermocouple drift Results from literature, or in situ comparisons
Reference junction uncertainty
Manufacturer or independent evaluation
Readout uncertainty Manufacturer or independent evaluation
Readout drift Manufacturer or independent evaluation
Items in italics—examples of components generally not addressed with liquid-in-glass thermometers
Examples of Subtle Device/Readout Failures
Long-term drift of readouts is expected, and addressed by periodic recalibration, but there are other risks:
Device loses calibration values in memory & reverts to default coefficients
Resistance bridge balances correctly, but circuitry to compensate for cable resistance is faulty
Incorrect entry of calibration coefficients into readout
Probes switched without updating coefficients
Consequence: Greater need for measurement cross-checks / measurement assurance / check standards
Routine checks of performance
Checks at ice point
Comparison of readings of different thermometers
Specific Types of Thermometers: Liquid-in-Glass Thermometers
Types of Liquid-in-Glass (LiG) Thermometers
Types of ASTM LiG thermometers– Over 120 types– Total Immersion– Partial Immersion
Liquids used– Mercury (Hg) - ASTM– Organic– Mercury Thallium– Proprietary (non-toxic)
Total Immersion Partial ImmersionTop of the liquid column about 1 cm
above the top of the bath fluid.Immersed to an indicated line or
depth
Emergent Stem-TemperatureCorrection
Partial Immersion Thermometers above 150 °C
Faden Thermometers used to make emergent stem-temperature (EST) correction
Uncertainty of EST correction
– Faden temperature measurement
where k is the differential expansion coefficient of Hg
and n is the number of scale degrees equivalent to the length of emergent stem
)( spobs ttknEST
Tolerances & Calibration Uncertainties
Typical ASTM tolerances: 1 or 2 scale divisions.
Typical calibration uncertainties: 0.2 to 0.5 of a scale division
Additional uncertainties in use:• Emergent stem temperature different from ASTM specification (partials only)
• Drift due to changes in bulb
• Drift due to capillary distress at high temperatures (> 150 °C)
Recalibration of LiG Thermometers
• NIST SP 819 “A Procedure for the Effective Recalibration of Liquid-in-Glass Thermometers” recommends recalibration by adjusting old calibration values by the ice-point shift.
• Presently, we believe that LiG thermometers used above 150 °Cshould additionally be periodically recalibrated at the highest temperature of use.
• New NIST study in progress to revisit SP819 results
Temporary Changes in Bulb VolumeAfter use at high temperatures, there is a temporary enlargement of the bulb
This type of temporary bulb expansion will cause a depression in the reading at the ice-point.
It takes 72 hours at room temperature for a thermometer to recover from exposure to high temperatures.
Easiest solution is to use thermometer with increasing temperature.
Alternative solution is to apply a scale correction (contact Dawn Cross)
Permanent Changes in Bulb Volume
High temperature use sometimes causes a permanent increase in the ice-point reading due to annealing effects
Non-Mercury Liquid-in-Glass Thermometers
• Organic liquids generally have inferior performance to mercury, but are a reasonable alternative if uncertainty requirements aremodest (ASTM standard just begun)
• Beware of drainage of organic liquid down capillary wall on cooling
• “Next-generation” organic liquids under development (Existing ASTM standard E2251); good accuracy, but check for separation ofliquid column
• For all non-mercury LiG thermometers, capillary and bulb dimensions will be different, with different time response and immersion characteristics!!!
Choice of a LiG Thermometer
Advantages of LiG thermometers– Relatively inexpensive
– When used at moderate temperatures (<150 °C), recalibration at the Ice MP suffices
– Damage to thermometer is usually visually apparent (!!!)
Disadvantages of LiG thermometers– Very difficult to automate
– Total immersion require adjustment of immersion with changing temperature/Partial immersion not too accurate
– Mercury is banned in some circumstances; prohibitively expensive to clean up in other instances
Thermocouples
Thermocouples...any two dissimilar conductors, joined at one end
Although total signal depends on temperature of two ends (1 & 4), thermocouples generate signal primarily in regions of strong thermal gradients. (Region 2-3)
The junction itself does not generate a voltage!!
T0
T1
Material A
Material BVoltmeter
Advantages of Thermocouples
CheapWide temperature range (270 C to 2100 C) Small (down to 0.25 mm diameter) Easy to integrate into automated data systems
Disadvantages of Thermocouples
Small signals, limited temperature resolution (1 mK to 1 K) Thermocouple wires must extend from the measurement point to the
readout. Signal generated wherever wires pass through a thermalgradient.
At higher temperatures, thermocouples may undergo chemical and physical changes, leading to loss of calibration.
Recalibration of certain types of thermocouples or in certain applications is very difficult.
Emf-Temperature Relationshipsfor the 8 Letter-Designated Thermocouple Types
BT
E JK
NR
S
Temperature / °C
Em
f / m
V
Notation: E = emf = Electromotive Force = Thermoelectric VoltageS = dE/dT = Seebeck Coefficient = Sensitivity
Base-metal types
ASTM E230 Tolerances for Type K and Type N Thermocouples
-10
-5
0
5
10
-200 0 200 400 600 800 1000 1200 1400
Temperature, °C
Tole
ranc
e, °
C
Standard toleranceSpecial tolerance
Soft-Insulated Thermocouples
• Choose polymer insulation based on upper temperature limit • High-temperature: use single lengths of alumina, not woven ceramics
Inner polymer insulationOuter polymer insulation
Weld or solder junction by any method
Mineral-Insulated, Metal-Sheathed (MIMS) Thermocouples
• MIMS thermocouples are available in small diameters (0.25 mm)• Sheath protects thermoelements from contamination
Selecting a Thermocouple Type
type E: High Seebeck coefficient, homogeneous materials. Good for low temperatures.
type J: Cheap!type K: Fairly cheap high temperature thermocouple.type N: Good base metal thermocouple for high temperatures.type T: Homogeneous materials. Direct connection of differential
pairs to voltmeters.
Use type K, E, or T at room temp., type K up to 200 C, type N in the range 300 C to 600C, type N or K above 600 C
type R, S: Noble metal thermocouple for range 0 C to1400 C. type B: Noble metal thermocouple used from 800 C to 1700 C.
Use type R or S below 1300 C, type B above 1300 C.
Emf MeasurementsVoltmeter
• voltmeter: accurate, but no direct temperature display
• Analog-to-digital PC card: beware high noise, high voltage offsets
• dedicated readout: equivalent to voltmeter + reference junction compensation + software.
Reference Junction Compensation• Thermocouple tables assume 0 °C reference junction temperature• Reference junction compensators mimic “missing” section of
thermocouple when reference junction temperature is not at 0 °C• Emf addition by hardware or software
Copper wires
DVMIce Point
TC
TC connector & extension wire
Care and Feeding of Thermocouples
• Monitor drifts in base metal thermocouples by in situ tests• Protect from mechanical strain and kinks
• Protect from contamination using alumina or silica tubes, or usemineral-insulated-metal-sheathed thermocouples.
• For each temperature environment to be measured, a new thermocouple should be made, and it should always be used at thesame immersion.
• Obey the ASTM upper temperature limits for bare wire thermocouples.
Limitations on Thermocouple Performance
• Chemical contamination (potentially very large)• Physical strain (moderate)• Preferential oxidation, volatilization (potentially large)• Hysteresis in structural phase changes (small to moderate,
worst for type E and K)• Extension wires, thermocouple connectors, feedthroughs
(potentially large, but avoidable)
Rule of thumb for fractional uncertainty of t, in °C
Base-Metal Noble-MetalVery best possible 3x10-4 10-4
(t < 100 °C)
Very best possible 10-3 10-4
(t > 100 °C)
Typical 10-2 10-3
Recalibration of Used Thermocouples is Problematic!
If the furnace is isothermal, there will be NO difference between the original TC calibration and the recalibration.
Platinum Resistance Thermometers (PRTs)
What is an IPRT?2, 3, or 4-wire resistance element –nominally 100 @ 0 °C
– Wire wound– Thick film– Thin film
Resistance changes as a function of temperaturePositive temperature coefficient Nominal temperature range of use:
– –200 °C to 850 °CNominal resistance at 0 °C
– 100
Pictures courtesy of SDI and Minco
IPRT Wiring Configurations
2-Wire: Rmeas = V/i = i(RL + RL + RS)/i = 2RL + RS
possible error of (2 RL/RS)(1/4103)°C
• calibration will change if RL changes
V
R L
R L
R S
i
4-Wire: No current passes through voltage-sensing leads. Consequently, there is no extra voltage drop along these leads, and measured voltage is RS.
I+ I-
V+ V-to DMM
to DMM
to current source to current source
• compensates for changes in RL
• preferred method
2-wire for non-demanding applications (±5 °C)3-wire for ±1 °C measurements, or ±5 °C over long cables4-wire for all high-accuracy measurements
Nominal Resistance vs. Temperature Curve for an IPRT
0
100
200
300
400
Temperature, °C
Res
ista
nce,
IEC 751ASTM E1137
ASTM E1137 “Off the Shelf” Tolerance and Uncertainty
0.0
0.5
1.0
1.5
2.0
2.5
Temperature, °C
Tole
ranc
e, °C
Class AClass B
Calibrate individual units for better uncertainty
Types and Construction of IPRT elements
Wire-wound element– Alumina insulator: –200 °C to 850 °C– Glass insulator: –200 °C to 400 °C– Often loose or compacted MgO powder
for support– Sometimes hermetic via glass
compound– Best accuracy with loose powder
Thick and Thin Film element– Alumina substrate: –200 °C to 850 °C– Polyimide: –200 °C to 200 °C– Small with fast time response
Trade off between shock resistance and high accuracy
2-wire IPRT element
Pictures courtesy of SDI and Minco
Thick Film IPRT element
Construction of IPRT ProbeMIMS IPRT Probe
– –200 °C to 650 °C– 2, 3, or 4-wire device– Element is either
wire-wound or film– Specify length,
fittings, element type
Support of winding– Not strain free– Thin film is bendable
(strain sensor)
Sealing– Element is usually
hermetically sealed– MIMS cable may not
be sealed
MIMS
Pictures courtesy of Minco
Which IPRT Should I Use?
Probes vs. Bare Element: probes recommended unless element is permanently mounted
Film IPRTs: good time response, small size, shock resistant
Wire-wound IPRTs with constrained coils: low accuracy, but shock resistant
Wired wound IPRTs with slightly constrained coils: best accuracy (approaching ±0.01 °C over 400 °C span), sensitive to shock. Performance is highly variable with model.
Resistor configuration– 2-wire for non-demanding applications (±5 °C)– 3-wire for ±1 °C measurements, or ±5 °C over long cables– 4-wire for all high-accuracy measurements
Considerations in Selecting IPRTs
ADVANTAGES– Wide temperature range– R vs. T is well characterized– Rugged construction– Cost is less than an SPRT– Available in different shapes and sized to meet most application
requirements– Can be used with a digital temperature read-out device
DISADVANTAGES– Resistance element is not strain free - hysteresis– Deterioration at elevated temperatures– 2- and 3- wire devices need lead-wire compensation– Non-hermetically sealed IPRTs will deteriorate in environments
with excessive moisture– Not as accurate as an SPRT– Not a defining standard of the ITS-90
Sources of Errors When Using IPRTsStability
– Repeatability at 0 °C or 0.01 °CImmersion
– Check immersion characteristics of IPRT on insertion in thermal environment
Insulation Resistance– May degrade at t > 500 °C and high-moisture environments
Self Heating– Calibration and measurement current must be the same (nominally 1 mA)– Thermal contact with temperature of interest is important– Air probe or fluid probe will influence calibration method
Mechanical Shock and Vibration– Vibration or dropping the IPRT will cause the IPRT to drift or fail
Hysteresis– Measurement of temperature with increasing temperature may be
different than measurement of temperature with decreasing temperature
Immersion Test of an IPRT
Immersion Profile of IPRT on Insertion– Different designs will exhibit different immersion characteristics
5 10 15 20
Immersion depth in oil bath, cm
T, °
C
IPRT 1IPRT 2
Hysteresis Test of an IPRT
Causes of Hysteresis– Strain in platinum – most IPRTs are not strain free– Moisture produces shunting of sensor
-0.4
-0.2
0.0
0.2
0.4
0 100 200 300 400 500 600
Temperature, °C
T, °
CIncreasingDecreasing
Recalibration Interval for an IPRT
Widely varies by designWidely varying performance based on use
– Thermal history– Mechanical shock
Behavior is not as predictable as an SPRTDrift at 0 °C or 0.01 °C does not always correlate well at other
temperatures
Recommendation:– Measurement at 0 °C or 0.01 °C as a minimum– Measurement at highest temperature of use is better
Thermistors
Thermistors (Thermal Resistor)
Semiconductors of ceramic material made by sintering mixtures of metallic oxides such as manganese, nickel, cobalt, copper, iron and uranium.
Temperature Range: –50 °C to 100 °C
Standard Forms:bead 300 to 100 probe bead in glass roddisc 0.5 cm to 1.3 cm thick, 5 k to 10 kwasher 2 cm diameterrod moderate power capacity, 1 k to 150 k
NTC: Negative Temperature Coefficient - The vast majority of commercial thermistors used as thermometers are in the NTC category.
Anatomy of a Thermistor
Pictures courtesy of Thermometrics
Temperature, °C
R(t
) / R
(25
°C),
k
IPRTThermistor
Sensitivity Comparison between Thermistors and IPRTs
Thermistor Sensitivity– Resistance changes by
more than a factor of 300 from –50 °C to 90 °C
– Higher sensitivity than an IPRT
– Non-linear• –50 °C: 5000 / °C• 90 °C: 8 / °C
Interchangeability of a Thermistor
Thermistors are interchangeable to within the manufacturer’s tolerance band
Practical limits from –80 °C to 105 °C– ±0.05 °C for a 50 °C span– ±0.10 °C for a 75 °C span– ±0.20 °C for a 100 °C span
For a greater accuracy than the tolerance band, then a calibration must be performed– 0.001 °C for a 15 °C span– 0.003 °C for a 70 °C span
Advantages and Disadvantages of Thermistors
ADVANTAGESEasy to miniaturizeRuggedFast response timeEasy to useDigital thermometer readoutInexpensiveHigh sensitivitySmall-size beads may be used for point-sensingStability: 4000 h at 100 °C
bead-in-glass 0.003 °C to 0.02 °Cdisc 0.01 °C to 0.02 °C
DISADVANTAGESSmall temperature rangeNon-linear deviceNeeds frequent checks on calibration when exposed to t > 100 °CInterchangeability is limited unless the thermistors are matchedSelf-heating may be large
Cautions for Using Thermistors
Self-heating error may be large– Use low current and match calibration and use current– Calibrate air probe in air-filled glass tube
Epoxy coated thermistors are susceptible to degradation from moisture
Glass sealed thermistors are susceptible to degradation if the lead wires are bent apart at the base of the glass– Micro cracks can form in the glass
Heat sink the leads when making electrical connections with solder
Do not expose thermistors above 100 °C– Elevated temperatures will excessively age the thermistor and
cause drift
Digital Thermometers
Digital ThermometersWhat is a digital thermometer?
– An electronic measurement box that converts either resistance or emf of a thermometer to temperature
Pictures courtesy of Agilent, ASL, Brookstone, Hart Scientific, and Omega Engineering
Digital Thermometers
Easy to use in conjunction with a resistance thermometer
Device displays temperature directly by using the calibration coefficients of the thermometer
Different temperature scales available
Uncertainty: 0.001 °C to 1 °C
Resolution: 0.0001 °C to 1 °C
Device may allow two thermometers to connected directly to unit for differential thermometry
Some have software that allow “real time” calibration
SMART thermometers have electronics in the thermometer head for an RS232 (or similar) port connection
Caveats
Careful use requires careful reading: look at manufacturer’s specifications & know the probe type and limitations
“Digital Thermometer” only means that you can watch the numbers on a display
Calibration can be as a system or as readout + separate probeSystem calibration is cheaper and simpler
Probe and readout calibration allows easier repair if one element fails, allows identification of source of drift
Conclusions & Roadmap
1. Identify the level of uncertainty needed2. Identify the temperature range3. Identify unique aspects of the test apparatus or method (e.g.,
inherent temperature non-uniformity)
4. Identify adequacy of presently specified mercury thermometer (anywhere from overkill to just adequate)
Make judgments on how tightly to prescribe the thermometerwhether to require calibration, measurement assurancewhat tests/round robins are needed to validate the
revised standard
top related