Calibration checks on temp transmitters

There are charts available for Pt100 RTDs, so no assumptions are necessary.

The word guarantee is a commercial term, not a technical one.

A two-point check confirms that the whole system is set up properly. An injection test only confirms that the transmitter is working correctly.

All systems operate with a tolerance.

I understand about the tables. I've been using them since I was an apprentice.

People tend to assumed that a two point check verifies that the RTD or Thermocouple will work correctly throughout it' operating range due to it's linear properties.

Have you ever found an RTD or TC to be non linear?

Hello Tim/Sparkytip,

There are many types of thermocoules. Whch type do you employ? e.g. Type B°C/E°C/J°C/K°C.? Are you utilising Tabulated values of Thermoelectric Voltage in millivolts? Are you Calibrating or only "Verifying"? via DC low freq cal.What is your component/Product tolerance parameter?

So many questions when I should be giving you answers, just need know a few things firstly.

Best Regards

Metman

My question is a generalised one.

These checks are to verify the exisitng calbration of RTD temperature transmitters used for custody metering.

I've been working as an IE tech for 15 years so I have had exposeure to differnt types of TC's, RTD and tables etc. I'm interested to see what other techs out there do for this kind of check.

I know most people do a 2 point check, then use a suitable type of calibrator such as a fluke 725 multi function meter to test the calibration of the transmitter. Can you assume that a two point check with the RTD connected is enough to be sure the RTD will perform correctly for it's whole range.

I've spoken to other people who do this sort of custod metering calibration checks and they do a 2 point check and then full range check of the transmitter. Is this good enough?

I've always done a complete test with a dry block calibrator such as a JOFRA to test the RTD and transmitter as a whole. The way I think of it is that to be sure to be sure then this is the way to do it, but man is it slow. It takes time for the dry block to ramp up and ramp down. I'm looking to see if other people do it this way.

IMHO you should be doing a two point check. Ice water and boiling water. These tests are not quite as simple as some believe. You don't just dump an ice cube in a glass of water and assume it reaches 0 deg.C. Nether do you just boil water and assume it reaches 100 deg. C. Neither thermocouples nor PT thermometers are linear. If they were, you wouldn't need tables to accurately measure temperature. My suggestion is that you contact the people at NIST fot a recommendation.

If your instruments are critical you should be using traceable thermocouples from an NIST lab that have calibration tables for any offsets. Also spare sets to run side by side comparison checks.

Your calibration can only be as good as your temperature calibrator accuracy if this gives identical readings on identical transmitters it is a pretty good indication of correct function all things being equal.

RTDs and thermocouples, by themselves, are NOT linear devices, they both have deviations from a straight line over their temperature ranges. Usually some form of active circuitry is used to compensate for the non-linearity or a chart is produced giving the deviations in resistance with temperature from nominal. If you are using a relatively narrow temperature range, the linearity may not be of significance. Assuming that these sensors are not abused during their normal use, their characteristics are quite stable and once they have been 'certified', usually three points need be checked, near their minimum, maximum and 25'C temperature. Drift from these points will indicate a change in absolute value and further checks to verify stability of the device.

If your application is of a relatively wide temperature range and tighter tolerance, five temperature points from minimum to maximum, evenly spaced, over the range should be sufficient to verify accuracy.

For example, a properly compensated RTD circuit can produce a 0'C-400'C range with an accuracy of +/-0.05'C, with calibration points of 0'C, 100'C and 400'C.

Thermistors have a larger non-linearity curve than RTDs but are nearly as stable given proper use, similar compensation circuitry can produce very similar results to the RTD's.

An accepted rule of thumb for calibration is that the calibration source should be 10 times as accurate as the calibrated device.Preferably a NIST traceable calibration standard.

What are your acceptable tolerances?

What is the temperature range?

What is the transmitter specified accuracy?

Is this a 2 wire,3wire, or 4 wire transmitter?

RTD transmitters use a wheatstone bridge and op amp,with the leads in opposite legs, and the op amp input across the diagonal.What is the accuracy of the op amp?

The display accuracy? If digital display, what about the A D converter?Flash,Successive approximation,single slope,dual slope conversion?If analog display, how about parralax error?

The questions can go on ad nauseum, but it boils down to the acceptable error for the application.Critical is not a precise term.Be more specific.

EZ Street

I assume that you are talking about the calibration of only Temperature TRANSMITTER and not the sensor part of it.

As you said there could be two types of sensors: RTD type (PT100 /PT 1000 / CU 1000....) and Thermocouples of various types.

First type of sensors is resistance. Thus the transmitter using RTD accept resistance as input. Now, you have tables for each type of RTD. To calibrate the transmitter, just connect known resistances (through calibrated Decade box) and calibrate the indicator. You may calibrate it at as many points as you wish. Obviously as any RTD is also not really / absolutely linear, you may not get liner indication also, unless your indicator is linearized. Or you have to define the accuracy band and calibrate the indicator with mean straight line.

In case of thermocouples, the output is mV, thus the transmitter will be mV input type. Always outputs of any type of thermocouple is non linear. Thus obviously you have to choose the range to indicate. Bigger the range, more the problem of linearizing. Even for smaller range, the indicators has to be linearized for the range. Now, using standard mV source input the mV to indicator as per tables of mV Vs Temperatures and calibrate the indicated value. This has to be done for many points because of non linear nature of output of thermocouple. Two point calibration is not sufficient.

Regarding ice and boiling water, please remember that these are never 0 and 100 deg C as per standard values. The temperature depends upon the purity of water as well as atmospheric pressure. Thus for same location these temperatures vary time to time.

Temperature transmitters can be linear to temperature or to the source (resistance for RTDs, millivolts for TC), so first you must know your transmitter's specifications.

Routine calibrations should be at the minimum input range and the maximum input range with a check on the mid-point, whether by temperature or source (Ohms or mV). The frequency of calibration varies based on the transmitter and components used in manufacturing. Analog transmitters that are subject to a wide range of ambient temperatures will require more frequent calibrations. Many of the mircoprocessor based devices that are prolific in today's market can provide accurate service for years with out "calibration".

Calibrate the readout device with a DC voltage calibrator per the proper TC table. Then align the TC to the readout at ice point. Then check the TC at one other point (usually boiling point of water). If all are good, then it can be assumed with a very high degree of confidence that the system is properly measuring temperature.