Increasing RTD accuracy

Hey guys,
I am using some Platinum 385 RTDs and need accuracy down to 1/10th of a degree. It is my understanding, that RTDs have, for the most part, a linear response.

However to improve accuracy, the Callendar Van Dusen equation can be used to achieve high accuracy throughout the full range of the RTD. Normally the RTDs are wired into their own separate unit(Instrulab 4312a). In the past, to "calibrate" the RTDs, they would calculate the Callendar Van Dusen coefficients (2nd order quadratic equation) and input those into the unit. This would "calibrate" the RTD for the whole range.

Now we are trying to wire the RTDs into a Compactlogix (L32), with a 1769-IR6 card. To get the most resolution I have it set to Raw/Proportional units. I then scale it to get my temperature.

Does anyone have any experience in trying to use their own custom coefficients, instead of allowing the IR6 card to do its own thing?

OR does anyone have other recommendations to improve the accuracy?

Thanks guys.

The most accurate that I have used in the past, was purchase probes that was matched to the temperature I was looking for. We wanted 155 DegF, so we ordered them that way. They were hand picked. Also, 4 wire are the most accurate. I would start out there.

George

Check the manual for the 1769-IR6 module. I believe you'll find that the best accuracy you can expect from the module itself is around 0.5 degrees.

Anything can resolve 0.1 deg, it takes some doing to get that kind of accuracy [comparison to a traceable standard].

People who are serious about 0.1Deg accuracy use high end temperature transmitters and either very high end analog input modules, or digital communications (Profibus, Foundation Fieldbus, HART) so that temperature value is not degraded in subsequent D/A and A/D conversions.

Rosemount's 644 (Profibus PA) claims 0.15 deg C.


Siemens TH300 (analog 4-20ma) claims 0.1 deg C.

The big problem here is that even if you work with a transmitter or take the RTD into a very accurate temperature controller and extract the temperature by Modbus or similar, the combined error could be greater than your 0.1 degree - it also could be better.
Take the Rosemount unit @ + 0.15 and the RTD at + 0.1, then the total error could be 0.25. The combined error could also cancel out and you may have no error. This error could vary over the temperature range as well.
High end RTDs can be obtained with known, very small error but are extremely expensive and difficult to obtain. If my memory serves me correctly a company called Chino in Japan can supply these but you will need big pockets.

In addition to RTD accuracy (which does vary with measurement range)and transmitter accuracy, you can add inaccuracy of thermowell installation and location of RTD in the process. You are very optimistic about the measurement accuracy you will achieve in reality.

I'll throw another problem into the mix, self-heating error. Just running 100uA through an RTD can cause it's temperature to rise by 0.1 degrees.

I would go back to whoever wrote the specification and ask them to confirm the justification for it. The last time I chased a problem like this the specification was created by rectal extraction.

It's simply impossible with a 1769-IR6, which has at best a 0.5 degree accuracy.

Timbert said:

I'll throw another problem into the mix, self-heating error. Just running 100uA through an RTD can cause it's temperature to rise by 0.1 degrees.

I would go back to whoever wrote the specification and ask them to confirm the justification for it. The last time I chased a problem like this the specification was created by rectal extraction.

It's simply impossible with a 1769-IR6, which has at best a 0.5 degree accuracy.

Click to expand...

Hmmm Rectal Extraction

Is that associated with a recently recognized syndrome ie
Rectal Cranial Inversion?

Prolonged and untreated, it may develop into
Optio Rectal Malitis
aka Shixxy Outlook on Life

Dan Bentler

Thanks for all the responses guys.

One thing that strikes me, if the IR6 module is only capable of 0.5C accuracy which is what spec does say, why even bother making an RTD module? A thermocouple would be just as accurate. The attractiveness of RTDs is that they are more accurate than thermocouples. 0.5C is a LOT. If people are using RTDs in their systems, in most cases it is because they want greater accuracy that thermocouples cant provide.

I disagree, we use RTD's because they require less maintenance than thermocouples.

Pardon my ignorance Ken but why do you consider RTDs to require less maintenance? We have thermocouple systems in our plant that haven't been touched in years.

1) With regard to the OP's AI spec for RTD:
Here are some specs for RTD accuracy for two different devices:.
Note this one is at 0.9° to 1.8°C,

The 2nd device has a tighter spec at 0.5°C but only when the device is at reference temperature of 25°C:


The OP's stated spec is quite a decent spec, actually.

2) The same level of accuracy is not available from thermocouples.
Observe the thermocouple sensor error in the ANSI standards table (below) for commercial thermocouples. Note that the error is sensor error only, and does not take into account analog input error or cold junction compensation error.

Any thermocouple accuracy spec has to include the cold junction error:


The multiple errors for thermocouple generally takes it out of the accuracy categories that RTD's can address.

The spec for the OP device is in the "better" category of industrial AI’s.

I don’t think the expectation of easily achievable 0.1% temperature accuracy is realistic.

Dan

mutabi said:

Hey guys,
I am using some Platinum 385 RTDs and need accuracy down to 1/10th of a degree. It is my understanding, that RTDs have, for the most part, a linear response.

However to improve accuracy, the Callendar Van Dusen equation can be used to achieve high accuracy throughout the full range of the RTD. Normally the RTDs are wired into their own separate unit(Instrulab 4312a). In the past, to "calibrate" the RTDs, they would calculate the Callendar Van Dusen coefficients (2nd order quadratic equation) and input those into the unit. This would "calibrate" the RTD for the whole range.

Now we are trying to wire the RTDs into a Compactlogix (L32), with a 1769-IR6 card. To get the most resolution I have it set to Raw/Proportional units. I then scale it to get my temperature.

Does anyone have any experience in trying to use their own custom coefficients, instead of allowing the IR6 card to do its own thing?

OR does anyone have other recommendations to improve the accuracy?

Thanks guys.

Click to expand...

PTI

why the need for 0.1 degree accuracy, and I am assuming F ?

I do a lot in this precision.
RTD is more stable as thermocouple.
i can reach stable of 0.01 celsius with no problem.
The RTD has a certain precision like 0.02 at 0 celsius. depending of the class.
The input of your card is the problem.
use the data you have and use a offset to be precise.

If your RTD/IO card system is repeatable you can compensate for the input card issues.

Use an ice bath, boiling water, and a known high accuracy thermometer. (The thermometer is needed because boiling and melting points vary with barometric pressure.) Even better if you can use a medium close the temperature you will be measuring.

At any rate, if you can take the output of the temperature of the IO card and create a correction factor. This will probably be an offset to the reading to correct to the "actual" temperature. This will have to be done for each RTD and IO card, and if the RTD is changed in the future will have to be repeated.