Author Topic: A temperature disconnect - understanding TC & RTD calibration specs  (Read 1396 times)

0 Members and 1 Guest are viewing this topic.

Offline binary01

  • Contributor
  • Posts: 15
  • Country: au
Hi all,
Love reading this section of the forum, however it often highlights flaws in my knowledge, so I always feel the need to apologise ahead of posting!  I am a metrologist, but with a mechanical background focussed on mass and pressure, so new and trying to learn in the area of electrical calibration/metrology.
Coronavirus shift work in the lab has given me time to work on some hopeful improvements from an existing very basic DC Volts and current capability.  At the moment I am trying to determine the best path forward and ensure any equipment purchases will meet the needs for the types of instruments we see for calibration.  I currently have a Fluke 7526A calibrator (basic process calibrator), Fluke 8588A and Keithley 2000 DMMs.  The DMMs help us to squeeze performance out of the 7526A, but it remains a limited function calibrator, so I am exploring options for its replacement.

One area where I cannot properly reconcile the accuracy specifications for these devices, and an area I need to achieve low uncertainty, is calibration of temperature scales by electrical simulation.  The disconnect might be with the specifications themselves, but more likely it is a disconnect in my own understanding, so I am hoping some of you might be able to enlighten me  :D

For example, RTD PT100-385 range calibration is, in my understanding, fundamentally a resistance calibration of odd values up to 400 Ohms.  Although I feel more confident with physical resistors, such as a good quality decade resistor, I know that many calibrators are capable of resistance simulation indended for these types of calibrations.  When I compare their specification in resistance (Ohms) with their specifications in temperature simulation (degC via Ohms), the specifications are drastically different. 

Consider the Fluke 5522A:

[attachimg=1]
(Green = resistance  Blue = temperature)

I have listed some temperature values and their corresponding ideal resistance value.  If I was to test these values according to the specifications from the Fluke 5522A - RTD Simulation, I would expect performance of the middle two colums.  If instead I was test test these values using the Fluke 5522A - Resistance Simulation function (final two columns), I would get much better performance according to the Fluke specification. Compare the two blue columns and I would get almost an order of magnitude better temperature values using the resistance range directly.  Why is this the case? 
If it is purely a resistance calibration, scaled in a different unit, why is there a disconnect between the Fluke RTD simulation specification and the Fluke resistance simulation specification?

The same occurs when I consider K-type TC ranges, fundamentally a mV calibration scaled in degC (if I ignore reference junction implications for this purpose).

[attachimg=2]
(Red = mV  Blue = temperature)

Again in this case, the resultant temperature accuracy is almost an order of magnitude better if I use the mV range compared to using the TC-K function.

If someone can shed some light on why this is the case, or where I am wrong in my assessment, I would be greatly appreciative.

More generally also -
I have been considering a few options for calibrators, including Fluke 5730A ($$$!), 5522A and Transmille 3010A.  I can't find many opinions on Transmille products (hardware or ProCal software) around here, or generally online.  I know Fluke is dominant in the calibrator rhelm, and making progress in the reference DMM space, but if anyone has any knowledge or opinions on the Transmille 3010A, please let me know. 
I originally intended on seeking out a used 5700A or 5720A on evilbay, but I have read quite a number of repair/rebuild threads on these forums and elsewhere. I grew concerned that these are aging products with an unknown history, and may suffer from degraded reliability now which is not great for a commercial lab - is this a fair assessment of the 5700A/5720A option for someone that is not experienced in repairs of high performance electronics?

Long thread - sorry.  If you have read this far, thank you for taking the time.
« Last Edit: April 16, 2020, 10:42:43 pm by binary01 »
 

Offline splin

  • Frequent Contributor
  • **
  • Posts: 999
  • Country: gb
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #1 on: April 17, 2020, 02:45:00 am »
The apparent mismatch between the thermocouple and the voltage source accuracy specifications is likely due to the uncertainty of the temperature measurement of the internal isothermal block for cold junction compensation. Of course they couldn't be a**ed to specify that unlike for most < $300 thermocouple calibrators.

After a superficial look (so yes I probably missed something important!), the user manual seems to be pretty atrocious for a $42000 very expensive instrument wrt to thermocouple calibration/measurement. For example:

Quote
REF SRC (Reference Source) Selects intrnl (Internal) or extrnl (External) temperature reference source. Select intrnl when the selected thermocouple has alloy wires and you are using the isothermal block internal to the Calibrator.Select extrnl when using an external isothermal block, and when the selected thermocouple has copper wires. Press the REF softkey to enter the value of the external temperature reference. The best accuracy is obtained when you use extrnl and the external isothermal block is maintained at 0 °C.

Okay, so are the specs for using the internal or external reference? What if your thermocouple has alloy wires and you are using the external reference OR you have copper wires and you are using the internal reference? Perhaps the technical writer(s) got their 'and's and 'or's  mixed up? :-//

As to the RTD specifications I have no idea - it makes no sense to me. I expect the temperature calibration features were marketing tick list items and not taken seriously.

[EDIT] I don't know how much these actually cost - I thought I saw $42000 but looking again it was an ebay listing for a refurbished item at £43,700!
« Last Edit: April 17, 2020, 02:56:25 am by splin »
 

Offline beanflying

  • Super Contributor
  • ***
  • Posts: 5160
  • Country: au
  • Toys so very many Toys.
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #2 on: April 17, 2020, 03:05:45 am »
Have a really good read of the Omega site https://au.omega.com/technical-learning/ for a starting point.

Generally K type T/C's are not all that accurate in absolute terms so they make a poor choice for Calibration where they win is obviously a high range of operating conditions.

Also watch out for different Temp Co's of RTD's the Fluke I got below for example is different to what my Keysight 34461A speaks  |O so I have to do a conversion on it.

Some of the gear I have assembled over the last year and avoiding a Triple Point Cell 'for now'  :palm:

https://www.eevblog.com/forum/testgear/kaye-140-4-ice-point-reference/msg2412237/#msg2412237
https://www.eevblog.com/forum/testgear/sika-18600e-dryblock-temperature-calibrator-teardown-pictures-testing/msg2249781/#msg2249781
and a Fluke 5616 Secondary Reference PRT from a member in Germany.
Coffee, Food, R/C and electronics nerd in no particular order. Also CNC wannabe, 3D printer and Laser Cutter Junkie and just don't mention my TEA addiction....
 

Offline binary01

  • Contributor
  • Posts: 15
  • Country: au
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #3 on: April 17, 2020, 04:23:25 am »
Thanks splin - I'm at least glad that you agree that the specs might be unnecessarily broad for the TC and RTD ranges compared to the underlying electrical quantities.  Electrical calibration references seem to be so heavy on accuracy specificaton detail in other areas, but they are very lean in this area...
I had considered possibilities:
- Perhaps they use a secondary circuit for both temperature functions, which have much less accuracy than the primary voltage/resistance circuits?
- Perhaps they didn't want to overstate the accuracy, in case people were not allowing for the correct reference junction - but why not just provide more detail to the user about the best possible outcome if these are considered correctly?

In any case, at this stage I intend to simply develop calibration procedures using the primary quantities of mV and resistance, and perform conversion to the temperature quantities for the expected readings in software.  Is this the approach you and/or others are taking?

Hi beanflying - thanks for the link.  I agree with your assessment of TCs and I don't use them directly for measurements in my lab, instead I use 4w RTDs logged using PICO RTD loggers, which work well and have proven stable.  I have them calibrated as a complete system to avoid any translation errors :) 
However in this case I have customers requiring calibration of the TC (mV) ranges on multifuction calibrators using electrical simulation, which happen to have very tight manufacurer's specifications, meaning that I need to consider the method of calibration carefully to get a reasonable TUR.  Using the 5522A calibrator PRT and TC specs directly would be inadequate.  So it is not so much about the direct use of TCs, but how to suitably calibrate the TC(mV) ranges on the multifuction calibrators.  I will need to verify them using their own internal cold reference junctions and also using an external cold reference junction, which I am still developing.  For this reason I actually read your post on the Kaye device a few days ago with interest.  Having a device like that would certainly be better than making up an ice slurry for every calibration!  How you have found the Kaye device over the past year?  There is one going on ebay AU at the moment, although the asking pricing is not throw-away money and it looks to be reading low (-0.51 degC) according to the pictures.  The Kaye manual suggests that low temperature is a sign of faulty/damaged cell, so I wasn't sure...




 

Offline beanflying

  • Super Contributor
  • ***
  • Posts: 5160
  • Country: au
  • Toys so very many Toys.
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #4 on: April 17, 2020, 04:35:08 am »
The Kaye I got was a 110V one I took a complete punt on if it was a working one or not. Fortunately it worked out well running on a Transformer I already had. The Adjustment on mine is working well but as shown in the thread it is fairly mechanical in nature relying on the expansion/ice to drive the microswitch.

That Aussie one looks ok too but consider trawling ebay.com and paying the freight first and team it with a stepdown transformer. Keep an eye out for the Omega version too (more common) eBay auction: #362619869604 or there is a Kaye one here eBay auction: #232590205227
Coffee, Food, R/C and electronics nerd in no particular order. Also CNC wannabe, 3D printer and Laser Cutter Junkie and just don't mention my TEA addiction....
 

Offline mzzj

  • Super Contributor
  • ***
  • Posts: 1011
  • Country: fi
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #5 on: April 17, 2020, 09:08:40 am »
RTD specification appears to be 1- year specification, 4 or 2-wire measuremt.
Resistance specification is with 7-day  since zero call and 4-wire only
 

Offline binary01

  • Contributor
  • Posts: 15
  • Country: au
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #6 on: April 17, 2020, 09:27:29 am »
[EDIT] I don't know how much these actually cost - I thought I saw $42000 but looking again it was an ebay listing for a refurbished item at £43,700!
Haha, I think your original guess was closer to the mark (a bit over perhaps, if USD).  I'm not sure the refurbished ebay unit at £43,700 (USD $55k) will sell in a hurry!  I've seen lots of used 5700As for USD $15k which are sometimes tempting, but I might regret it later...

That Aussie one looks ok too but consider trawling ebay.com and paying the freight first and team it with a stepdown transformer. Keep an eye out for the Omega version too (more common) eBay auction: #362619869604 or there is a Kaye one here eBay auction: #232590205227
Thank you for thos suggestions.  There are quite a lot more to choose from on ebay.com compared to ebay.com.au + worldwide filter... The Omega and Kaye units look to use the same ice expansion mechanism, so I guess they would be equivalent? I assume your stepdown transformer is still 50hz without any issues?

RTD specification appears to be 1- year specification, 4 or 2-wire measuremt.
Resistance specification is with 7-day  since zero call and 4-wire only
Good point mzzj - I was considering 12 month ppm spec + 12 hour-since-zero floor term for the 5522A resistance, whereas the RTD range doesn't seem to rely on zeroing the resistance range prior or 4W, so that would be one source for the difference.
« Last Edit: April 17, 2020, 09:33:10 am by binary01 »
 

Offline beanflying

  • Super Contributor
  • ***
  • Posts: 5160
  • Country: au
  • Toys so very many Toys.
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #7 on: April 17, 2020, 09:37:27 am »
I haven't noticed any issues with the 50Hz operation down to the noise level of my 34461A so either should be fine. I got the transformer for an Advantest bench DMM which also remains low noise but for some bench gear it can be an issue with some HP Gear I know came with different versions.
« Last Edit: April 19, 2020, 07:25:05 am by beanflying »
Coffee, Food, R/C and electronics nerd in no particular order. Also CNC wannabe, 3D printer and Laser Cutter Junkie and just don't mention my TEA addiction....
 
The following users thanked this post: binary01

Offline splin

  • Frequent Contributor
  • **
  • Posts: 999
  • Country: gb
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #8 on: April 18, 2020, 11:41:11 pm »
This is a fluke paper about the uncertainties involved with thermocouple simulation and the calibration of a 5520A's isothermal CJC block:

http://download.flukecal.com/pub/literature/9010016_a_w.pdf

It is interesting but doesn't really answers any of your questions. The relevant DCV, resistance and temperature specs for the 5520A appear to be identical to your 5522A. About the only useful data point given is the calculated uncertainty, 0.038C, of calibrating the isothermal block. Of course that doesn't tell you if all 5520As are calibrated to that uncertainty and more importantly it doesn't tell you anything about the maximum 90 day and 1 year drift included in the specs.

One other point that struck me about the 5522A thermocouple specs is the amount that the uncertainty varies with temperature, which is far more than the DCV calibration output uncertainty varies. This obviously cannot be influenced by the CJC  temperature accuracy which is independant of the thermocouple temperature. About the only possibility I can think of are errors due to interpolation errors of the thermcouple table values as mentioned in the above link but that seems a far stretch - I would expect any such errors to be virtually negligable.

One other possibility that would apply to both the RTD and thermcouple specs is that the specs don't differentiate between simulated and measured temperatures. The 5522A is principally designed to generate calibration signals rather than measure them, so the poor specifications could be dominated by relatively poor DCV measurement specs rather than the calibration source voltage. This again seems to be unlikely though given the relative ease with which the instrument could self-calibrate it's internal voltmeter (and it may use a high precision voltmeter for calibrating its outputs anyway). If the volmeter is only required for the temperature measurements then it's just possible that they simply decicded not to include a more accurate meter on cost grounds.

I would suggest it would be a good idea to address these questions to Fluke - including why the manual and specs are so vague wrt temperature accuracy (eg. wrt the varying sensor connection variations (2,3 or 4 wire, thermocouple leads or copper wires) and using internal v external CJC. If they do provide any insightful answers please update us here so we can learn from them.
« Last Edit: April 19, 2020, 03:37:27 pm by splin »
 
The following users thanked this post: binary01

Offline binary01

  • Contributor
  • Posts: 15
  • Country: au
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #9 on: April 19, 2020, 11:13:10 am »
Many thanks for doing some more reading about this splin.  I read that Fluke paper yesterday and did find it quite interesting.  It did highlight some complications associated with this type of calibration, especially when using the CJC in either the simulating or measuring device (I will need to calibrate both measure and simulate functions on the UUT).
I guess the case they discusss where both devices are connected using copper hook-up wire and both interal CJCs turned off is the purest electrical calibration, effectively simulating a perferct 0 degC external RJ. In this mode it might offer the types of accuricies I have calculated based on the mV accuracy conversion.  Using this pure copper mode also negates the need for an a physical ice-point external RJ (bonus!).
I would still need to do a verification of the UUT internal CJC at some stage, so imagine I will need to use proper TC hook-up wire and produce an external RJ at some stage in the calibration, and this will add some uncertainty components relating to the quality of my RJ.  However, given how thorough Fluke and others are in specifications of other functions, I would have imagined that would still state the best possible accuracies based on the pure copper case or an external RJ, and list added components when using of the internal CJC.
Similarly, if the 5522A TC measurement capability is a limitation (a function I don't need), then I'd hope they would list the measure and simulate accuracy specs seperately.

This specification disconnect is not specific to the Fluke, as have done the same exercise with the Transmille 3010A specifications, and the situation is the same.  So I must be missing something!
Given that there is still some mystery after posting amongst the experts here, I might see if Fluke and/or Transmille are willing to shed any light on it, and I'll be sure to feed back if I hear anything.
 

Offline splin

  • Frequent Contributor
  • **
  • Posts: 999
  • Country: gb
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #10 on: April 19, 2020, 04:31:15 pm »
It just occurred to me that some, if not all, of the additional thermocouple error comes from this, from that Fluke "The Calibration and Use of Thermocouple Simulators" paper:

[attachimg=1]

[attachimg=2]

The noise, gain and drift errors of those x 1/10 and x10 amps will directly impact the accuracy. The gain setting resistors of the LTC1050 amps are likely those custom thin film resistors standing up at the back of the photo. The gain error will obviously increase the error at the higher and lower temperature extremes which is consistant with the specs. However, assuming the thin film 1:10 gain setting resistors have 2ppm/C ratio TC (10ppm over the +/-5C spec) and say 20ppm/year ratio drift, that is still less than .1C error at 1320C according to my calculations which is still a long way from the 0.4C spec.

I wonder if they provided switching on that PCB to allow those amps to be self-calibrated? The block diagram doesn't show it but the ADG511 has 4 switches so it could (but only two have obvious trace connnections on the PCB). It's quite possible that the internal architecture doesn't support self calibration given that the temperature calibration function seems to have been bodged into an earlier design:

Quote
The Fluke 5520A uses entries in degrees Celsius or Fahrenheit to direct the output to the TC terminals. If the instrument is asked to output a voltage, the output will be directed to the Normal terminals. Since the only way to get voltages to the TC terminals is to enter a temperature, a thermocouple type, 10 μV/degC, was created. This conversion factor was chosen because it is the same order of magnitude of the change in voltage vs. temperature for a number of the more popular thermocouples. So, calibrating the full range of the thermocouple terminals from zero volts to full scale of ±300 mV in the 10 μV/degC mode, results in displayed temperatures from -30,000 degC and plus 30,000 degC. On calibration certificates, reported values of temperatures exceeding those of the surface of the sun and those far below absolute zero have sometimes raised eyebrows, but all can be made in a calibration lab in one’s shirtsleeves. Just recognize it is a DC Voltage calibration of simulated temperature with a sensitivity coefficient of 10 μV/degC. It was a designer’s quirk to make entry simple and consistent, though not always intuitive.

In other words, the software wasn't designed to support temperature calibration/measurement so this was the easiest hack design modification to the UI to meeet the new marketing feature requirement.

Something similar might apply to the RTD side, though the existing resistance calibration function has enough resolution to provide the .003C resolution without modification, unlike the thermocouple output which needed the x 1/10 amp to provide .01C resolution. I aslo note that they don't specify the RTD measuring current. It's probably 1mA but a surprising omission for such an expensive piece of kit.
 
The following users thanked this post: binary01

Offline binary01

  • Contributor
  • Posts: 15
  • Country: au
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #11 on: April 19, 2020, 10:12:52 pm »
Aaah, I think you might have it.  So it is indeed a different circuit added in for the TC function, and perhaps rather than tightly controlling the amplifier gain/drift, they just allow a big bucket in the accuracy specification.
In fairness to Fluke (and Transmille if the same case), the specifications are probably still perfectly adequate for many TC calibrations - it's just that some of the instruments I see for calibration have very tight accuracy specifications on these ranges, which makes our job difficult!

A similar trick might be being done for the RTD side, and I imagine RTD circuits would typically be very low current to avoid heating, which may be harder to simulate with the same accruacy as "normal" resistance simulation?

Increasingly the argument for the TC calibration is to use the mV ranges and handle an RJ externally (or not at all for the pure copper mode).
For the RTD ranges, I may need to experiment.  If the 552A RTD range limitation due to low current required for RTD calibration, I may end up back with my good quality decade resistors to simulate the RTDs instead.
« Last Edit: April 19, 2020, 10:18:20 pm by binary01 »
 

Offline thermistor-guy

  • Regular Contributor
  • *
  • Posts: 230
  • Country: au
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #12 on: April 20, 2020, 12:23:56 am »
... an area I need to achieve low uncertainty, is calibration of temperature scales by electrical simulation. ...

I take it you are calibrating readouts (meters) only? You are not (a) calibrating temperature sensors, or (b) readouts + sensors as a system?

For (a) and (b), you will need a metrology-type temperature well, perhaps more than one, depending on the temperature ranges of interest.

Do you plan to stick to electrical simulation only?
 

Offline binary01

  • Contributor
  • Posts: 15
  • Country: au
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #13 on: April 20, 2020, 01:38:46 am »
Do you plan to stick to electrical simulation only?
Yes, we are only looking at verifying the measure and simulation functions on the instruments electrically.  In most cases the customers do not even provide a TC or RTD with the instruments, but they are multifunction devices and they want all functions calibrated.  In the case where they provide a probe and want it calibrated fully (as a system), we would send out to an accredited temperature lab intead.
 

Offline splin

  • Frequent Contributor
  • **
  • Posts: 999
  • Country: gb
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #14 on: April 20, 2020, 10:25:09 pm »
I was looking at the 552A service manual and its supplement but could find no reference at all to calibrating the RTD functions! The only mention of RTD is in the specification section - it's as if they took the 5520A manual and updated it for the 5522A but simply forgot to add in the RTD calibration - priceless!

The thermocouple calibration procedure calls for an ASTMS 56C thermometer which is .1C accurate with .02C divisions. It also requires a J type thermocouple supplied as part of a Fluke test lead set. That means the cold junction reference calibration unceertainty will be at least 0.1C plus the thermouple error which is not stated. It doesn't even specify how often the thermocouple should be re-calibrated, but perhaps this is documented in the test lead kit?

One other small additional uncertainty will be due to the voltage source noise specified at 1uV, 0.1 to 10Hz, for the 329.99mV range. There is no spec for noise in the resistance calibrator which must be included in the floor spec.

As to the RTD, I think it would be a good idea to ask Fluke - especially as to why isn't the RTD calibration procedure documented in the service manual.
 

Offline binary01

  • Contributor
  • Posts: 15
  • Country: au
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #15 on: April 21, 2020, 02:49:29 am »
The feedback from my local Fluke Cal agent is that the TC spec is broad due to internal RJ uncertainty and trade-offs to make the common component suit all types of TCs.
I have also seen that the 5522A calibration is performed by Fluke over the range -30000 to 30000 degC using the 10uV/degC fictional TC, and a single point calibration of the output at 23 degC with J-type TC.  This relates well to the information you have found in your last two posts.  Also as you have found, there are no specific calibration points for the RTD range, so it's hard to claim any traceability there?  Seems strange for an advertised function of the device.

In any case, I think I have fully resigned to the fact that I will need to perform calibrations using direct mV stimulus and measurement, and handling temperature conversion and any RJ implications manually in post.  Likewise for RTDs, I will need to do some exploration and proficiency testing to see if the resistance simulation will be adequate for RTD ranges, or if I need to resort to a decade resistor instead, characterised with the 8588A.

I don't think we have solved all the mysteries, but the picture is much clearer with your assistance.  :-+
 

Offline dl1640

  • Regular Contributor
  • *
  • Posts: 212
  • Country: cn
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #16 on: April 21, 2020, 01:58:35 pm »
The CJC of 55xx itself is referenced at ambient temperature, so it may not be that stable . We use external ice point bath as CJC, as well as a calibrated thermocouple for calibration of temperature indicating device . For RTD calibration mostly is to use 3-wire, so the 4-wire mode and 2-wire comp are useless of 55xx, for fixed point calibration I use calibrated resistors which are *low noise*.
 
The following users thanked this post: binary01

Offline itsbiodiversity

  • Regular Contributor
  • *
  • Posts: 77
  • Country: us
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #17 on: July 11, 2020, 08:46:09 am »
I have a Transmille 3041A and find it a delight to use. The TC simulation adapter truly has highlighted all of the concerns you laid forth in your comments. With Auto CJC enabled I am able to get very precise TC measurements/source , but I am amazed at the number of products that barely meet specification.  I believe it is due to errors in factory cal due to poor CJC methodology.
 

Offline bgugi

  • Contributor
  • Posts: 12
  • Country: us
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #18 on: July 16, 2020, 06:02:18 am »
A few notes on the RTD disparity: Footnote three in the resistance specification table has the resistance floor increase with decreasing excitation current - many industrial RTD transmitters operate much lower than the metrology pseudostandard of 1mA (The first one i found in a quick google search indicates "<200 microamp" https://www.omega.com/en-us/communication-and-connectivity/signal-conditioners-and-transmitters/signal-conditioners/p/TX12-Series ). In addition, the RTD spec is for COMP-OFF and 2- and 4-wire compensation.

as others have mentioned, the TC spec includes the uncertainty from CJC. If the instrument you're calibrating supports direct voltage measurement, you're typically better off performing the voltage and cold junction calibrations separately. If this isn't an option, you can get away with a calibrated "ice-point" probe (cu-cu to tc wire extension) in a lag bath (thermos full of ambient-temp water) measured by a precision thermometer (if you don't plan on extending your range much beyond ambient, you're probably best off with a reference thermistor).
 

Offline binary01

  • Contributor
  • Posts: 15
  • Country: au
Re: A temperature disconnect - understanding TC & RTD calibration specs
« Reply #19 on: July 16, 2020, 11:54:35 am »
Yes bgugi, I think the mystery is finally solved on this matter, and I have been largely succesful in overcoming the accuracy specification limitations by using mV and 4-wire Ohm source and measure directly. I don't use the TC or RTD functions directly at all on my DMM or calibrator.

For tests where I need to assess the internal CJC of test instruments, I am using an thermocouple reference probe (cu-cu to TC wire) in an ambient temperature lag bath, measured with an RTD, exactly as you have mentioned.  It is working quite well after spending quite a bit time developing it.  The excercise of characterising the thermocouple reference probe has been interesting, as I have found its errors vs the expected reference values to be too large to ignore for my target uncertainty.  With a typical lag bath temperature of 20C, and the test instrument connection typically 22C to 24C depending on the model (i.e. two junctions are very close in temp), I expected the probe errors to be insignificant, but this assumption proved wrong   |O
In other cases where internal CJC isn't used, it is much simpler, and I source/measure mV directly according to reference tables.

You are correct regarding RTDs also.  Especially when calibrating RTD simulation functions, I need to use low current modes or select higher Ohm ranges on my DMM to keep the force current low enough.

Thanks for your feedback on the Transmille 3041A, itsbiodiversity.   I actually ended up buying a 5522A but I'm very glad to hear that the 3041A is working well for you.  I would have liked to support an alternative brand, but in particular the improved resistance simulation function on the 5522A swayed me in the end.  The TC simulator module on the Transmille did look like a good add-on, but it's spec was still a little short in my partcular case when you add the CJC uncertainty component.  The DC current spec on the Transmille 3010A was attractive, and much better than the 5522A.

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf