My buddy and I were arguing over if you need a bath to calibrate Fluke 52's?

[USMCPMEL]

He says you don't I think you do. he calibrates it directly from the Fluke 5520A. Opinions? Evidence to support said opinions please.

Thanks in advance.

[OlDave]

A Fluke 52 can measure type J, K, T and E thermocouples. The temperature ranges go from -250°C for Type T to 1372°C for type K. I doubt you have sources to produce temperatures covering this range.

Typically these units are calibrated by electrical simulation using a standard similar to the Fluke 5220 to verify conformance to NIST-175 so the customer can use the thermocouple of their choice.

We do have special applications where the customer will send a thermocouple with the meter and request a system calibration of the meter and their thermocouple at select temperatures. In those cases we perform the electrical simulation on the meter first and then verify the system accuracy at their test points using a bath or dry well.

[USMCPMEL]

That is what got us started on the subject. He had a customer arguing with him over the readings he was getting. My friend calmly explained to the customer that the thermocouple also has a tolerance. Some people do not understand that they think that if the 52 is dead on when they plug in they should get the same readings and they do not take into account thermocouple error.

[metrologygeek]

Our procedure is that if the customer does not send in the thermocouple they are working with we put a Limited Use on the 52 of ±2.2°C or ±0.75%, whichever is greater, in addition to the Fluke 52's spec. This would be the worst case spec on the base metal thermocouples that it reads, and it wakes the customer up to the fact that they need to consider the tolerance of the thermocouple in addition to the spec of the meter. If they do send it in with the sensor, which they usually do to avoid the Limited Use, we can go from -200°C to 1450°C, but we always try to get the temperature range of interest so we can hit points generally within the range of use. When we simulate a sensor we use an Ectron 1140A, which has always given us good results. Digital thermometers stay in our temperature lab rather than going to the general purpose electronics lab.

[sdmetrol]

There are different series of the 52.  If you look at the calibration procedure for both of them,  they call for an ice bath and a Fluke 5440, 5700, or equivalent.  The 5520 may meet the specs. for dc output but we use our 5720A and an ice bath.  The T/C output of the 5500 & 5520 do not meet the 4:1 ratio throughout the range of the 52.  I would have to agree with you.

MGeek, the 52 does have user offset adjustments to compensate for t/c errors, so limited use seems harsh to me.  My clients would push back on that for sure.  You are covering your bases by calibrating the unit with an ice bath and accurate dc source.  The thermocouple error is only relevant when it is sent in with the 52 and matched by some means that prevents tampering.  They could easily replace the t/c at any time.

[Mike]

I don't think a bath is entirely necessary, but a little deodorant would be nice.

<Ba Dum Bump>

[Hawaii596]

Good one, Mike.

We treat the thermocouple and the meter as two separate calibrations.  At least on the original model 51/52, the zero Celsius offset was considered an operator adjustment, as no two thermocouples connected are going to read exactly the same (since the metal purity, surface oxidation, etc. is going to vary from thermocouple to thermocouple and connector to connector.

So unless the thermocouple probe is calibration void sealed to the meter, I would have a little trouble calibrating the meter with an ice batch, especially since that is only for setting the zero offset of the given thermocouple.  Matter of fact, even the (let's say K) male-to-male mini jumper from the calibrator to the meter will have variability as well.

CAVIAT:  If you do a higher accuracy calibration on a 51/52, and you are using a 5700A with precision divider, a cold junction and ice point, etc.  In that case, an ice batch may be of use if your lab does not have an ice point cell (such as the Kaye 140 or Omega TRC-III, etc.). 

So, ice bath needed only if you are using the latter method above.

[sdmetrol]

Point well taken on the ice point reference as opposed to bath only.

Regardless, I assumed the question was regarding direct calibration with a 5520A without using a zero reference.  The specifications of the t/c output on the 5520A does not meet a 4:1 ratio throughout the range (if I have done my math correctly).

[Hawaii596]

Although it is, I think, common to calibrate Fluke 51/52 directly with 5520A T/C output, I believe you are correct that it isn't 4:1 across all of the ranges.  It's been a while since I've looked it up.  Guess I need to look it up myself to update my mental database.