Low Temperature Deviation Issue with Hart Scientific 5615 RTD

 metroman I'm no expert, but the problem is intriguing.  Is the Hart the "odd man out" or do you have 5 unique instruments?  Have you moved the sensors between instruments to isolate it to the 5615?  With the sensors being as precises as they are, maybe it's the difference of a picogram of solder somewhere in the circuit.  Are there any probe adapters or anything like that in your sensor circuit that makes that instrument different?

I have no actual experience with these devices.

I can make some suggestions as to how you might proceed.

It's not quite clear to me what your set up looks like - does each sensor have its own read out device or are you using 1 read out device with 5 different sensors.

If the 5615 sensor is always used with the same read out device ( by which I mean the same single physical device, not just the same make and model)  then you can not tell if the errors comes from the read out device or the sensor.

How does the calibration work - is it electronic or paper - what does the calibration of the 5615 look like compared with the other sensors.

What does the error profile look like "Below 0 °C, the 5615 starts to deviate by approximately 0.2°C" isn't specific enough - a graph would help.

Is 20 minutes long enough (a graph of reading against time would tell).

What are the differences between the sensors (dimensions, cable, connectors etc.)

What do Fluke have to say about it ?

MK

Thank you very much for your reply I really appreciate it!

To clarify:

• All five sensors are Hart Scientific RTDs, but the one showing the deviation is of a different model than the others  that’s the only notable difference. I didn't know it but when I looked I saw that it had a slightly different shape.
• Each sensor is connected to its own dedicated readout, and they are calibrated as a pair (sensor + instrument), so swapping instruments isn’t an option without breaking traceability.
• I’ve tested the 5615 in different environments (oil bath and air oven), and even in isolation with just one other sensor the deviation at low temperatures remains consistent.
• There are no adapters or special connections in the setup all sensors are connected identically, I also did some tests with different connections but nothing changes.

The deviation is about 0.2 °C below 0 °C, which is well beyond the expected tolerance. It's true that with sensors this precise, even the tiniest physical or electrical variation could potentially have an effect.
That said, it would be extremely difficult to detect or confirm something like a microscopic solder variation in practice, especially since the setup is otherwise very clean and consistent.

Thanks again for your input your questions helped me think it through more clearly!

Thank you so much for your reply.

Each sensor in my setup is paired with its own dedicated readout device, and they are calibrated together as a system. Because of this, I can’t swap instruments. The sensor that shows the deviation is a different model from the others, although all of them are Hart Scientific RTDs and those particular sensors were calibrated by the same lab on the same range of temperature -50 to 150 °C.

The deviation of about 0.2 °C below 0 °C is clearly beyond what would be expected for this class of sensor. To be more precize for instance when I go to -20 °C the 4 other sensors show values around -19.923°C and the 5615 sensor shows a value around -19.710°C. This is very strange because it behaves like that only for those low temperatures and behaves correctly for the rest. I will try and simulate a stabilization curve to be sure (thanks for the idea), however 20 minutes should be more than enough for thermal equilibrium in this setup (according to specifications and my past measurements done with this sensors.)

All sensors use the same cables, connectors, and immersion depth P1, so there are no physical differences that could explain the issue. In the image above you can find my attempt to schematize my setup.  I haven’t contacted Fluke yet, but that’s the next step I plan to take.

Thanks again for your insights they’ve really helped me clarify my approach and think more.