I have problems with thermocouples type T on CEM20. The readings are erroneous and might change between -6.0 to 26.8 C between two scans (RA1M) even though that soil temperature is a stable substrate.

I have both tried single ended and differential connection, but neither of them works. I have checked/logged the 1REFT and it is stable. If I use an immediate scan, it seems that the measurement is Ok, but when I am using a scan schedule something goes wrong.

The exact same configuration have been used on a DT505 with CEM. Is there a special software switch that should be add to the scanning list?

Best regards,
Niels Andersson

I am having a similar problem with a DT80 (without the CEM). We are using T-Type wire that is twisted together at the end to form the junction.

These are then buried in concrete some also in just soil. Yes they are not coated or covered to insulate them electrically. Unfortunately I inherited this project after the installation. In my case there seems to be a stray electrical current in the ground causing the problem because it starts at 6:31 pm and goes to 7:01 am every day.

I tried using two thermocouples that were just exposed to the air. These did not have the same noise problem, which makes me think there is some cross current between the different channels. We were using a different data logger at first that didn't show this problem, but they were poor quality and failed often. I will post here if I find a solution.

Good morning Neils,

Thermocouples have a very small output. For a T type thermocouple the output is about 39 millionths of a Volt per degree C and is non-linear. This makes them quite difficult to read and susceptible to noise sources.

For example if we ignore the CJC, a change of -6 to 26 Deg C is about 1.26 mV for a T type thermocouple. With bare tipped thermocouples in soil it is quite possible that minerals in the soil along with moisture are causing a galvanic cell (battery) that is creating voltages at the tip of the thermocouple.

If this is what is happening then the only thing you can do is electrically insulate the thermocouple from the soil. Try cleaning the tip and dipping it in epoxy resin or a neutral pH silicon sealing compound.

Cheers,
Roger

Good morning Jeff,

Continuing on with other noise sources. Having noise at a very specific time in the morning does strongly indicate equipment being turned on.

If the thermocouples are wires as differential inputs between the +/- or */# then they are completely isolated from one another and there won't be cross currents or ground loops. If you are using single ended inputs (Common ground) then there can be ground loops between the thermocouples on the same channel. If you can possibly wire the inputs as differential then it might help the problem.

There are other sources of noise that we need to consider. It is possible that the noise can be coming from the mains power supply. A simple test is connect DT80 to a nice big external battery and let it run for a couple of days. Then down load the data.

If there is no noise then you will need to look at using power supply filtering. If it is EMF being picked up from by the thermocouples then there is very little you can do. I don't think your boss would like you to dig up the thermocouples and replace them with ones that are shielded twisted pair.

Cheers,
Roger

The thermocouple I use is hermetic sealed and it is working without problems on the DT85. Is it possible to change the cold junction to the one for DT85 instead of using the one in the CEM20?

Best regards
Niels Andersson

Good morning Niels,

The addressing of the CEM20 tells DT80 to use the CJC in the expansion module.
If the CJC in the CEM20 fails then you will see a reference error. You can check the reference with REFT command, if the CEM20 is on address 1 send the command 1REFT.

It is very important to keep the CEM20 at a constant temperature. Please make sure it is not exposed to anything that might cause a change in temperature such as sun light, air conditioning, sudden changes in temperature etc.

Cheers,
Roger