weird problem max31855

I've connected a max31855 with a type K thermocouple connected to it, to an atmega328, which I'm powering via a USB port on outlet like this one:

All works fine until I stick the metal end of the thermocouple probe like this one:

into the soil (that I'm measuring temperature of), at which time I get unpredictable readings. If I wrap duct tape around the probe to insulate it from the soil, again all is fine. At first I thought maybe just noise from the switching regulator, so put a clamp-on bead on the USB cable, but that didn't help.

I'm guessing some kind of ground loop through the switching regulator for the USB port and back to house mains, but not sure how to fix it, while still using this port, which is very convenient for this project. Any ideas?


Is that outlet a mains and USB outlet?
If so then the gnd of the USB is probably connected to the mains gnd.

If the temperature probe case is connected to the USB gnd, you will be making a ground loop when you push it into the gnd.

Insulating the temp probe will break that loop and hence any noise on the system.

Tom.... :slight_smile:

Thanks TomGeorge. Kinda what I figured about the ground loop which is why I wrapped the end with duct tape. I was wondering if there is another solution. The reason being, is that even if the wrap over the wires touch the soil, I also get the error. Apparently, being just a cheap probe from china, the wire junction inside the probe is touching the probe casing? Since it's hard to keep that wrap off the soil because it's kinda long, I also had to wrap it in tape which is really kinda weird. I obviously can't connect the DC ground to the mains common, but I wonder if I connect it to the PE (green) if that would do anything? Maybe easier for me to just try it than to make anyone give it much thought as to whether it would work. :-/

Just checked, and the case isn’t connected to the USB ground - not connected to anything.

You probably have a "grounded" TC, test each lead to the sheath with an ohmmeter, you need an ungrounded or "isolated" TC.

outsider; as I suggested above, being a cheap probe from china, the junction inside the probe is likely touching the probe inside and thus the sheath, and even though the sheath isn't connected to the USB ground, it's likely making a loop since it's obviously touching the soil. Likely no fixing that. I have some TC extension cable, maybe I can make a junction at an end of it, and poke it into the ground so I only need to insulate that end. I'll try that.

Thanks to both of you for your input. :slight_smile:

. . . and good point outsider. Found this:

"Grounded vs. Ungrounded

The grounded thermocouple loop has a faster response time to temperature change, but the disadvantages lay in that the hot end, grounded to the sheath, will exposes the circuit to transient earth loops and this causes noise on the circuit.

The ungrounded thermocouple response time to temperature change, has a slower response time than the grounded thermocouple. In this circuit the hot end will not have been directly coupled to the ground, so therefore regarded as isolated from the ground, by mineral insulation. The advantage being, that the loop will no longer have the same transient noise on the circuit, as the grounded thermocouple."


Maybe I can take the probe apart and just silicone or somehow insulate the junction. I don't need fast response time, or even very precise measurement. :slight_smile: