We actually built 3 “in-house” prototypes that utilized different style thermistors(rivet style, tab w/ screw, etc…) and housed them in a single pvc pipe that was capped. We filled the pvc with spray foam to try and deal with moisture issues. All wires from thermistor were ran to a campbell datalogger at our Sterling, VA test facility and all seemed to work out great. Problem is choosing one and finding economical production of such a device.
Our plans to move forward with soil temp data gathering with this style of probe will be forthcoming in the next year or so. As I get more info or find a source I’ll update the community.