0
\$\begingroup\$

I have a 2 layer board that I am mounting directly to an aluminum housing. The circuit contains a temperature regulation circuit to monitor housing temperature by way of an NTC thermistor. As the resistance drops, power is reduced until steady state is reached.

Below is a pic of how I did it. The thermistor pad is connected to the ground plane with some vias. Hopefully the heat conducts from the housing, into the ground plane, and into the thermistor. Could this work? Does the thermistor body need to be mounted to the surface to work properly? Any idea how accurate this might be and is there a better way or standard practice?

I should note this is a little 0805 package thermistor and current is extremely low, it's just a sensing circuit not a power circuit.

Thank you

thermistor

\$\endgroup\$
4
  • \$\begingroup\$ How does heat transfer from housing to GND plane? You can do better than "hopefully" by ensuring a good contact area. \$\endgroup\$ – user_1818839 Jan 10 at 12:40
  • \$\begingroup\$ Where is the source of heat? Is it the PCB, or something else on the housing? What you have is going to measure the PCB temperature. How that relates to the housing is a WAG (OK, maybe a SWAG) unless the proper thermal analysis has been done that shows how you can infer the housing temperature from that of the PCB. \$\endgroup\$ – SteveSh Jan 10 at 12:54
  • \$\begingroup\$ @SteveSh The housing generates the heat. The board (1" square) has full surface contact to the housing with thermal paste and is screwed down. Therefore the entire ground plane should be at the temperature of the housing. I guess my concerns are: will the vias sufficiently transfer the heat? And is heating a thermistor through the pad acceptable? \$\endgroup\$ – kobra20 Jan 10 at 18:50
  • 1
    \$\begingroup\$ "is heating a thermistor through the pad acceptable?" Yes. We use thermistors in vacuum (no convection) environments all the time to measure PCB temperatures. So long as your PCB is tightly coupled thermally to your housing, you should be OK for what you described. \$\endgroup\$ – SteveSh Jan 11 at 0:19
1
\$\begingroup\$

If you want it primarily coupled to the housing, eliminate the any copper pour thermal connections to the rest of the PCB, use thin conductors (perhaps in serpentine pattern) and consider slotting the board to further isolate the sensor.

You can also insulate the PCB around the sensor to attempt to isolate it from the air in the enclosure. Something like self-adhesive polyurethane foam tape (but be careful that the electrical properties of the adhesive do not come into play).

\$\endgroup\$
0
\$\begingroup\$

Perhaps consider to add vias to the left and right of the thermistor as well. If possible you could also use the mounting as a thermal conductor (assuming that is a mounting hole visible)

How accurate do you need this to be? The accuracy is also very dependent on the coupling between the heatsoruce and aluminum.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.