I'm building an outdoor sensor system (a bunch of sensors and a Raspberry Pi inside a plastic enclosure). A few weeks ago, one of my prototypes got too cold (ambient temperature went below 0deg Celsius). While the sensors can handle sub-zero temperatures for a short time, it does reduce their lifespan, so I want to heat the components up slightly.

I bought a heating pad and have got it working (using one of these as a controller). However, it seems to keep all of its heat within the pad, rather than heating up the air around it. For instance, if I put my hand near the pad (2-3cm above it), I don't feel any heat, but if I touch the pad, it is very hot. My thermometer showed it reached above 50deg Celsius, so I know it's working.

How do I go about distributing this heat towards specific components within my enclosure?

Should I just stick a big heat sink onto it, and then have a fan pointing at the heat sink and facing in the direction of where I want the heat to go? Or is there some more efficient method?


  • 1
    \$\begingroup\$ Part of what seems to be missing from your attempt is an insulated enclosure. You need to prevent the heating element from getting dangerously hot, but you'll have an easier time keeping the other parts warm if you minimize the heat lost from the system, which will minimize the heat which needs to be produced and flow from the heater element to the other parts. \$\endgroup\$ Commented May 15, 2020 at 1:01
  • \$\begingroup\$ Yes, the enclosure will be well-insulated when I build it. I'm just experimenting on my bench-top at this point. Do you think an insulated case would be sufficient? Thanks \$\endgroup\$
    – ZPMMaker
    Commented May 15, 2020 at 1:11
  • \$\begingroup\$ How big of an enclosure? Seems like you could build a cheap heater to keep the temp above your desired temp and use a thermistor to turn it off. You may have the opposite problem when the weather heats up... \$\endgroup\$
    – Ron Beyer
    Commented May 15, 2020 at 1:12
  • \$\begingroup\$ Maybe 12in/30cm diameter, 6in/15cm tall. So nothing massive... And yes I am using a thermistor to control the heater. Cooling will indeed be another issue, but I'm tackling this one step at a time. I do already have some ideas on how to keep it all cool (Peltier coolers and heat sinks). \$\endgroup\$
    – ZPMMaker
    Commented May 15, 2020 at 1:28
  • \$\begingroup\$ Couple to a metal sheet. Better still couple to enclosed water - which resists freezing as a bonus. \$\endgroup\$
    – Russell McMahon
    Commented May 15, 2020 at 4:35

2 Answers 2


I ended up applying thermal paste to the heat pad, and then attached a low-profile computer CPU fan, which I then powered separately (direct from 12V battery, via a step-down converter to 5V to suit the fan). That was enough to spread the heat around the enclosure.


Just use 4 resistors, each rated at TWO WATTS, axial leaded, placed near the edge (4) of your PCB, with the heat conducted thru the leads into your PCB and keeping the various GND and Power planes warm.

I suggest TWO WATT resistors because of the thick leads.

You might also use 4 time 270 ohms, to pull about 0.1 amp from the 117VAC power line, daisy-chained about the PCB periphery; use a SCR or TRIAC with opto-isolator that is activated when below +10 ° C.

Make all the "daisy-chain" intermediate-connection traces WIDE and located over the GND plane, so the heat gets dumped thru the FR-4 into the GND plane. Of course, some displacement current coupling will occur from the 117VAC into the Ground plane.


Can you build the heater into a NEW layer inside the PCB?

Conaider a 4" by 4" layer, with 10 mil traces separate by 10 mils. Thus 50 traces per inch, 200 traces total, of length 4" * 200 = 800 inches.

The Length/Width tells us the # of squares.

800/0.01 = 80,000 square of copper.

At 0.000500 ohms/square (for standard thickness 1 ounce/squarefoot foil), or 1/2000 ohms almost exactly, there are 40 ohms of resistance.

With 5 volts across that 40 ohms, the power = 5 * 5 /40 == 0.625 watts.

  • \$\begingroup\$ I wonder how a dedicated inner sadwiched between two planes, densely packed and weaving 10 mil , thin heater traces would fare. Kind of a waste of a layer though. \$\endgroup\$
    – DKNguyen
    Commented May 15, 2020 at 13:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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