14

I have to develop a microcontroller circuit that will run at low temperatures (as low as -60C). I want to heat the FR4 PCB board until it reaches the commercial temperature range above -40C. I found some flexi heaters to heat the PCB. What other options do I have? Is it possible to use a PCB layer as heater? I couldn't find any information about using a layer as a heater.

PeterJ
  • 17,131
  • 37
  • 56
  • 91
zontragon
  • 261
  • 2
  • 8
  • 3
    Just use a few surface-mount resistors, just be careful of heat cycling cracking the solder joints over time. – John U Nov 01 '13 at 14:42
  • What mechanism will be controlling your heater, before MC starts (-60C - -40C)? For heater, see this link - [Heated-bed](http://reprap.org/wiki/PCB_Heatbed) – Oleg Olivson Nov 01 '13 at 14:51
  • 1
    I'd love to see an internal layer being used as a heater. You'll want to meander the trace to get it nice and long, but all the calculations for power dissipation and temperature rise of internal copper traces are readily available on the internet. Is power consumption not an issue? If it is, you can use a RTG and kill 2 birds with 1 stone :> – HL-SDK Nov 01 '13 at 14:51
  • Try this: http://www.pcbco.com.au/tracecalc.html – HL-SDK Nov 01 '13 at 14:53
  • I'd contact the microcontroller manufacturer and ask for info on running the device at low temp – Gustavo Litovsky Nov 01 '13 at 15:43
  • @JohnU It's good question. My Microcontrollers specs says it's minimum working temperature is -40C. I'm not sure if it's possible to run the microcontroller at -60C to control warming process. – zontragon Nov 04 '13 at 08:50
  • Use PTC resistor(s) connected across the power rail, no control needed. Could also use one to hold the RESET controller / line to prevent the micro from starting when very cold. – John U Nov 05 '13 at 17:11
  • @JohnU How to tell when the temperature is good enough to start the system? – richieqianle Jun 19 '15 at 02:09
  • 1
    What about some sort of _temperature sensor_, or a small thermistor connected to a switching circuit? – John U Jun 19 '15 at 09:13
  • Sounds reasonable to me. I will try to design one. Thanks John! – richieqianle Jun 20 '15 at 02:19

2 Answers2

13

I would be inclined to use traces on or in the PCB as a direct heater, as you suggest. Obviously, you'd start by insulating the board as much as possible to minimize the energy required to maintain temperature. Pay particular attention to external electrical connections, which can also be good conductors of heat. Extra lengths of wire buried in the insulation will help quite a bit.

I would put the trace(s) mostly around the edge of the board, generally keeping them away from other heat-producing components in order to keep temperature gradients across the board to a minimum. I would design a current source that has temperature feedback to drive the trace, with a "maximum" setting that's guaranteed not to burn it out.

The flexi heaters you found are essentially doing the same thing, but they probably consider the applications data somewhat proprietary. You'll have to work out the details yourself; things like trace widths and thicknesses, current levels and temperature rise, based on the basic resistivity of copper, which is 1.68×10-8 Ω-m @ 20°C.

Dave Tweed
  • 168,369
  • 17
  • 228
  • 393
  • 1
    is it wise to place traces on a internal layer becouse of the low thermal conductivity of FR4 epoxy? – zontragon Nov 04 '13 at 13:27
  • 2
    @zontragon: It really doesn't matter all that much. You're not trying to move a lot of heat, like you are with a heatsink, just raise the temperature of the system overall. – Dave Tweed Nov 04 '13 at 14:19
  • Why would you rather to use pcb traces instead of using external flexi heater? What advantages does it have? – zontragon Nov 04 '13 at 14:45
  • @zontragon: Because it's basically zero incremental cost -- included in the cost of the PCB. – Dave Tweed Nov 04 '13 at 21:55
8

I had to design a 400MHz radio receiver data-demodulator that could sit down at -60 ºC and I put a bunch of surface mount resistors in a matrix formation underneath the PCB on the bottom trace (4-layer board). The matrix meant if one resistor failed it didn't kill the whole heater idea.

I used a temperature sensor to control the heating and at round about +10 ºC this turned off because the same module could also sit out in the baking heat all day. The only problem that the unit initially had was when heating back up through 0 ºC - condensation formed on the circuit board and for 5 minutes the radio didn't operate as expected. This was cured with selectively lacquering vulnerable areas of the PCB.

There were still a few noticable glitches as it rose back through 0 ºC but these were very short-lived. The unit was designed for an aero engine test cell - not a nice place to be when the engine was on! The radio receiver was collecting data from the engine.

Andy aka
  • 434,556
  • 28
  • 351
  • 777
  • Nice design! But if the system sits below -60 degree and is shut down, how to start the system without running MCU under 60 degree? – richieqianle Jun 19 '15 at 02:12
  • @richieqianle it didn't have an MCU. – Andy aka Jun 19 '15 at 08:33
  • Could you please introduce how to wake up/close the temp sensor? – richieqianle Jun 19 '15 at 14:58
  • @richieqianle it was all done in analogue with a comparator as I remember. – Andy aka Jun 19 '15 at 15:23
  • To avoid starting the other part of the circuit(except the heating part) at low temperature, what was done? – richieqianle Jun 20 '15 at 02:20
  • @richieqianle I'm not sure what you mean. From memory, everything remained powered all the time and the box was tested like this in our lab in a special freezer. – Andy aka Jun 20 '15 at 11:28
  • The circuit can be divided into two parts: heater part and function part. The function part is designed to work above +10 degree, in your case, therefore a heater is needed. If the whole circuit is put into -55 degree, the heater part should be started first and the function part shall not work until the environment is heated up to +10 degree. How was this done? – richieqianle Jun 22 '15 at 02:28
  • @richieqianle I'm not sure where you are going with this. Do you have a question that maybe should be raised as a proper question on stack exchange? – Andy aka Jun 22 '15 at 07:05
  • Thanks for your reminder. I started a new post: http://electronics.stackexchange.com/questions/176720/how-to-design-circuits-to-work-under-55-celsius-degree – richieqianle Jun 22 '15 at 11:56