15

I need to design a PCB circuit to work at (or even below) -40 degree centigrade. Normally, I use FR-4 PCB board at room temperature. However, I have learned from the question "What is the minimum temperature for FR-4 PCBs?" that FR-4 PCBs may have problems below -30 degree centigrade.

So what kind of PCB material can handle such low temperature?

Community
  • 1
billyzhao
  • 623
  • 6
  • 20
  • 1
    -40 °C to 85 °C is an *extremely* common temperature range (generally called "industrial", but different mfcs have different bounds) for components and particularly modules (as we're concerned about PCBs here). At previous companies I've worked at (who mostly made automotive-type modules), all used bog-standard FR-4. – Nick T Jun 15 '15 at 18:11

3 Answers3

14

I've used FR4 at 4K and others have used it at much lower temperatures.

The physical characteristics go somewhat downhill at low temperatures, but board failure such as delamination does not normally occur from mere exposure to cold temperatures. The Charpy tests referred to in your linked answer are a measure of strength of a notched specimen to shock (add a stress riser then whack it with a hammer, basically). If you are in a severe mechanical environment you may have to consider the lower impact strength and use a thicker board or support it better.

Solder joint failure due to differences in coefficient of thermal expansion can be a factor, especially with lead-free solder and things like large BGA packages.

-40°C is just a nippy day in some parts of the world, and -55°C is the lower end of the military temperature range, both limits are within the normal range of epoxy-glass boards, and there are plenty of reasonably-priced components available that have guaranteed specifications at those temperatures (especially -40).

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
  • Thank you! 4K means -269 degree centigrade? And it works good even exposed to such low temperature for a long time? – billyzhao Jun 15 '15 at 11:46
  • 5
    @billyzhao Yes, that's right. Of course we're not whacking it with a hammer inside the cryostat. – Spehro Pefhany Jun 15 '15 at 11:51
  • That`s great. But one more thing: at the moment starting the circuit, the regulator or the processor would produce heat. The heat changes the temperature of the PCB board. Would such temperature change cause failure? – billyzhao Jun 15 '15 at 12:22
  • Generally there is no problem with moderate relatively slow changes of temperature. The glass transition temperature marks a point of inflection in the CTE (it significantly increases at temperatures *above* Tg) but that's typically at ~130°C (more like >170°C on higher performance materials). – Spehro Pefhany Jun 15 '15 at 12:30
  • 2
    How does one go "much lower" than 4K? – crocboy Jun 15 '15 at 13:22
  • 7
    @crocboy You can only go <4 degrees lower, obviously, but some things change significantly through single digits Kelvin. Resistivity of copper levels off to RRR, heat capacity drops precipitously etc. It's also much, much harder to get to millikelvins than 4.2K. – Spehro Pefhany Jun 15 '15 at 13:55
  • In a broad sense most of the (mechanical) thermal effects present at 4K will be present at 77K (liquid nitrogen, which is *much* easier to get hold of and test). This includes most of the thermal contraction and embrittlement. At 233K (-40°C) that will be less true, but in general plastic materials will be more brittle and non-metal components will have lower thermal conudicivities. – Chris H Jun 15 '15 at 19:57
  • 1
    I forgot to mention it before, but: at these temperatures, [tin pest](https://en.wikipedia.org/wiki/Tin_pest) will likely be a problem. You allude to problems with lead-free solder, but this is a far more serious issue than thermal expansion mismatch. For short duration of service at cold temperatures, tin pest will probably never develop, but if it is constantly at -40°C, it certainly will. – Oleksandr R. Jun 11 '16 at 23:29
3

You can use a ceramic substrate: see e.g. here. Of course it will be expensive, and may require special tooling, and I should say that have no personal experience with these. But it is the usual solution for circuits that must operate at extremes of temperature.

If it is not absolutely ruled out by the application, you might also consider using a heated enclosure. This will be less costly and will allow you to use conventional materials and techniques.

Oleksandr R.
  • 1,644
  • 11
  • 14
1

My solution would be to enclose the FR-4 PCB in an insulated hermetic enclosure, with an insulation gradient of at least 20 degrees C. If more temperature control is needed, I would add a bi-directional heating/cooling device. this way, the temperature outside the enclosure can be more extreme and still not affect the temperature inside the enclosure. This method also avoids large local PCB temperature gradients.

Guill
  • 2,430
  • 10
  • 6