2

I am working on a custom PCB design. The SoC chipset which we use supports only commercial grade temperature (minimum operating junction temperature of SoC chip is 0ºC.) We want to operate the device in cold ambient temperature (say -30ºC.)

  1. Are there any validated simple solution for heating solutions for PCBs to operate in cold conditions?
  2. What will be the time required for temperature rise of 10ºC?
  3. Will there will be any impact in the life time of the heaters when it is operated in longer duration under cold temperatures?
JRE
  • 67,678
  • 8
  • 104
  • 179
Raman
  • 21
  • 1
  • 1
    Welcome to the site. I would say the answers depend on how you build the heater. – jay Sep 01 '21 at 17:36
  • What power is available? – Andrew Morton Sep 01 '21 at 17:43
  • There's something called "The Heat Equation" (iirc) that might be partly an answer to your question. It states that q = Cp ∙ m ∙ ∆T, which is to say the amount of energy in Joules (q) required to increase the temperature of an object with a certain specific heat in Joules per gram per Kelvin (Cp) and of a certain mass in grams (m) by a certain amount in Kelvin (∆T) is the the product of those three terms. It's "complicated" because your PCB is a heterogeneous material including copper and fiberglass, but I believe it would have an "effective specific heat" that you could deduce empirically. – vicatcu Sep 01 '21 at 17:44
  • Is this a commercial product or a one off design? If the latter I would consider buying the chips and then thermal cycling them to see if they actually fail under your conditions, especially if you don't expect to hit the upper end of the thermal range. Most logic will work fine down to moderately low temperatures, but thermal expansion can be a problem if you want to do -30C to 85C regularly. – user1850479 Sep 01 '21 at 17:56
  • Some things **are** rocket science... not being able to source SoCs with extended temperature range for usage higher-atmospheric temperature ranges does sound like the kind of rocket science I'm instinctively sceptic about. – Marcus Müller Sep 01 '21 at 17:56
  • Do not switch the SOC off. Usually, it will heat itself efficiently enough – 0___________ Sep 02 '21 at 01:26
  • The simplest method is put a PTC heater near/on top of the chip. I used PTC heater plate(Self-regulating heater), just wire a power supply to it and it will heat itself to a fixed temperature. But if you want a more precise temperature, simple power supply + PTC is not enough. – Chengxian Zhang Sep 02 '21 at 01:36

5 Answers5

2

You might research near-space balloon technology, which relies on heaters and insulation to use consumer-grade electronics in cold environments. For example, Improving a Low-Cost Thermal-Vac Chamber for Testing Stratospheric Ballooning Payloads by Meyer, Straub, and Flaten discusses use of heaters an insulation. L. Paul Verhage has published a series of articles in Nuts & Volts Magazine on designing payloads to operate in the cold.

Perhaps you need to test if the heater must be left on at all times, to prevent damage to components such as batteries, as well as IC's, even in the off state, or if the components only need to be heated while operating. In that case, you'd also need to determine how long a "pre-heat" time is needed. Though math may help, ultimately, testing under realistic conditions is the only way to check your calculations for thermal mass, internal thermal conductivity, power and external heat loss.

DrMoishe Pippik
  • 3,634
  • 5
  • 13
0

The easiest way would to place some resistors around the device and run current through them to warm the area. You can control the temperature if you want.The time to raise the temperature depends on the physical properties of the board etc, what you are going to lose heat by radiation and conduction. There is no one simple answer unless we have a lot more details. If you are using them in large quantities by the manufacturers definition you may be able to get them pre sorted to your temperature requirement and or a specification at that temperature. If the design is not finalized is there another solution that will meet your temperature etc requirements.

Gil
  • 4,951
  • 2
  • 12
  • 14
0

The moisture ingress ratings around interconnects and plastic determine the temperature barrier of 0'C.

The only catastrophic failure mode for low voltage is not -30'C, but the transition from +1'C to -1'C which can occur going to sleep in cold weather if self-heating is not sufficient.

SMD Resistors on FPC board with a foam insulator with self-adhesive and thermistor + comparator on board is all you need to add a Watt or so of power to keep the epoxy above 0'C. As long as the chip still has room to cool at max temp, you can make tradeoff's for P vs Rth.

For unpowered transport, lots of insulation can reduce the rate of change of T such that the popcorn freeze effect does not occur.

Tony Stewart EE75
  • 1
  • 3
  • 54
  • 182
0

IMO, heating is fairly simple by use of some heating element + fan. But the real concern is the condensing. When you cycle the heat, then the air inside expands and shrinks, if its sealed it would pressurize and vacuumize. But if not sealed, the humid air will intake and then after cooling down it will condensate. You'll get water on your PCB.

Marko Buršič
  • 23,562
  • 2
  • 20
  • 33
0
  1. Are there any validated simple solution for heating solutions for PCBs to operate in cold conditions?

Yes. Commercially available heater strips are available in a variety of wattages and voltages.

  1. What will be the time required for temperature rise of 10ºC?

All depends on how big the heater is, and what your thermal loss is to the environment. I'm sure I could find a heater that would warm your PCB up 10 deg C in less than a minute (think of a toaster).

  1. Will there will be any impact in the life time of the heaters when it is operated in longer duration under cold temperatures?

Hard to say. A lot depends on the design and quality of the heater. Usually the weak point is in whatever is being used to control the heater. Mechanical thermostats tend to be less reliable than solid state control but, like many things, it all depends.

SteveSh
  • 9,672
  • 2
  • 14
  • 31