Designing for low temperature: Must-knows, Recommendations, Tips?

[igendel]

Hi all,

I recently designed a PCB for a relatively simple system (an 8-bit MCU @16MHz, some low-speed communications over RS485, a Flash memory IC, stuff like that). It's supposed to operate in cabinets which are located outside, so it's protected from water/dust but not from temperature. Now the client suggested using this system in a cold country (think down to -30C in winter), which was not part of the original requirements.

I started checking the components I use, so far - except for the crystal oscillators which have low-temp equivalents - they are all rated -40C to 85C for normal operation, so they should be ok. But what else should I be looking at? PCB material? Specific conformal coating? Cold solder joints ?

Thanks!

[Messtechniker]

Better heat it to prevent condensation problems
(unless it is filled with dry nitrogen).

[beanflying]

If you are running on batteries triple check their spec some will lose a huge lot of any performance at that sort of temp. NiMh start to go off near Zero C if you are trying to draw any sort of power let alone -30.

On mains or a power supply then a simple heater system would make a lot of sense. I have been playing with aluminium cases and peltier modules. Mount an RTD inside with your gear and all sorted.

[Bashstreet]

Better heat it to prevent condensation problems
(unless it is filled with dry nitrogen).

That will not be necessary but he might have to make design modifications.

OP it is best consult a company or organization that specializes in cold environment electronics or you will have quite few pitfalls. (Such information or design help wont be free)
As i originate from such "cold" climate country that -30 even -50 temperatures are possible such information is quite widely shared in engineer circles but for you might be best find consultant.

Condensation at itself will be least of your worries as humidity will be almost nonexistent although relative can be very high. (I assume you know about how temperature effects relative and absolute humidity)

[fcb]

Test. Test and test again.

Watch electrolytic capacitors, batteries, relays.

Depending on the sealing of your casing - humidity/condensation could be a huge problem.  You can conformally coat some stuff, but it isn't a panacea. Parylene might be worth looking at though.

When we started testing our clients products, we re-purposed a domestic chest freezer - we could get that down to -30C. Pretty soon we invested in proper thermal chambers (we can go down to -65C now).

Also, understand the effect of thermal cycling, e.g. your product might live at -30C then it activates and warms up to +10C, then deactivates, etc...  If you can afford the power - look at adding a heater & insulation.

Personally I'd recommend testing in-house as soon as you possibly can, and way before you fix your design. It amazes me the number of people that test (thermal, vibration, EMC, etc...) AFTER their design is complete.

[igendel]

ok, first of all I forgot to mention that the system will be powered from mains, and that it has no moving parts; however there's a backup battery (CR2032) for a Real Time Clock IC. And one electrolytic capacitor, though it doesn't really have to be, for what it does I can use ceramic.

The system is also always on (except for the occasional powerout), no sleep modes, high currents or anything that should make the components temperature change fast.

I'll talk to the client about warming the cabinet, though I'm not sure he'll like that idea

Thank you all for the input, I imagined it won't be very easy... but if you have any more suggestions or experiences to share I'll appreciate it.

[Bashstreet]

ok, first of all I forgot to mention that the system will be powered from mains, and that it has no moving parts; however there's a backup battery (CR2032) for a Real Time Clock IC. And one electrolytic capacitor, though it doesn't really have to be, for what it does I can use ceramic.

The system is also always on (except for the occasional powerout), no sleep modes, high currents or anything that should make the components temperature change fast.

I'll talk to the client about warming the cabinet, though I'm not sure he'll like that idea

Thank you all for the input, I imagined it won't be very easy... but if you have any more suggestions or experiences to share I'll appreciate it.

Take my advise you will not need to "warm" the cabinet.

[BrianHG]

Relying on warming means loosing the warming during power failures.

Also take a look into conformal coating sprays designed for low temperature and to protect against humidity.
Also with the clock battery, make sure it is soldered, not in a battery holder which is a failure point in wide temperature variations.  Or at least, a battery holder really suited structurally to handle with wide temperature deviations and conformal coating.
Make sure the case is water tight.
Also look into what automotive manufacturers use in their designs to protect against the cold and wet/humidity.

[igendel]

Take my advise you will not need to "warm" the cabinet.

Probably so, but probably also way out of the client's current budget, and obviously less fun 

Relying on warming means loosing the warming during power failures.

Also take a look into conformal coating sprays designed for low temperature and to protect against humidity.
Also with the clock battery, make sure it is soldered, not in a battery holder which is a failure point in wide temperature variations.  Or at least, a battery holder really suited structurally to handle with wide temperature deviations and conformal coating.
Make sure the case is water tight.
Also look into what automotive manufacturers use in their designs to protect against the cold and wet/humidity.

That makes lot of sense, thanks!

[Bashstreet]

Take my advise you will not need to "warm" the cabinet.

Probably so, but probably also way out of the client's current budget, and obviously less fun 

Relying on warming means loosing the warming during power failures.

Also take a look into conformal coating sprays designed for low temperature and to protect against humidity.
Also with the clock battery, make sure it is soldered, not in a battery holder which is a failure point in wide temperature variations.  Or at least, a battery holder really suited structurally to handle with wide temperature deviations and conformal coating.
Make sure the case is water tight.
Also look into what automotive manufacturers use in their designs to protect against the cold and wet/humidity.

That makes lot of sense, thanks!

It would be quite a notion in my country of origin if all electrical and electronics equipment would have to have "warmed cabinets" 

[igendel]

It would be quite a notion in my country of origin if all electrical and electronics equipment would have to have "warmed cabinets" 

Well, a solution is a solution...  Also, I bet the local wildlife will appreciate the heat! That can have some unexpected consequences...

[Bashstreet]

Any case you should consider using inert as possible materials also temperature (cold) do not as such increase corrosive effects.

The most important aspect is to consider effects of thermal changes.

Placement of equipment (placing equipment in area that will have (sun) can lead to 10-25 (If not higher) decree daily variation causing physical stress on board solder joints connectors etc.

Also you will want to encase the equipment in such way that it will not be exposed to dust moisture etc.

Consider best surface coating of connectors etc.

Component operating temperatures.

The challenges of protecting electronics from corrosion are much the same in colder and warmer climates and protecting in colder climates are generally easier.

[Bashstreet]

It would be quite a notion in my country of origin if all electrical and electronics equipment would have to have "warmed cabinets" 

Well, a solution is a solution...  Also, I bet the local wildlife will appreciate the heat! That can have some unexpected consequences...

Yep lil nice cabinets for the mice

[David Hess]

When I have done this, we used an environment chamber cooled with liquid nitrogen making it very easy to test to much lower temperatures.  I completely avoided aluminum electrolytic capacitors and used solid tantalum capacitors instead.  Analog circuit operation which included bipolar ICs, plastic packages, and surface mount construction worked fine down to liquid nitrogen temperatures.

The largest problem we encountered was contamination through the conformal coating in the field but that was beyond my purview.  Surface adhesion of the conformal coating was a problem.  Now I might apply the conformal coating under a vacuum.

Digital logic should not be a problem but beware of clock oscillators not starting at low temperature despite their specifications.  A discrete transistor oscillator with bias compensation should be more reliable than an integrated oscillator if component variation is taken into account.  Watch out for manufacturing process and lot changes.  Thermal cycling should be done to destruction if possible to find weak aspects of the design.

When I was first using the environment chamber, my boss walked in and asked how low the circuit would function.  My response was, "I do not know.  The thermometer you gave me only reads down to -299F."

[Niklas]

Forward voltage drop of diodes usually increases when the temperature drops. That could be an issue for circuitry powered by a battery, for instance cold crank of a car where minimum battery voltage and max diode drop coincides. Some manufacturers, for instance Vishay, do not even show the graphs below room temp.

Current gain of bipolar transistors drops. Vgs(on) of MOSFETs change.

Thermal expansion of all parts can be an issue over time. Have a look at connector and other parts that are connected at two points. For instance a connector that is soldered both on the PCBA and fixed to the housing. Possible source for fatigue cracks on solder joints.

Our customers (automotive OEMs) typically demand conformal coating or similar protection. So far I have only seen specs for what is acceptable at high temp related of thermal expansion of the coating. We are using a soft potting or overmolding instead. The thermal expansion of the flux residues can peel off/delaminate the conformal coating.

Check the temp specs of connectors. Many of the small JST type connectors are only rated down to -25'C.

We have not used electrolytic caps due to space constraints, but apart from that we are using most common component types.

Don't rely on heating or try to find the perfect sealing, i.e. trying to work against mother nature. You can have a power failure and the ideal gas law (p*V=n*R*T) can create differential pressure that defeats sealings over time.

[jbb]

...a backup battery (CR2032) for a Real Time Clock IC.

Coin cells can be a problem at low temperatures. You’ll find that their open circuit voltage drops and their internal resistance goes up a lot, especially towards the end of discharge.

To control this I suggest you make sure there’s as little load on it as possible, ideally just an RTC chip. Be aware that some of the temperature compensated ones draw current pulses from time to time to do temperature measurement. You should check

Also, the datshetts for some coin cells are ... optimistic. I suggest you test several of them for yourself at the right temperature.

Finally, you might want to check out the chemistries available. The lithium carbon monofluoride might be better at low temperature (can’t remember exactly). A suitable lithium thionyl chloride would certainly be possible, but they are more hazardous than standard coin cells and require special disposal.

[viperidae]

...a backup battery (CR2032) for a Real Time Clock IC.

Coin cells can be a problem at low temperatures. You’ll find that their open circuit voltage drops and their internal resistance goes up a lot, especially towards the end of discharge.

To control this I suggest you make sure there’s as little load on it as possible, ideally just an RTC chip. Be aware that some of the temperature compensated ones draw current pulses from time to time to do temperature measurement. You should check

Also, the datshetts for some coin cells are ... optimistic. I suggest you test several of them for yourself at the right temperature.

Finally, you might want to check out the chemistries available. The lithium carbon monofluoride might be better at low temperature (can’t remember exactly). A suitable lithium thionyl chloride would certainly be possible, but they are more hazardous than standard coin cells and require special disposal.

I just found a data sheet for a CR2032 that says leakage may occur below -20 due to the plastic seals deforming. They're rated -20 to +70
The Energizer cells are rated  -30 to +60

4.2.2 Storage at less than -20? can deform the plastic parts and may cause a leakage.

http://www.farnell.com/datasheets/1496885.pdf