When does epoxy potting compound cause stress-damage to SMT components.

[Psi]

The reason for this thread is to discuss SMT stress damage caused by potting compounds from thermal expansion and how to remove/mitigate the risks.

I'm trying to decide what potting compound to order to seal a small PCB (90x57mm) in a small box.
The board temp when operating is max 95C and for the environmental temp I would like to be ok down to -10C or at least -5C.

I have done some testing with epoxy potting compound and it is working well so far.
At 20C I can easily dent it with my fingernail so I don't think it's putting too much force on the SMT parts.
However I'm a little concerned it might put force on the parts if ambient temps get down to -5C or -10C,  since epoxy is harder at colder temps.

So I'm trying to figure out if epoxy should be ok or if I should go with a silicone potting compound to get more flexibility at colder temps.
The smallest parts on the PCB are 0603 resistors/caps if that matters?

Does anyone have any recommendations for good potting compound at a fair price which is ok for this sort of application and temp range?

Does anyone have any knowledge about how/what SMT parts get damaged from potting component stress and what to do to mitigate it?

Is there any 'rule of thumb' for when epoxy is out and silicone should be used?

Maybe Urethane potting compound is another option.

Thanks.

[Niklas]

The highest risk of damage is for the larger components where the potting will more area to apply pressure on during thermal cycling.

On a previous job we used a couple of different potting material for different generations of an automotive sensor. Typical ambient operating temperatures in the range of - 40'C to +85'C

For the oldest one they used a yellow epoxy that was relatively hard, but became slightly softer when heated. Cannot recall that I heard about any potting related failures on the components, but some customers opted for a thin barrier around the PCBA with softer potting.
There was also a small sensor cavity that was potted with a hard, black polyurethane.

Another generation of sensors used Dow Corning Dowsil 3-4241. Probably a bit more expensive, but it was softer and transparent enough to see through and also possible to probe through. This potting was preferred by one of the German customers.

The last generation used a polyurethane, don't remember exactly which one or the manufacturer. Soft as a rubber eraser and fully transparent.

Of these potting, it was only Dow Corning and the last polyurethane ones that had good adhesion to both the PCBA and connected wire harnesses. The yellow epoxy and the black polyurethane both allowed leakage of liquids by capillary force in the interface between the potting and the wire jacket. We also noted that PCBAs with the newer transparent thepolyurethane potting could fail when exposed for high temperatures and high humidity, but only if they were powered up. The root cause was related to chemical degradation of the potting that dissolved component terminals and, in some cases, also allowed dendrites to grow. Not all terminals were affected, there was a correlation between higher potential (5-28 V) and more dissolved material.

Have you considered to perform so simple thermal cycle testing? Two buckets with ice water and hot water could also be used to apply thermal shock on some samples. It is also possible to put the DUT in the freezer and oven for slower transitions.

[Psi]

Thanks for replying. i was worried the thread would get buried.

The highest risk of damage is for the larger components where the potting will more area to apply pressure on during thermal cycling.

My biggest part is two electrolytic caps. 8mm diameter case. They poke out half way above the potting compound.
I wonder if SMT vs TH caps would be better. Currently SMT.

For the oldest one they used a yellow epoxy that was relatively hard, but became slightly softer when heated. Cannot recall that I heard about any potting related failures on the components, but some customers opted for a thin barrier around the PCBA with softer potting.

That's good to hear. Yeah this clear epoxy gets soft when heater and harder when cold.

Have you considered to perform so simple thermal cycle testing? Two buckets with ice water and hot water could also be used to apply thermal shock on some samples. It is also possible to put the DUT in the freezer and oven for slower transitions.

Yes, i was pondering putting a unit in the freezer with some wires poking out the door seal and then cycling it on/off every hour or two to get some extreme thermal cycles. 
I wonder how long it would have to survive in that extreme state to be considered 'good enough'.
The product isn't something that will be used that often, maybe once a week if that so i guess i can expected extrapolate uses per year with freezer cycles with 5 or 10x safety margin.
Probably need to use production samples for this, I reckon my hand soldering is probably stronger than factory soldering.

The yellow epoxy and the black polyurethane both allowed leakage of liquids by capillary force in the interface between the potting and the wire jacket.

Not an issue for me as I don't have any wire coming out of the potting compound. (There's a 34p superseal connector on the PCB for I/O)


I'm not too worried about a very small number of failures. I just want to do all the things i can to avoid any nightmare scenario where they all die the same way a year later.  I think the freezer idea is probably the best plan.