Potting Electronics

[yramgu]

Hi everybody,

I need your help/advice on a problem we're facing at my company.

We've designed a product that works with a battery (primary cell) and that is potted after manufacturing/testing. It has radio connectivity and we have absolutely no physical access to it after potting. We've noticed (well the customer noticed :s) that some products were already dead right out of production.

These products were perfectly fine during testing, they failed after the potting process

I opened some with a dremel (lots of fun) and noticed that the batteries were dead. I managed to isolate it and plug in a power supply and noticed that these products had a permanent current consumption in the range of mAs when it is supposed to be in the range of µAs. That explains the dead batteries

So far we've tried:

- "Classic" X-ray
- Tomography
- Melting the resin with chemicals => kinda works when the resin peels off components go with it so that's not an option

We haven't seen anything obvious so we're stuck

Anyone had previous experiences with potting processes? Any advice on what to look out for? I'm currently looking into thermal expansion that happens when the resin hardens.

Any help would be greatly appreciated

[@rt]

I did a lot of potting and conformal coating,
but don't really understand the question.
Do you think the potting is conductive before it has set?
I couldn't really tell you that :O but if that's the case
you could insulate it prior, and probably wouldn't be asking.

[yramgu]

Do you think the potting is conductive before it has set?

It's not, we already have successfully produced more than 2000 of these products. 1800 have been installed and are functional. ~10pcs have this problem of dead battery. But we still need to identify what's happening

and heat released in the process is not a problem.

It's not the heat I'm worried about, it's the pressure it could cause against some components. I read that in this article:

http://blog.mtfwalker.com/?p=7

[DTJ]

Is the epoxy a low Thermal Coefficient of Expansion (TCE) type intended for potting electronics?

I guess nothing odd like adding 'carbon Black' pigment to the epoxy mix has happened???

Are they cast in a mold? How forceful is removal and are they all removed after the same amount of curing time?

Do you check current draw prior to potting? I've seen some low power designs where the quiescent current has increased by a factor of 100 by using semis from different manufacturers.

[StuUK]

Why not measure the draw during the entire potting process, it could be revealing...

[Koen]

What is the potting compound and what is its peak exotherm ?

[yramgu]

Thanks for all that feedback. I've attached the resin's datasheet

To answer some questions:

Is the epoxy a low Thermal Coefficient of Expansion (TCE) type intended for potting electronics? -- Yes

I guess nothing odd like adding 'carbon Black' pigment to the epoxy mix has happened??? -- No, it was black from the beginning

Are they cast in a mold? -- No, it is poured directly in the product's own enclosure

Do you check current draw prior to potting? -- Absolutely. We even check for various states (sleep, RF transmission, etc...)

Why not measure the draw during the entire potting process, it could be revealing... -- We thought about that but it's really not easy to do while you're manufacturing thousands of products. We're still trying to figure out a solution to do this. In the meantime we took 10 products in the lab and added wires that would give us access to current draw after potting. But I agree we need to be able to do this in the production process.

What is the potting compound and what is its peak exotherm? -- Datasheet joined. Problem is I'm really not familiar with this (yet, I just started documenting myself) so I don't really know how to interpret the data.

[Alex Nikitin]

I managed to isolate it and plug in a power supply and noticed that these products had a permanent current consumption in the range of mAs when it is supposed to be in the range of µAs. That explains the dead batteries.

I suppose it should be fairly easy to determine what part is actually consuming that much current on faulty units. As a working hypothesis I would suspect a static damage to some sensitive components during potting.

Cheers

Alex

[georges80]

Spraying the board with a conformal coating prior to potting could provide a datapoint on whether the problem is conductive/capacitive or more mechanical.

And yeah, verifying the handling during the potting process from an ESD perspective would be a good idea. There could be ESD issues as the potting compound transfers from the dispenser to the PCB...

cheers,
george.

[PChi]

Years ago I worked for a company that made encapsulated power supplies. They used epoxy but they found it was necessary to first coat the transformer lead out copper wires with silicone rubber otherwise after a while the wires fractured due to thermal expansion but you are suffering excessive current consumption. Are there any micro controller pins that can become open circuit and float causing higher current consumption?
You could try some DIY thermal cycling on a good unit to see if it's possible to induce failure.

[calexanian]

For many years I chased my tail on situations like this with epoxy potting for electronics. We abandoned it for battery operated products. Just too damn hard to QC. Some things I learned over the years. First you need an epoxy specifically for electronics and as a "Potting" type. Generally they will have some compliance to allow movement particullarly under thermal stress conditions. Microcontrollers will go to strange power consumption states when under physical pressures from curing epoxy. Ceramic capacitors can short or go open under the right conditions with epoxy. These problems appear to be made worse by a non replaceable primary battery.

We have switched primarily to Humiseal type products. They stay pliable and can be probed through for QC purposes. Since we did that our mystery failures have gone down significantly.   

[Koen]

They use polyurethane. It does not have the high peak exotherm and mechanical stress issues of epoxy.

[sprocket]

I don’t really have an answer for you question only speculations or theories I can’t back up with hard evidence.

I don’t like using hard resin for potting. It makes virtually impossible to do any sort of repair work, fault finding, repairing or things like that on the product, as you have found out. And to it also looks like resin shrinks a tiny fraction (using my completely unscientific eyeball measuring) when it have cooled down this might just come down to my perception and it’s not something we have actually measured as we haven’t had the need to do so, but if in fact that’s what it does, it could induce tension on components and the PCB

So what we sometime use when we need to encapsulate electronics to keep out moisture or mechanically stabilize components are magic-gel and magic-rubber from raytech (RS-online nr. 703-1231 and 703-1228). They do they trick for us, and we can actually get to the electronics if we need too.

Mind you, I work at a marine research faculty at a university here in Denmark, so I never work with bulk production, only prototypes, one-off’s and very low volume products for the researchers and scientists. So those sorts of products might not be viable in your case.

 Only place where we use hard resin (3M scotchcast 2131) is for submersible applications with connectors and cables.

[yramgu]

Interesting replies. So far from what I can see in the comments we could check for:

- mechanical stress
- ESD

They used epoxy but they found it was necessary to first coat the transformer lead out copper wires with silicone rubber

-- We also had the idea to use softer resins but they were much more expensive !

Are there any micro controller pins that can become open circuit and float causing higher current consumption?

-- Unfortunately yes. But since the resin is very hard, impossible to access anything after potting. I've had a lot of trouble just accessing the battery without damaging the board with a dremel

You could try some DIY thermal cycling on a good unit to see if it's possible to induce failure.

-- Very good idea! we have a climatic chamber where I can do this.

For many years I chased my tail on situations like this with epoxy potting for electronics. We abandoned it for battery operated products. Just too damn hard to QC.

-- I agree so much !

They use polyurethane. It does not have the high peak exotherm and mechanical stress issues of epoxy.

-- I got mixed up, sorry The datasheet I joined is for a new resin we're gonna test to replace the current one. The current one is epoxy-based but the datasheet is in french.

I don’t really have an answer for you question only speculations or theories I can’t back up with hard evidence.

I don’t like using hard resin for potting. It makes virtually impossible to do any sort of repair work, fault finding, repairing or things like that on the product, as you have found out. And to it also looks like resin shrinks a tiny fraction (using my completely unscientific eyeball measuring) when it have cooled down this might just come down to my perception and it’s not something we have actually measured as we haven’t had the need to do so, but if in fact that’s what it does, it could induce tension on components and the PCB

-- I'm looking for speculations Since we heard of possible mechanical stress from several sources, I think we're going to contact labs to try to record (with a strain gauge for example) the differences in pressure during the curing process. Even if it doesn't lead anywhere, it could be an interesting exepriment!