Potting horror...

Capacitance changes?

How do you know the MCU stays cool when it is potted? You milled away the compound and it stays cool 'now'. Does the unit still lock up that you cleared the compound away from the MCU?

Could the compound have flowed somewhere it wasn't suppose to and it is now mechanically causing an intermittent connection?

Also, how is the clock generated?

Repeat the process, retested at various stages, including shipping, to ensure that potting really is the significant factor.

Right now you seem to have only verbal assurances that proper processes were used on the first 4.

Any open magnetics, resonators or electromechanical components on-board?

RL

I only milled away a small part of the compound (just enough to reach the 48 pin package and get to the pins) and covered the hole while testing. The MCU is not consuming a lot of power.

Yes, the unit still bahaves odd.

That could be the case, but that wouldn't explain odd behaviour on multiple modules.

A 10MHz crystal in a HC49 SMD metal case with the usual capacitors (around 20pf). I also measured the crystal's frequency on a potted and non-potted PCB, the readings where correct. I didn't measure the amplitude yet.

Potting is always a horror. It's expensive, messy, often thermally horrible, can wreck parts, makes debugging almost impossible. Some potting compounds exotherm big time, or shrink and crush parts. Even when it seems like a good idea, it's usually not.

John

Is it possible some unused pins on the MCU or other logic are not terminated properly (pulled up/down) and now you have a big slab of ESD producing plastic?

Just wondering if the potting compound might change the waveform.

I assume the boards all tested ok and worked properly before potting. Were all functions tested? Is there an unpotted working control board? How were the boards cleaned and dried before potting? Was the urethane properly and sufficiently mixed? Was the urethane thermally cured, what temp? Could there be any other surface contamination, flux or other contaminants? Could there be tin whiskers from unleaded solder propagating along the urethane-board interface. Could there be other creepage issues within the urethane-board interface such as moisture, etc. Are you sure about the urethane, could it be a thermal but poorly insulating compound, aluminum filled for example. Was the potting vacuumed or could there be bubbles on internal surfaces with possible condensation or other contamination.

Many questions here but one main one is the proper operation of a working unpotted control board to insure that problem is in the potting and not just the circuit or software such as open pins, etc.

Was this a two part polyurethane that has big trouble if there is any water about? I assume it is.

Never apply power to the board until the reaction is totally finished.

Did someone make the mistake of using "no clean" flux? If they did abandon all hope of potting the boards in question. Make some more with washed flux and wash them. Beware of the water from washing them.

Is this polyurethane colored or is it the sort of clear stuff? The sort of clear works better than the stuff with pigments.

Is the failure heat related?

What silly fool decided to pot them? Perhaps you can delete the potting and et on with production.

er hammer!"

did the boards work before potting did you unpot one of the bad boards did it start to work after unpotting

what speed is the uc what is the dielectric constant of the potting

are there ANY hi impedance nodes anywhere in the system including power supply, reset circuit, or anywhere else

what about the busses, do they go hi Z sometimes

was the potting fully cured and dry before powering up

let us know what you find

Mark

Boards were dirty or contaminated. The potting probably never fully cured. That?s really the first place to look.

Cheers

If it is an RoHS process, I'd look for tin whiskers that propagate through potting easily. However...

Sounds like some capacitances have changed though is really likely all it is. Some ringing. Terminate the memory lines, even though most chips already do.

I think his determination was good enough. He could use IR devices to prove a rise or not. It likely doesn't get hot in open air either, so why would it potted?

He was probably more definitive with it than YOU are... now.

Probably.

That sounds like a solder flaw or design flaw. There are NO places on ANY circuits where the potting cannot go. An HV circuit gets fully potted, and a vacuum gets utilized, so the potting goes EVERYWHERE.

SOME problems MAY arise with EL caps (low likelihood, but other than that, no. Not in a proper solder process.

You really are dumb, John.

The potting materials discussed here have elasticity, not hard epoxy crush capacity, like 'stycast' has.

Properly chosen media, and properly utilized, potting has a VERY SPECIFIC place in the electronics industry.

Sad that you are disconnected from the facts to the point that your opinions make you a really bad 'engineer'.

That's OK. You only recently got out of plain cardboard stock bins. Sad. Do you think there is no use for ESD precautions as well?

It was professionally potted, idiot. The contractor would not have delivered the finished assemblies with full curing.

Read his post, idiot!

That's a good suggestion. We've received boards from outside assemblers that would drip water if you shake them.

It's hard to find assemblers nowadays who use solvent wash. And sometimes they use dirty water, or don't completely dry the boards, especially under parts. Potting will lock in any wet crud, and electrolytic corrosion will eventually mess things up.

John

At least I have a name.

Sometimes. But more often it's a bad idea, often driven by paranoia.

I'm a pretty good, degreed engineer, without quote marks. I've posted lots of links to, and pics of, my stuff, so people can judge for themselves.

You never post any. You don't even have a name. I think you're a frustrated tech.

Why sad? We made a ton of money out of the parts that were stored in those bins. Several tons, actually.

And our ESD controls are better than they need to be. No problems at all.

The reshuffle of the stock room was driven by new inventory/BOM software and a total renumbering of all the parts in stock. It all worked great... 4800 parts, about 1.6 million pieces, relabeled and rearranged on the shelves, about 600 BOMs converted to the new number system, all the computers changed to use the new system, all done in two days. We love it.

All the parts are now described in a standardized way, and all part numbers are logical. For example, all resistors are now in order by both part number and resistance. Datasheets and notes/pictures are linked to part records, and you can instantly see all the places where a part is used. And you can pull up a BOM and see the total parts cost, instantly. Searching is so fast you usually can't see any delay at all.

John