Potting horror...

er hammer!"

Hi Nico,

Looking to your profile, you are in NL? Did you work with Purtec?

When the PU company did a good job, and you did a good design job with sufficient margin, I would first look to contaminants on the board that may influence the curing of the PU or even lead to gas production (results in high stress). Even residues of adhesive tape can cause problems.

Cut one PCB and look for the uniformity of the PU. For the other boards, run a long test at elevated temperature and see what happens over time. It is possible that other board will fail also.

To check proper functioning (of the clock oscillator circuitry), you might run a temperature sweep. When you do not have sufficient margin in the oscillator, it will fail at low or high temperature. You may also check for slight frequency drift after potting (of course this requires to measure the frequency before potting).

Though shrinkage may be low, you will always have some mechanical stress due to shrinkage.

Best regards,

Wim PA3DJS

In article , snipped-for-privacy@highNOTlandTHIStechnologyPART.com says...>

So does DimBulb.

The *one* time I've had my designs potted it was for security reasons. It was an *expen$ive* PITA.

DimBulb is a tech, certainly not a degreed engineer (you've noted the phobia of all things math). Nothing wrong with being a tech, though not nearly as fun. Notice how Dimmie always coaches his words. "We make" or "our product", never "my design".

Why does Dimmie think cardboard can't be ESD treated? Many parts come in card bard containers.

Oohh. Our purchasing people would croak.

Ours *did* have a standardized part numbering system with the value as the last digits. From what I'm told, they went to a new release system that wanted to sequentially assign part numbers, though any string *could* be assigned. Whoever put the software in set it up so the old P/Ns were preserved but new ones were sequential. What a PITA. They *HAD* it right, then blew it. I'm sure the owner has no idea how much that decision cost him.

This is a big accomplishment. Can you share your experience in more details; i.e. what softwares did you use; how do you link the sch/pcb databases with the inventory; how do you link the BOMs to the specific vendors, who is responsible for keeping the system and how many persons are involved. Do you have the mechanical parts in the same database with the electronic components?

My biggest frustration with the systems like that in the past was because of the hell of bureaucracy, inconvenience and associated overhead. So people tend to fool the system and do the things in their own old ways, which makes the system worseless...

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

It's our own thing, a single PowerBasic program, 17k lines including comments and whitespace. There's a master parts file of fixed-length binary records, a folder full of text-mode BOMs, and a folder that contains folders of miscellaneous data, like datasheets, photos, notes, one sub-folder per part.

At startup, it reads the entire parts file into memory arrays, and indexes the BOMs and data folders, which takes a few seconds. After that, it's blindingly fast. Most commercial packages seem to build on top of some hairy general-purpose database manager, and then do everything over the network, and get slow.

A part number is of the form nnn-mmmm, where nnn is a category, like RES 0805 or some such. The mmmm is encoded somehow, like by resistance in this case.

This system was designed by engineering and manufacturing people to do exactly what we want. It works, like its predecessor DOS version (which used 5-digit numbers that eventually got to be sort of random) because the people make it work, and follow the rules, which are all available in writing.

We may decide to sell it. People wouldn't be buying just the software, but the whole set of procedures and concepts, which might be very healthy for some companies without that discipline. Some people really don't understand the whole thing: dash numbers, assemblies and subassemblies, revision control, ECOs, BOMs, parts numbering, all that.

John

I would if there was a gap in the same place in all the units. Stranger things have happened.

Good call. Check wave form shape AND amplitude. I had a similar problem with a product 15 years ago. Changed units over to an oscillator module and that was that. The can of the module became the equivalent of a bubble around the crystal and caps.

Ah, another student of Phil Allison.

Off to the killfile for you dipshit.

Good question. I'll check that with the company that did the potting.

Sounds reasonable. What is the problem with no-clean flux and potting?

Its black.

I doubt it. I already checked that with the one I milled open.

Potting it is an affordable way to get a slick looking custom housing.

Potting is better than air for heat conduction, so I doubt that's it.

Potting can tear parts from the board as it cures. Probably not likely, though.

You mentioned a crystal. The potting compound will change the dielectric constant which could alter the capacitance around the crystal. This can also affect high-speed lines. If you have fluorinert, you can try dunking the board and see if you can reproduce the problem. Mineral oil will work, but is messy. If it is repeatable, you can partially dunk the board to see which area is sensitive. I have come across this sort of problem when operating electronics in castor oil. We had a high-speed diff line that was marginal.

Forget that companies background; i would suspect that kind of potting material. Many polyurethane potting compounds "foam up" as they harden; that foaming can cause a *lot* of static and so zap a lot of parts.

Check, double check and even triple check!

The "no clean" stuff leaves a sort of plastic stuff that is supposed to trap the salts etc from the actual flux. The potting compound eats into this plastic and frees up the salts.

a solid material. I am not sure about the details of how this leads to tr= ouble but I think that it provides a surface that has uncured material on i= t long after most of the chemical reaction is done.

"affordable" may not turn out to be true in this case.

er hammer!"

------------------------------------------------------------

The problem with no-clean and potting is that the potting compound does NOT adhere to the flux locations, making the potting benefit moot.

OIt also inhibits the cure of some compounds.

Since most potting is fairly chemically neutral when cured, I doubt seriously that his claim of a chemical action taking place is valid... at all.

I find it more and more difficult to put ANY credence in any of the MooseFart's spew.

That should read "Hard potting compounds can tear..."

Elastic potting compounds rarely stress parts.

He has lots of names. Archimedes'Boytoy. MiniProng. AlwaysWrong. Scatman.

John

Many polyurethane potting compounds have no foaming issues whatsoever.

If you experienced one that did, that was your experience with ONE material. Your blanket remark points to a lack of experience, since it is not true.

Potting compounds are specifically made to be potted into locations using very low pressure to remove all air and chemical gasification bubbles. Even a full (near full) vacuum.

It is not a good compound if it exhibits gas while curing. If it foams, it is a media that was specifically meant to foam... pretty much.

"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...

It would be Interesting if a malfunctioning board could be x-rayed. Possibly by a board house that is setup for BGA stuff. I'm not sure if it would show anything. Un cured potting may be the same density as the cured stuff. Cutting and digging may be the only solution to find the problem.

Cheers

The foaming action is often the result of humidity or water. You are best to work in a dry environment. If you use the bipack method of mixing, and start with fairly pure materials and keep the water out, foaming in not going to be an issue.

Mixed in with other "turd-rated"...

Author: makenoattempttoadjustyourset | itsasecretdummy | higgsfield | archimedes' lever | snipped-for-privacy@infiniteseries.org | valleygirl | snipped-for-privacy@likeiwouldgiveit.comeon | ultimatepatriot | itissoeasyacavemancandoit | fatbytestard | stickthatinyourpipeandsmokeit | the great attractor | massiveprong | the last mimsy | bungalow bill | klipstein | tutamongus | spurious response | falk willberg | dorothy with the red shoes | do i really need to say

...Jim Thompson