3
\$\begingroup\$

I'm experiencing a strange failure of electronics after potting which I'm hoping to get some more insight and explanation to. The device is a custom, rugged USB drive in an Aluminium/Titanium body (see www.viud.net for pictures and more details). I'm potting the USB board (a variety of off-the-shelf boards) in Aremco 526N epoxy and curing in the oven. The USB drives are verified to work fine before potting but after potting I'm getting a high failure rate (17% Aluminum body, 78% Titanium body).

Possible explanations for the failures I've considered are:

  1. Heat - Since I'm curing the devices in the oven I thought that maybe they were getting too hot. The first batch was cured at 90C for 2 hours but after testing I found the oven does get considerably hotter than that. The next batch I did at 55F for 4 hours and confirmed that nothing gets hotter than 80C but the failure rate didn't change.
  2. Epoxy - The Aremco 526N is not technically advertised as a "potting" epoxy although its electrical properties are better or similar to other potting epoxies I've used with no issues.
  3. USB Drive - The failures occurred with 4 different USB boards so it is unlikely related to that.

None of these possibilities explain why the Titanium models have a much higher failure rate than the Aluminium. If it was heat related I would expect the Aluminium models to have a higher failure rate anyways (thermal conductivity of the Al is x36 greater than the Ti).

The only possible cause of the failure I have left is related to internal pressure of the epoxy while it is curing in the oven. The epoxy is essentially sealed completely into the body with only a very small escape through/around the USB connector. Since the epoxy around that area likely sets first the rest of the epoxy in the body is likely under some pressure due to thermal expansion. This also explains why the Titanium models fail more than the Aluminium: the Al conducts the heat faster through the body which means the epoxy sets more evenly reducing the internal pressure.

Unfortunately, I have no way of verifying this theory as to open the drives I have to destroy the USB. I also cannot believe there is enough internal pressure from the epoxy to actually damage the circuit board. I can try curing the epoxy at only room temperature but it then takes weeks to set and the epoxy is not as strong.

Any ideas, theories or comments on my issue are more than welcome.

\$\endgroup\$
  • 1
    \$\begingroup\$ As a WAG, how about too much pressure inside. The Titanium is stronger and doesn't give as much as the epoxy expands. I think sometimes people put a pliable coating over the electronics first. But I've never done any potting of electronics. \$\endgroup\$ – George Herold Sep 4 '14 at 14:10
  • \$\begingroup\$ I was thinking of trying a conformal coating of the board before potting but have never done it and know nothing about it really. Looking at the USB board the memory chip does have space under it which might let it crack if there is too much pressure on it from the outside. \$\endgroup\$ – uesp Sep 4 '14 at 14:17
  • \$\begingroup\$ I've only expoxied two drives over the years with the basic hardware store epoxy, but both failed before their time. So the theories above do seem correct. \$\endgroup\$ – grabbers Apr 6 '17 at 16:56
5
\$\begingroup\$

After a good deal of testing I believe the problem is actually a combination of three things:

  • Temperature
  • Sensitive USB flash drives
  • Air pockets

On the initial batch the temperature was definitely a little too high which may have contributed or exacerbated the issue.

I've used several different models of USB flash drives and one in particular seems extremely sensitive to temperature and shock.

However, it seems the root cause of the issue is large air pockets. If any significant portion of the circuit board is not covered by epoxy it seems to fail very easily. For example, one unit failed after just falling 1 meter to the floor and was confirmed to have a large air pocket at the top of the flash drive. Another unit test with no epoxy at all similarly failed after a low speed drop test.

I've changed my potting procedure to try and eliminate air pockets and ensure the entire flash drive is covered. This is a little tricky as the design is completely enclosed and there is no (easy) way of confirming how much or little air is left inside. But recent tests seem to confirm that careful assembly to minimize air pockets leads to a more reliable product with no premature failures.

\$\endgroup\$
  • 1
    \$\begingroup\$ Way to return with results! +1 \$\endgroup\$ – ACD Nov 5 '14 at 15:57
  • 1
    \$\begingroup\$ If you're getting air-inclusion issues, you should really look into vacuum potting. It basically entirely fixes the airspace issues. \$\endgroup\$ – Connor Wolf Nov 5 '14 at 18:17
3
\$\begingroup\$

At an internship quite awhile ago, we would pot connectors with some kind of black two part potting compound. We would outgas the epoxy in a vacuum chamber, the vacuum doesn't have to very good. Maybe there is some trapped gas causing your problems.

\$\endgroup\$
  • \$\begingroup\$ While I don't think there is trapped gas I'm willing to entertain any possibility. I don't have definitive proof (only a few samples) but it seems there is more failure rate when there is more epoxy in the case that leaks out during curing. \$\endgroup\$ – uesp Sep 4 '14 at 15:25
2
\$\begingroup\$

At a 78% failure rate, I don't understand your hesitation to open a drive and see what's going on. For two-hour soaks, I wouldn't think that the thermal conductivity of the case has anything to do with things -- though maybe thermal expansion does, creating an interesting shear on your PCB. No way of knowing without getting in there. Can you xray through the case??

\$\endgroup\$
  • \$\begingroup\$ I'd like to keep the cases as these were a pre-production run and they are expensive. I actually just managed to open one from the last failure while keeping most of the USB board intact. There were no signs of any of the ICBs cracking or other obvious damage. \$\endgroup\$ – uesp Sep 4 '14 at 15:21
  • \$\begingroup\$ @uesp -- If, as you describe in your blog, the epoxy expands as it cures, there's your answer right there. Dollars to donuts, it's causing shear of the circuit elements with respect to the PCB. I'd venture that expanding epoxies are not suitable potting agents \$\endgroup\$ – Scott Seidman Sep 4 '14 at 18:50
0
\$\begingroup\$

According to the datasheet for that range of epoxies, the 526N you're using has among the highest cure shrinkage in the range.

0.01 in/in may not sound like much, but its more than enough to, at the very least, cause hairline fractures in joints on the PCB. I have experienced this exact phenomenon when potting accelerometer sensor PCBs into an aluminum enclosure.

It you're set on using that epoxy, you could give the PCBs a conformal coating of some sort before potting - but make sure its flexible. I used a thin layer of low viscosity RTV silicone (Dow Corning 3140) and my failure rate dropped from over 80% to well under 1%.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.