19
\$\begingroup\$

Our vendor pulled some PCBA out of storage, been sitting for about a year.

They found about 5% delaminated. They claim it was 'because of the storage duration', but I'm not buying that. I think these boards were defective from day one.

Yes, I know a hatched pattern for the pours is better - These aren't my design.

Anybody think this is a "storage" issue? They're (supposed to be) FR4, so presumably they are fiberglass resin, and shouldn't be hygroscopic. Vendor is in south-east China (Shen Zhen) where the humidity and temperatures are "jungle like" most of the year. I'm sure there was no climate control during storage, I wouldn't be surprised if they were subjected to months of 130F+ temperatures. Maybe storage EXPOSED the issue, but I don't like the theory it CAUSED the issue.

In field, these boards will get quite hot during normal operation. There's a number of high-power MOSFETs on the other side, and the board regularly will see 15A, sometimes much higher during surges (about 35A max or so)

Further, the vendor is suggesting subjecting the entire inventory to a heat-test, basically a shorter version of a reflow oven cycle, to ferret out the bad ones. They're suggesting use some kind of profilometer to detect the 'bubbles'.

This is a 4-layer PCB. We almost "never" use multilayer boards, and I doubt our vendor sees them often either (maybe "never").

This is also a 'safety critical' component.

Any and all comments & suggestions welcome!!!

\$\endgroup\$
  • 12
    \$\begingroup\$ Has your vendor got a recognized quality system and do they have quality accreditation by any well-known bodies? If not, then it's likely the tip of an iceberg. \$\endgroup\$ – Andy aka Aug 19 at 17:37
  • 25
    \$\begingroup\$ If it's safety critical I wouldn't hesitate to toss the whole batch, then order new ones from a supplier with an established track record and quality system. \$\endgroup\$ – John D Aug 19 at 18:55
  • 8
    \$\begingroup\$ Regardless of whether or not the boards were damaged by storage or not, they're evidently not suitable for a safety-critical application, so either way they should be dumped. Whether the vendor should also be dumped is a different question, but the answer there is also pretty simple... if the delamination is caused by manufacture then yes; if the delamination is caused by improper storage then also yes. \$\endgroup\$ – Ian Kemp Aug 20 at 11:45
  • 25
    \$\begingroup\$ I've worked with many Chinese manufacturers and when you catch them on stuff like this they will say literally anything and even fabricate evidence to keep your business. The only "proof" you need to move your business elsewhere is the fact that the PCBs are de-laminating. Well-made PCBs would not do this under any conditions. \$\endgroup\$ – Geoff Griswald Aug 20 at 12:24
  • 5
    \$\begingroup\$ Safety critical hardware must usually be qualified through tests, including environmental and ageing tests. It's likely that these boards will fail at an unacceptable rate. The earlier in the design and production process they are replaced with quality material the cheaper in the end. This is independent of the reasons for their condition and independent of any possible refunds. \$\endgroup\$ – Peter - Reinstate Monica Aug 20 at 14:51
42
\$\begingroup\$

The fact that they delaminated in only a year calls into question the quality of the workmanship, regardless of the fact that they were in storage. Like you said, if they delaminate in storage they most certainly would delaminate in the field, and in a safety-critical application this is absolutely unacceptable. I would agree with John D that these should be scrapped (depending on the cost, you may be able to work out a partial refund from the supplier) and order new ones from a reputable source. Also, if you have any authority to do so, I recommend pushing a redesign to hatch the copper pours.

| improve this answer | |
\$\endgroup\$
  • 22
    \$\begingroup\$ +1 If (as the vendor claims) the boards delaminated in a year just sitting somewhere, with no local resistive heating, no vibrations or mechanical stresses to speak of, this batch would have been a total disaster in the field. \$\endgroup\$ – TooTea Aug 20 at 8:42
  • 2
    \$\begingroup\$ You don’t need to hatch the planes in a modern manufacturing process. \$\endgroup\$ – user110971 Aug 21 at 6:22
  • 1
    \$\begingroup\$ @user110971 I disagree. There are a number of reasons why hatching is beneficial in with modern boards. The biggest one is if one side of your board has more copper than the other, then in any case other than a tightly-controlled thermal chamber, the temperature cycling will cause uneven heating of one side of the board or the other. Over the long term, this will cause repeated warping and straightening of the PCB which can lead to delamination. Hatched pours helps reduce the effect of the "heat sink" behavior of the copper pours. \$\endgroup\$ – DerStrom8 Aug 21 at 15:01
  • \$\begingroup\$ @DerStrom8 My experience is with boards in machine tools. In a machine tool the temperature gradient is typically at least 30 degrees C every 24 hours. The boards last for 10+ years without any warping. A new product is also tested at 50 degrees C temperature gradient for a 1000 cycles. \$\endgroup\$ – user110971 Aug 21 at 15:13
  • 1
    \$\begingroup\$ @user110971 Allow me to clarify. I am not saying that a board will delaminate due to temperature fluctuations if you don't use a hatched pour. I'm saying it could, and in order to minimize the chance a hatched pour could be beneficial. Realistically speaking, any decent board house will be able to manufacture a board that won't delaminate during normal use. Period. However, as a PCB designer who constantly has boards going out into the geotechnical field, I strive to take every possible precaution I can to make sure my boards won't fail in the field. \$\endgroup\$ – DerStrom8 Aug 21 at 15:19
23
\$\begingroup\$

If the storage conditions were relatively benign (eg. not cycling back in forth between temperature extremes in an environmental chamber many times), looks like their process was defective, maybe improper process control in their lamination press. Which means that the entire panel that was in the press on that run would be scrap and maybe (probably) every panel made around that time.

Given the low cost of 4-layer boards, you should scrap them and move on. Even if it was not safety-critical the cost of field failures is too high.

There are hundreds of PCB suppliers in the SZ Bay area, you should be able to find a better one across the street.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ PCBs should still survive extreme temperature cycling in an environmental chamber (at least within the bounds of normal storage, transportation, and operation conditions). That's how you prove they won't fall apart in the field or in storage. If you're building a device that has to be ok for -30C transport or +60C service outdoors the PCB also needs to survive those conditions. \$\endgroup\$ – J... Aug 22 at 15:19
6
\$\begingroup\$

Definitely toss them. 4 layer boards even made domestically are inexpensive these days.

FR-4 will absorb moisture, and when laminated at high temperature and pressure the moisture turns to steam and creates pockets in the epoxy/glass substrate. My bet is these were already delaminated when they shipped them.

If you are even considering using them, even if you have a few that do not look delaminated, I would bake them at 250 degrees F for a minimum of 4 hours. Then reinspect them.

If you run them through reflow or wave and there is moisture in them it can be very ugly.

| improve this answer | |
\$\endgroup\$
2
\$\begingroup\$

I'm going to agree with what seems to be everyone else here: scrap them and try to get a refund. Also stop ordering from this supplier, you shouldn't need to worry about this in your supply chain.

Even if their story checks out and this is caused by storage within a year, that's going to happen again on other units. Maybe if you told its for a household low power toy where margin is tiny and stakes are low, but all of those aren't the case and this is a no-go in your stated application.

| improve this answer | |
\$\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.