General > General Technical Chat
Bizarre and Unexplainable High-Impedance "Shorts" Across PCB
(1/3) > >>
tom66:
Has anyone seen this behaviour before?

We had a number of high-density boards back from a UK manufacturer (PCB manufactured and assembled by them.)  Roughly 10-12 layers, microvias and laser-drilled vias used, density roughly 4/4mil and a few small BGAs/QFNs but nothing beyond 0.5mm.   

I am getting all sorts of strange logic level signals on my oscilloscope or multimeter.  So for instance we have a 3.3V programming signal to our microcontroller but the logic levels are all over the place, highs of 2.5V and lows of 1.5V with the high and the low jumping all over over the course of a few minutes.   This then disappears and comes back at arbitrary times.  LVDS signals are also affected, but very low impedance power supply rails seem to be mostly okay.  When I measure with a meter, I might see some 50-100 ohms between adjacent signals, even when all that is on that board is a connector and traces without any actual logic on that board.   Most recently I have a board that just smells a little bad and gets warm and draws more current than expected when 3.3V is applied to it, but a thermal camera and my finger can't locate a specific source - it's almost as if the whole board is just a bit leaky and something is burning somewhere.

The boards had connectors fitted to the board by us, so our suspicion was at first that it might be conductive flux causing shorts between adjacent nets, as I've heard this can be an issue after some years but maybe we were getting unlucky.  I found the technician was using a standard paste flux (years old and awfully contaminated) that required cleaning and that he might not be cleaning the flux off after soldering, so I made sure the boards were thoroughly cleaned before testing them.  This, for the time being, seemed to resolve the issue, however it has begun rearing its head again even after we use no-clean flux AND make sure to properly clean the board with flux-off.  We threw away the old contaminated flux altogether.  The issues usually appear some time after rework and appear to be located in the area work was done.  We are using 60/40 SnPb solder for most of the work from RS.

Has anyone seen this issue before and are they aware what causes it?     Could we be seeing some kind of internal layer shorting that becomes more prominent after soldering is done in that area, exposing the board to heat?  Some kind of 'tin whiskering' effect but only apparent internally in the board?  Contaminated FR-4 or prepreg that isn't well-insulating?  I'm tearing my hair out!

As long as I know how to avoid it I can take steps in that direction, but right now we only have a vague sense of the triggers and not really an understanding of the fundamental effects.  I've considered x-raying boards or even doing a SEM scan on solder joints, but the cost of doing that isn't trivial and I'm not sure it'll give us that much. 

Any thoughts appreciated ...!
Tomorokoshi:
1. Do you have blank boards?

2. How much variation is there between the blanks?

3. Could metallization from the laser drilling be recondensing somewhere?

4. Does it change when flexed?

5. Can you cut out an area with isolated copper planes only and measure between them?
Marco:
How about first doing a cross cut after thermal cycling, polishing it and looking at it under low magnification?

If there's voids (crazing/delamination/whatever) there's a lot of other possible problems which become worse. Easily visible and easy to slap the PCB manufacturer over the head with for.
jmelson:
Many years ago, I had a computer tape controller that had worked for several years, but then failed.  Checking the internal reset/ signal showed it was stuck low (in the reset state.)  Checking all ICs that drove this node, and that received it, did not show a failure.  I cut the trace, which ran all over the board, to isolate the cause.  I finally found a 1 inch strip of trace that was shorted to ground.  Peeling that trace from the 2-layer board revealed no visible anomaly, but there was some kind of conductive contamination inside the fiber-epoxy substrate that had become conductive.  One of the wierdest cases from my years of fixing stuff.
Jon
jmelson:
Are the blank boards 100% tested on a flying probe tester?  I've had lots of boards that failed after assembly, where I had paid for electrical test, which had inner-layer shorts.  The probes leave tiny dimples in the center of the pads that are easily detectable under a microscope.  This was the one cardinal sin that made me blacklist a PCB fabricator.  If I paid for electrical test, damn it, they need to DO it!

The kind of conductivity you describe OUGHT to be picked up on a flying probe tester.

Jon
Navigation
Message Index
Next page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod