Weird PCB failure

[cloudscapes]

Hello,

Hope this is the right place to ask for advice on a problem I'm getting. These last few years I've gotten in the habit of making my own minimal microcontroller dev boards that can be plugged right onto a breadboard, with most of the relevant pins broken out. Power stuff, programming connector, and that's about it. They have worked fine until now. They tend to look like this:

https://oshpark.com/shared_projects/ih8ImCyE

I pulled a couple of these out to start a new project, I had already populated them a year ago and they had worked just fine. Trying them now, they appear dead, and the programmer won't even talk to them. I did some debugging and found out both of them had failed in the exact same way, while in storage for less than a year!

In this screenshot:
http://dronecloud.org/stuff/pic32_board_via_fail.jpg

There is no more continuity between any the ground vias (under the chip) and their adjoining pads. On both boards I have, all four via->pads have failed where they worked before. The underside of the via I can probe does have continuity with other ground points elsewhere on the board.

I'm just a hobbyist, I don't have any formal EE training, any PCB design knowledge I have is self-taught from the internet, plus a lot of trial and error. Did I design this wrong? Is the via ring thickness too small? Is the trace too thin? Is it a stress issue, a heat issue? The micros get warm, but not hot. I use isopropyl alcohol to swab PCBs after soldering. Even though the PCB worked for a past project, could the alcohol have damaged the pcb long-terrm? I tried reflowing the pads with no success.

The PCB service I use is oshpark, so should be good? 1oz copper thickness.

Thanks in advance for any help with this. I don't understand how this broke aside from a vague "this is probably a stress failure", and I really need to avoid this in the future.

[Alex Eisenhut]

Pictures of the real thing?

[cloudscapes]

Pictures of the real thing?

http://dronecloud.org/stuff/pic32_board_via_fail_real_thing.jpg

No microscope sorry. I have a 10x magnifier that I use to inspect stuff, that's it.

It's a little messier now than it was earlier before I started because I had a go at remelting pads, scratching traces to probe, bodging stuff, etc. And like I said, both of these exact same populated boards worked fine several months ago.

No picture of the other one with the fault. I tried removing the micro to see if I could confirm that the traces between the pad and the via broke, but that was way too messy to be useful.

Interestingly on this specific board, I found another broken trace that wasn't broken before. The programming header, the second-from-the-right pin. Its pad no longer had any continuity with its trace (which I had scratched to probe). There's also some greenish stuff on another of the header's pad. Corrosion? I kept these in an anti static bag in a drawer for 10 months.

The smallest traces (the ones that broke) are 10 mil.

[rob77]

greenish stuff might be a sign of corrosion, but most likely caused by some acids... what kind of soldering flux are you using ?
also check the continuity between the via's internal wall and the annular ring (kind of tricky to check that) but if it's disconnected then it's clear - the plated copper was too thin and was eaten away over time (edge of the via between inner wall and ring) - again the flux would be my first suspect.

[cloudscapes]

greenish stuff might be a sign of corrosion, but most likely caused by some acids... what kind of soldering flux are you using ?
also check the continuity between the via's internal wall and the annular ring (kind of tricky to check that) but if it's disconnected then it's clear - the plated copper was too thin and was eaten away over time (edge of the via between inner wall and ring) - again the flux would be my first suspect.

Flux pen I'm using is this one:
http://www.kester.com/products/product/2331-zx-flux-pen

PH neutral, apparently.

The flux kind of makes sense the more I think of it. Where would there have been remaining flux I couldn't swab off with the alcohol? Under the microcontroller, and under the plastic header (I had removed the plastic before taking the photo). Where are there issues with broken continuity around vias and pads? Under the microcontroller and under the header.

I don't think I can test the inner wall and ring of these vias. They are ridiculously tiny. I have magnet wire that will just barely fit if I jam it in.

[elecman14]

Probing in the vias could do more damage than good. Here is how you could isolate your problem to a via failure. This can be done by scraping the solder mask on the traces on opposite sides somewhat close to the via. Then you could measure the resistance with a 4 wire kelvin style resistance meter. If you find no problems, a non-matalic (don't buy gold, silver color etc...) nail polish or solder mask repair can be used to cover up the scrapings.

[wraper]

Flux pen I'm using is this one:
http://www.kester.com/products/product/2331-zx-flux-pen

PH neutral, apparently.

That says it all:

2331-ZX Flux Pen® is an innovation in organic acid water-soluble flux chemistry

It cannot be used without further washing with deionized water.

[rob77]

Probing in the vias could do more damage than good. Here is how you could isolate your problem to a via failure. This can be done by scraping the solder mask on the traces on opposite sides somewhat close to the via. Then you could measure the resistance with a 4 wire kelvin style resistance meter. If you find no problems, a non-matalic (don't buy gold, silver color etc...) nail polish or solder mask repair can be used to cover up the scrapings.

the vias are already damaged, so probing them will not cause fujrther damage... the task would be figure out what happened to avoid it in the future.
btw.... the best way to close scraped areas of soldermask is to apply UV curable solder mask and cure it... nail polish might contain aggressive components causing further damage.

[rob77]

Flux pen I'm using is this one:
http://www.kester.com/products/product/2331-zx-flux-pen

PH neutral, apparently.

That says it all:

2331-ZX Flux Pen® is an innovation in organic acid water-soluble flux chemistry

It cannot be used without further washing with deionized water.

agree, for manual assembly i would definitely suggest a no-clean flux. even the cheap rosin based Chinese flux is better than acid based fluxes.

[slurry]

OSHpark is really not the best choice for fine pitch,
recently i had to scrap two out of three boards due to short circuits, my recommendation is to find another fab for your boards.

[wraper]

OSHpark is really not the best choice for fine pitch,
recently i had to scrap two out of three boards due to short circuits, my recommendation is to find another fab for your boards.

This has nothing to do with fab, although I would change it purely because of the cost of the boards. Op used corrosive flux which must not be left on the board. It's a very good but active flux. And it must be washed away, just cleaning with brush in water won't do the job. Either washing machine or at least ultrasonic cleaner should be used.

[cloudscapes]

Thanks again guys for all the advice/info. I guess nothing says more hobbyist/amateur than not knowing the flux I had been using needed to be washed using special processed/deionized water. I did say I learn a lot by trial-and-error! I remember getting the flux on sparkfun some years ago, "hobby site, guess its easy to use" is probably what I thought.

I don't really need to salvage the boards. I was debugging mainly to figure out the problem so I could avoid it in the future. Into the bin they go.

I use osh park mainly because they offer regular post office shipping (dont have to use a courier). I know some other prototype fab places now also ship via regular mail, but I just stuck with osh out of habit. Personally have good experience with them, and their specs are more than good enough for my needs.

I'm looking for no-clean pens. I'll probably get something like this:
https://www.amazon.com/Kester-951-Soldering-Low-Solids-No-Clean/dp/B004X4KOWS/

I can still clean it with alcohol, but it isn't the end of the world either if some is left under chips and in vias. In theory.

[wraper]

I would suggest buying genuine Bonkote Bonpen BON-102. Ebay is full of cheap counterfeit crap, genuine is expensive but worth the cost. And fill it with a flux of your choice.

[timb]

Here's how you should clean your boards: Get a plastic container with a lid (Rubbermaid, Tupperware, etc,) big enough for your boards and fill it with 90% (or better) ISO.

Soak the boards in it for 15 minutes after soldering (assuming there's no non-sealed relays, membrane switches or something else that cannot be submerged with on the board).

Now, use a *nylon brush to scrub the board while still submerged. Scrub until you cannot see any flux remaining.

Soak for another 5 minutes then rinse with clean Iso (I use a spray bottle and do it over the container of dirty Iso so you're not wasting anything).

Finally, rinse with distilled water and let air dry.

Put a lid on the dirty Iso so you can reuse it. You can reuse it until the Iso turns "asparagus pee yellow" at which point it's mostly flux and won't do you any good.


*On brushes: You can use a "hard" bristled toothbrush in a pinch, but a pack of anti-static brushes on eBay or Amazon costs basically nothing. Another option is to get a pack of "Gun Cleaning Brushes" which will include a brass bristled brush (brass colored), a hard polypropylene brush (white) and a medium nylon brush (black). I find the polypropylene brushes are great for getting stubborn flux off and the nylon one perfect for general cleaning. Don't use the brass one, obviously. I bought "Hopps" branded set about 3 years ago and both plastic brushes are still going strong (meanwhile I've worn out about 10 cheap anti-static brushes from Amazon in that time).

[wraper]

Here's how you should clean your boards: Get a plastic container with a lid (Rubbermaid, Tupperware, etc,) big enough for your boards and fill it with 90% (or better) ISO.

Soak the boards in it for 15 minutes after soldering (assuming there's no non-sealed relays, membrane switches or something else that cannot be submerged with on the board).

Now, use a *nylon brush to scrub the board while still submerged. Scrub until you cannot see any flux remaining.

Soak for another 5 minutes then rinse with clean Iso (I use a spray bottle and do it over the container of dirty Iso so you're not wasting anything).

Finally, rinse with distilled water and let air dry.

And nothing of this will work with the flux he used. First of all, it won't dissolve in IPA, water / water based cleaning fluid is needed. Secondly, flux under components will still remain, you cannot get with a brush under SMD resistors,  capacitors, and under that PIC MCU on the photo too.

[timb]

Here's how you should clean your boards: Get a plastic container with a lid (Rubbermaid, Tupperware, etc,) big enough for your boards and fill it with 90% (or better) ISO.

Soak the boards in it for 15 minutes after soldering (assuming there's no non-sealed relays, membrane switches or something else that cannot be submerged with on the board).

Now, use a *nylon brush to scrub the board while still submerged. Scrub until you cannot see any flux remaining.

Soak for another 5 minutes then rinse with clean Iso (I use a spray bottle and do it over the container of dirty Iso so you're not wasting anything).

Finally, rinse with distilled water and let air dry.

And nothing of this will work with the flux he used. First of all, it won't dissolve in IPA, water / water based cleaning fluid is needed. Secondly, flux under components will still remain, you cannot get with a brush under SMD resistors,  capacitors, and under that PIC MCU on the photo too.

It's been my experience that a soak will dissolve most of it from under components. (You need an ultrasonic cleaner to really get it.)

Also, there's still water in the Iso. If you use a completely water based flux you could use 60% Iso, or dilute it with distilled water.

No mater what you use, the soak-scrub-soak-rinse is the key.

[wraper]

It's been my experience that a soak will dissolve most of it from under components. (You need an ultrasonic cleaner to really get it.)

Also, there's still water in the Iso. If you use a completely water based flux you could use 60% Iso, or dilute it with distilled water.

No mater what you use, the soak-scrub-soak-rinse is the key.

My experiencing is that often even ultrasonic bath won't remove all of the flux (I check under the microscope). The issue with IPA is called chemistry. For example, try to dissolve salt in the pure IPA. Or rosin in water. None of that will work. Yet both of them good solvents when used for right compounds. Therefore it makes me cringe when I see something like when someone on youtube cleans the corrosion on the PCB with IPA, as obviously, that someone did not learn chemistry well.

[timb]

It's been my experience that a soak will dissolve most of it from under components. (You need an ultrasonic cleaner to really get it.)

Also, there's still water in the Iso. If you use a completely water based flux you could use 60% Iso, or dilute it with distilled water.

No mater what you use, the soak-scrub-soak-rinse is the key.

My experiencing is that often even ultrasonic bath won't remove all of the flux (I check under the microscope). The issue with IPA is called chemistry. For example, try to dissolve salt in the pure IPA. Or rosin in water. None of that will work. Yet they are both of them good solvents when used for right compounds. Therefore it makes me cringe when someone on youtube cleans the corrosion on the PCB with IPA, as obviously, they did not learn chemistry well.

You're right, it depends very much on the chemistry of the flux. My initial reply was a copy/paste from another thread about cleaning boards. I guess I should have qualified it a bit better.

[T3sl4co1l]

Tenting vias that are very close to pads helps prevent them being gooped up with flux or solder (solder being more of a problem with paste reflow).

They probably wouldn't corrode as fast if the board were ENIG, but that's so thin, it's just a matter of time anyway..

Tim

[free_electron]

water soluble flux. MUST be washed off ! . tghat stuff abosrbs misture out of the air and becomes an acid ... and eats plating.

Use no-clean flux instead.

[Jay1011]

OSHpark is really not the best choice for fine pitch,
recently i had to scrap two out of three boards due to short circuits, my recommendation is to find another fab for your boards.

Why do you say this?  OSHPark gives you ENIG finish, which helps quite a bit with getting fine-pitch parts to solder reliably (in my experience at hand soldering and hand reflow).  You also get high-Tg FR-4.  The price isn't bad for small boards, and can actually be a whole lot less expensive than the China fabs if you want ENIG and high-Tg (price those out at PCBWay, for example).

I've done many 0,5mm pitch parts on OSHPark-fabbed boards and never had a problem.

[wraper]

OSHpark is really not the best choice for fine pitch,
recently i had to scrap two out of three boards due to short circuits, my recommendation is to find another fab for your boards.

Why do you say this?  OSHPark gives you ENIG finish, which helps quite a bit with getting fine-pitch parts to solder reliably (in my experience at hand soldering and hand reflow).  You also get high-Tg FR-4.  The price isn't bad for small boards, and can actually be a whole lot less expensive than the China fabs if you want ENIG and high-Tg (price those out at PCBWay, for example).

I've done many 0,5mm pitch parts on OSHPark-fabbed boards and never had a problem.

Price for high tg at pcbway is ridiculous. You can get TG 170 at pcbwin for much cheaper than TG 150-160 at pcbway. tg 180 for a little bit more. On the other hand, you don't need high tg for most of the boards. Also OSHPark is very limited. Min 2.5mm slots is ridiculous. Plated slots with of 1mm with is limiting too. One color only.

[thm_w]

Sparkfun stopped selling this, likely in part due to what OP experienced: https://www.sparkfun.com/products/retired/8967

However, flux is mildly corrosive over long lengths of time (months). The water soluble flux will cleanly wipe away after wetting the soldered surface.

They didn't have a good warning or instructions in the description that it needs to be cleaned, but the comments have lots of good info.

[nctnico]

Here's how you should clean your boards: Get a plastic container with a lid (Rubbermaid, Tupperware, etc,) big enough for your boards and fill it with 90% (or better) ISO.

Why not use flux remover? I always use Chemtronics 'Flux-off No clean plus' which dissolves rosin and lead free fluxes quickly and has very little tendency to form white residues (if it happens, clean again).

[technotronix]

Probing in the vias could do more damage than good. Here is how you could isolate your problem to a via failure. This can be done by scraping the solder mask on the traces on opposite sides somewhat close to the via. Then you could measure the resistance with a 4 wire kelvin style resistance meter. If you find no problems, a non-matalic (don't buy gold, silver color etc...) nail polish or solder mask repair can be used to cover up the scrapings.

Great. Solder mask is the best option to cover up the scraping.