Electronic PCB Intermittent Vias

Fix the material or process problem causing the defects or wave them first.

Use either a soldered wire or eyelets to repair current defects.

Wait one minute, why do you have vias at all with a single sided board? A via permits conduction from one plane of a circuit board to another plane. The conductive artwork of a single sided board resides only on one side. So any through hole components do not require any vias, a solderable eyelet on the conductive side will do.

The components are single sided surface mount as opposed to double sided. The PCB themselves are four layers.

If they are single sided surface mount, there should not be any vias. are you sure they aren't multi-layer boards? if they are, there is no fix for it. the internal layers won't make good connection to the via. If they are double layer only, then it is IS possible to fix them but the cost could make rework prohibitive. you might do better by having the more valuable components removed from the defective boards and recycling the components and chucking the defective boards.if you send them through wave soldering, you might fix them, or you might just make the problem worse. ultimately you have a plating problem. until you fix the plating problem, this is going to keep happening.

Thank you for your comment.

I will be addressing the plating issue with the PCB manufacturer.

For the moment, I am just trying to save a hundred of boards from being trashed. Each board has hundreds of vias. Too many to do by hand. We did solder the vias for a few boards manually and they passed the high temperature test.

My question is: Is it a viable way to fix the boards? If not, I will trash them. One failed board in the field cost more than a dozen boards.

The loss of reputation is even more important.

I don't want to take a chance. I am looking for some "assurance" based on real experience that this could solve this problem for the already produced boards.

I remember many years back that a company I worked for had a large number of multi-layer boards with a similar problem. Before new, well plated boards could be obtained the questionable boards went through a modified assembly process. Through hole and vias were first soldered in a wave soldering process. This required well controlled soldered temperatures so that solder wicked into the vias filled them completely. This was also with a lead based solder, too. The surface mounted parts (a small number of components for this board) were then had a lower temperature solder paste applied by hand and baked for soldering. This was not a cost effective solution, but it permitted time critical jobs to still be completed on time with complete reliability.

Sounds logical, but aren't the OP's boards already populated?

I was only explaining what we did in a similar circumstance many years back. Only our OP has any idea how critical this product is, the cost of the soldered components, the expected service life of the product, the lead time for new boards and the warranty if any from the board manufacturer. These and many other factors must be known to make a sound judgement on what to do with these defective boards.

In my circumstance the questionable boards deployed were all part of time critical installations. The boards were installed with the stipulation that a robust free replacement would be sent to the customer when they became available. It was up to the customer to schedule down time for their facility to swap out the questionable boards with the replacements.

OP,

I don't see what the boards are used for, but it seems that you've got a LOT riding on these things. With that in mind, if at all possible I'd send the problem back to the manufacturer. This especially applies if the boards are to be used in any kind of medical or other product that could potentially have a high liability if they have a field failure. In that case, I don't think I would even try salvaging components.

I can sympathize with your plight and this is just my 2 cents so please feel free to disregard it, but I'm with Tobugrynbak: doing anything else amounts to assuming ownership of a problem that rightfully belongs to someone else.

No life danger but annoyed customers and expensive replacements of bad boards. We try to minimize the impact on our customers.

This is why we took all the sensitive boards out of production.

I hate to scrap so many boards but will do it if we cannot find a reliable (and cost effective) way to fix them.

The boards are TRASH!!!

Even your oven testing is not an assured process of removing all defective vias, and thermal cycling in product life will certainly break those that have a tiny whisker of conductivity across the via.

Wavesoldering will probably not work. (It might work 95% of the time, but with hundreds of vias involved, some will still fail.)

Failed vias in the internal layers will not be fixed by wavesolder.

Spend your efforts expediting good product rather than trying to make a silk purse from a sows ear. (Even if it means using a more expensive board supplier under deviation for a short run until "production" boards are available.)

If the vias are that bad there is NO reliable way of repairing them.

soldering through the vias or even adding a wire and then soldering through is not guaranteed to achieve a reliable connection to the buried layers.

It appears that the bare PCBs were not manufactured to industry standard, however you NEED to do more investigation.

Take a NEW blank board and a faulty board - identify some faulty vias on the faulty board and have both boards professionally (and independently) microsectioned and the sections photographed through those vias on both boards.

You now have some evidence to determine if the fault lies in your assembly and testing process or the bare board manufacturing process.

From there proceed accordingly.

We had produced those boards before with good results.

The design is conservative with features about twice what the board manufacturer claims to be able to do.

Thank you for the info about the Tg. I will make good use of it.

Did you change your PCB supplier?

This sounds like a quality control problem, as the PCB expands though heat, the vias are damaged....

Trash them all......

It is also possible that the error is in the photoplotting process too. An incorrectly typed/transcoded Dcode aperture width in the Gerber file could result in a donut or trace that is still there, but smaller than expected/intended. That is less of a problem these days with automated optical inspection of both the mask and the finished boards, but it is still possible.

It's good to know the 1980s-based predominently Japanese trend towards zero defect manufacture is alive and kicking.

I would be curious...

These PCBs that are bad: Are these simply defective boards purchased from a reputable manufacturer you have used for an extended period of time? Or, are these boards knockoff (counterfeit) components that are considerably less expensive than another, and thus the manufacturer is reluctant to replace them? Or, are you in fact, manufacturing these boards?

Are you experiencing counterfeit convulsions that we have recently discussed?

We don't buy from abroad if we can help it.

This PCB mfg has been in business for a very long time and has given us a good service. They are not a cheap house. Two other types of boards were made at the same time and are OK.

I will be looking into Daffy's comment above to see if we would not have a critical feature or if some dirt/hair on the mask would not have caused this problem. Maybe they had a new employee working on this batch.

I intend to contact them as soon as I can prove my claims.

Here is some info to keep you all informed of the developments.

I contacted the PCB manufacturer and he was very helpful in reviewing his data from the batch. He also did some destructive tests to measure the copper thickness in the vias etc...

Everything checks out on his side.

He noticed that the boards were warped by a few percent. He said that it is probably because the board assembler used a ROHS temperature profile on a standard PCB or he did two or three re-flows because of missing parts.

I don't know what happened as the assembler is not cooperating in the inquiry.

The PCB manufacturer suggested to use a material with a TG of 180C next time. This will make the boards more resistant to thermal abuses.

Thank you all for your comments.

P.S. We are going to scrap the left over stock of boards even if the high temperature test seems to work properly.

So Daffy was spot on - Kudos to Daffy

how so? Tg was 140 and the assembler used 170 in the wave soldering cycle?

It might be a good idea to take a statistical sample of raw boards from each batch and thermal cycle them for a whole bunch of cycles then put them on a bed of nails to verify that the board itself stood up to the abuse before sending the rest of the batch to the assembler to be populated. of course the samples would need to be sectioned and inspected afterwards.

Military applications do not like through plated holes because in battlfield conditions, say inside a tank, the shock vibrations can delaminate the boards and break the vias. To get around this problem all through holes are pinned and soldered both sides. Most boards are pinned by machine, Harwin, Elete and Y5 were three manufacturers I dealt with but there are several others, some with fully automatic machines, some with semi-automatic. The Y5 machine was by far the best but they ceased trading about 1990. Check out the makers of 'through pins' for current solutions.