Another PCB FAIL!

[DrGeoff]

Just finished stuffing a proto board only to find a short between power and ground. Closer inspection revealed no less than 16 shorts between power, ground and other traces on the board. This one was made by Itead Studio, which I've not had a problem with before. Maybe they've changed manufacturers. Of course, some of the shorts were under connectors and chips. Here's a photo showing a few. Some under U3 are already removed, but the clear shorts under C14 are visible. You can also see a defect in the track along the bottom of the photo.

[FreeThinker]

Contact Itead and explain the problem. If they fix it (redo the boards correctly) then all well and good. If not, dump them. No one is perfect, addressing your limitations is the sign of a good company. Did you inspect the boards before assembly?


<Edit Typo>

[Kiriakos-GR]

The parts holes looks more than big, the the PCB trace around them nonexistent,
this is another major problem other than the random grounding issue.

Looks like that some one tried to squeeze the PCB size, and he created a mess. 

[ejeffrey]

Contact Itead and explain the problem. If they fix it (redo the boards correctly) then all well and good.

Not really.  Assuming the board meets reasonable design rules for the service, replacing the boards is not really a solution.  The cost in time and parts of assembling boards that turn out to be defective is much worse than paying a few bucks extra for the PCBs.  If they have consistent QC problems they need to fix the underlying problem: change their advertised design rules to something they can reliably manufacture or fix their manufacturing process.  Replacing defective boards should be considered automatic, but isn't enough.

Those pads do look rather small compared to the hole size, and not really enough given the apparent drilling accuracy, although that doesn't really explain the shorts you show.

[mkissin]

That seems really odd. Their e-test should have caught those errors, which are really pretty horrendous. Having said that, was this particular PCB one of the boards that was tested? Their website indicates that by default they only test half of the boards, and all PCB houses will have manufacturing defects occasionally. So, that should guarantee that you get 5 working boards and 5 that "might be OK" (going off information from their prototyping service page).

ejeffrey: You wouldn't have to assemble defective boards with 100% e-testing, which is available from itead for a few extra bucks. This shows the use of such a service. If these boards were e-tested, then that's pretty much negligence.

[DrGeoff]

@FreeThinker: I don't really need any more boards, they made 12 and I only need 1 for the proto.
@Kiriakos-GR: I don't know what you are talking about, do you? The layout is a 10/10, except for the MCU which is 0.5mm. The via hole size is 24mil. There's nothing squeezed about this and it is not a mess. The minimum annular ring is 12mil and, having made thousands of other PCB's with these same footprints over decades, find this to be adequate. It's not a hobby-magazine board for beginner solderers who would probably butcher something like this.

Have examined the 12 board received. Only 3 have defects, includingthe one I fitted, which was selected at random after examining another board as a reference. The defects are also random. Murphy's Law of course, since the one I grabbed for fitting is the worst of them all!

[ejeffrey]

ejeffrey: You wouldn't have to assemble defective boards with 100% e-testing, which is available from itead for a few extra bucks. This shows the use of such a service. If these boards were e-tested, then that's pretty much negligence.

True.  I was under the impression that the whole batch was bad, but this is evidentially not the case.  I guess in that case the real moral is to get as many boards tested as you want to assemble.

[Uncle Vernon]

@Kiriakos-GR: I don't know what you are talking about, do you?

The answer to that one is obvious DrGeoff. I'd treat the poster's dig at your competency with the contempt it deserves. A quick glance at your attached photo shows no evidence of poor design, overall that layout looks more than competent.

It appears from other threads that Itead offers both excellent price and inconsistent output. I guess the cost of their super low prices is a need for vigilant board checks before assembly. In a hobby situation that can be quite acceptable, in more demanding circumstances their budget offers are probably not the best choice. I've been considering sending a few boards that way but will watch for some further feedback first with interest.

[DrGeoff]

It's been so long since I've had a bad board, since these days the processes are far superior to what they were even 10 years ago. I used to get bad boards regularly with the 2:1 red-blue taping and photography registration errors, but that was back in the 70's.

I guess I got lazy and caught off guard, not scrutinizing the board under the microscope before assembling it! I put these few down to chemistry problems, since it is only a few and the problems, whiilst all on the top side of the board, are all in different places. Overall I've had lots of good boards from Itead, but I guess it depends on what manufacturer they are using at the time as well.

The PSU shorts are gone and we now get 5V on the supply rail OK. Now to see if the MCU can be programmed!

[FreeThinker]

This thread has stated me thinking ( ) .Just exactly how does a pcb get made? Do they use the same techniques as home made ones? ie sensitised boards and masks or a litho process or something else? Looking at the faults on the board I would suspect the former ( exposure ) technique as this would explain nicely the random faults (dirt or debis on mask or copper board blocking exposure) Any one any info /links on the process? I think it would be interesting and informative to see exactly how these are made in a production environment.

[Rufus]

The result of poor process control followed by lack of inspection and test. They didn't etch long enough and/or the etch was weak. The etch isn't completely uniform so when you don't etch enough some areas of the panel being OK and others bad is to be expected.

Couldn't see the track defect well enough to say if it is the same problem or dirty/damaged artwork or materials.

[metalphreak]

Do any of the faulty boards have a black marker line on the PCB edge? The 50% that they e-test will have marks on them (and that's 50% of the ten specified, so 5 out of 12 should have been tested).

If you aren't going to use all the boards, then make a habit of using all the tested ones first


Also, they charge $8.25 for 100% e-test on the basic 10board 5x5cm package, which is only $9.90... You're better off just buying two batches of 10, getting 10 tested boards as well as 10 untested ones for like $1 more... ?!

[ejeffrey]

The drill registration doesn't look so great either.  The drills are all over the pads, and not all in the same direction. 

[DrGeoff]

Do any of the faulty boards have a black marker line on the PCB edge? The 50% that they e-test will have marks on them (and that's 50% of the ten specified, so 5 out of 12 should have been tested).

If you aren't going to use all the boards, then make a habit of using all the tested ones first

I should have checked, and yes should have used the tested one for assembly. My mistake, I got used to these fairly low density and large clearance boards being free of defects.

All is not lost, so far testing has not revealed any other shorts or problems.

BTW, I got another batch of 24 other boards back from Itead today and they have perfect registration and no apparent defects.

[IanB]

This thread has stated me thinking ( ) .Just exactly how does a pcb get made? Do they use the same techniques as home made ones? ie sensitised boards and masks or a litho process or something else? Looking at the faults on the board I would suspect the former ( exposure ) technique as this would explain nicely the random faults (dirt or debis on mask or copper board blocking exposure) Any one any info /links on the process? I think it would be interesting and informative to see exactly how these are made in a production environment.

There is an article on Wikipedia about this. For run of the mill boards I believe they start with a blank copper clad board, drill the holes first, plate through the holes using the copper cladding to conduct electricity to the holes, print the etch resist using a screen printing process, etch using a temperature and time controlled spray jet system, then lastly print the solder mask using a further screen printing process. Or something like that. I've often wondered myself exactly how they do it.

I was coincidentally reading that a similar printing technique can be used by the home fabricator using an ink jet printer and a CD printing tray. Some Epson inks contain microscopic plastic particles that will melt into an etch resist layer upon heating gently in an oven. So you basically print your design on the copper clad board using the ink jet printer, fix it in place by heating, etch it, clean off the etch resist using acetone, then drill the holes. You can even put the board back through the same printer and print the silk screen on top. For a finishing touch you could probably silver the copper traces using some of that rub-on silver plating compound?

[FreeThinker]

Yes I've heard of the same technique with epson printers... magenta ink seems to give the best etch resist for some reason. I might give it a go at some point

[Semantics]

There is an article on Wikipedia about this. For run of the mill boards I believe they start with a blank copper clad board, drill the holes first, plate through the holes using the copper cladding to conduct electricity to the holes, print the etch resist using a screen printing process, etch using a temperature and time controlled spray jet system, then lastly print the solder mask using a further screen printing process. Or something like that. I've often wondered myself exactly how they do it.

I found this tour of Advanced Circuits quite interesting, to put a picture to each of those steps.

[LEECH666]

You  get what you pay for.

My experience with Itead so far -> *click me*

I am still happy with their service.

The Fusion PCB Service of Seeedstudio is equally cheap. Maybe they have a better process?

[shadewind]

I have have ordered several designs from iTead (most with clearances slightly below 8 mil) and I have never had a faulty board.