[SOLVED] I am furious - Ordered PCBs are defected.

[Asim]

Hey guys,

So, I designed  my circuit and sent it for manufacturing in China as I usually do. I also paid for DHL shipping because I was in hurry for this design.
Order was placed on 11th and reached today the 19th. So far so good.

Me being a fool started assembling immediately. After finishing a portion of the circuit, I started testing this is when Murphy came to say HI.

after only 10 minutes of troubleshooting, the problem was found. A via is not plated correctly/fully.

At that point, I decided to check the other batch of PCBs I got, all had the same issue BUT IN DIFFERENT AREAS and some PCBs are worst than the others.

who's the supplier you ask ? That would be seeedstudio

Their specification for via is 12 mil (3 mm) minimum  . my design was 15.75 mil  (4 mm)

what should I do now?
how should I approach them ?

Your valuable comments are appreciated.

Regards,
Asim

UPDATE:

SeeedStudio offered me a compensation for the defected PCBs

[Kalvin]

Have you tried contacting them, like writing an email, and explaining your problem and waiting for a reply. Writing a rant in here doesn't really help with the problem and fix your PCBs.

[wraper]

You should certainly take some photos and send them as evidence and request compensation / boards redone. As of PCBs themselves, if you are in hurry, maybe make another order somewhere else to not waste time on communicating them back and forth, before they agree to remake them. PCBway is very fast, however PCBwin offers best quality I've seen from China.

[wraper]

BTW it also suggests they didn't run the boards through automated electrical test.

[Asim]

Have you tried contacting them, like writing an email, and explaining your problem and waiting for a reply. Writing a rant in here doesn't really help with the problem and fix your PCBs.

This all happened in the last 4 hours. I will be contacting them shortly, but I had to rant it out of my system.

[Asim]

BTW it also suggests they didn't run the boards through automated electrical test.

Yea, I came up with the same conclusion. or they did a partial test ?!!

[wraper]

BTW it also suggests they didn't run the boards through automated electrical test.

Yea, I came up with the same conclusion. or they did a partial test ?!!

Are any boards marked somewhere on the edge? Any decent enough PCB house makes 100% testing these days. All what I've seen were boards with crappily "repaired" defects. They didn't retest boards, though. As there were a few occasions they cut through the short, but a bit of it still remained.

[Asim]

BTW it also suggests they didn't run the boards through automated electrical test.

Yea, I came up with the same conclusion. or they did a partial test ?!!

Are any boards marked somewhere on the edge? Any decent enough PCB house makes 100% testing these days. All what I've seen were boards with crappily "repaired" defects. They didn't retest boards, though. As there were a few occasions they cut through the short, but a bit of it still remained.

no marks( other than the serial number they love to attach to the pcb)

[wraper]

Should be like this. If they are not, likely they were not tested.

[nctnico]

AFAIK electrical testing has been replaced by automated optical inspection a long time ago due to cost and ever decreasing pad sizes.

[AndyC_772]

For completed PCB assemblies, sure, but how would you AOI a multi-layer PCB to ensure all the vias are correctly connected to internal layers? There's nothing externally visible to inspect.

True story... (wavy lines)

We ordered a batch of several hundred multi-layer PC cards from a major UK supplier, and as usual, paid for the bare boards to be 100% tested. The PCBs were duly assembled, and they arrived with us for functional testing and configuration.

The failure rate was 25%, all the same problem. A track, clearly visible on an outer layer, had a tiny break in the same place on each one.

We took it up with the board manufacturer, who shared with us the results of the investigation:

- the board was made 4 up on a panel
- one copy of the artwork for the bottom layer had a tiny scratch on it
- the operator carrying out the bare board testing had deliberately waived the test of that particular net because "too many units were failing it". He thought there was a fault with the test jig, so just ignored the faults rather than investigating. Thirty seconds with a multimeter would have found the problem 

[wraper]

AFAIK electrical testing has been replaced by automated optical inspection a long time ago due to cost and ever decreasing pad sizes.

Nonsense, you cannot optically test vias. manufacturers advertise E-test, not optical test. As well I see flying probe marks on all of my boards (under microscope, hard to see on ENIG, very noticeable on HASL). Optical testing is valid only for single layer boards. As well widely used on PCB assembly.

[technix]

This is very strange.

I almost always order my boards from a local no-name board house, but so far there have only been one board being messed up - the first run of M180v4 had a partially missing ground plane on the bottom - which cannot be caught by electrical testing but easily identifiable in visual testing. After I sent them a photo of the messed-up board and a reference image exported from CAD, they very quickly sent me replacements. (Although they did mistook M180v3 for M180v4, which is a fairly minor revision,  and a second replacement have to be sent. I ended up getting to keep a working-but-incorrect batch of M180v4, an extra batch of M180v3, and a correct batch of M180v4, all for the cost of one batch of boards)

[Asim]

I contacted them and explained the issue.
Instead of waiting for their reply, I ordered another batch from a different manufacturer

Will wait and see

[Mattylad]

What they may do is re supply you with foc replacement boards and at a faster delivery time so may arrive before the others you ordered do.
Seedstudio are cheap, you gets what you pays for with them, when you order you keep your fingers crossed.

[ludzinc]

Once received boards from Elecrow, without any drill holes!

Checked the files I sent - yes I had included the drill files so the problem wasn't at my end.

Sent an email with a pic, no questions asked a new batch was sent with their deepest apologies.

I always like to judge  customer service not on them getting it 100% right, but how they react when it goes wrong.

Elecrow FTW!

[Ice-Tea]

Aside from drill diameter, did you check their anular ring requirements?

[janoc]

What they may do is re supply you with foc replacement boards and at a faster delivery time so may arrive before the others you ordered do.
Seedstudio are cheap, you gets what you pays for with them, when you order you keep your fingers crossed.

I believe Seeed and ITead are both using the same fab and they are both the cheapest of the lot (or used to be, didn't check the prices recently). So you get what you pay for in a way - an occasional mess-up like this is to be expected. On the other hand, Seeed has decent customer service and you shouldn't have a problem with getting this rectified.

If you are making something where quality is critical, don't choose the cheapest hobbyist-oriented supplier.

[nctnico]

Once received boards from Elecrow, without any drill holes!

I always like to judge  customer service not on them getting it 100% right, but how they react when it goes wrong.

I agree. Every now and then PCB manufacturers mess things up and what counts is how they make it right. Over the years I have received several PCBs which where made wrong by various suppliers. People make mistakes so switching to a different supplier isn't the solution.

BTW in my experience the boards from Seeedstudio are just as good as the ones I get from Eurocircuits. The price difference is in the cost of labour.

[Asim]

Aside from drill diameter, did you check their anular ring requirements?

Their minimum is 4 mil, I designed for 8 mil

[Asim]

Once received boards from Elecrow, without any drill holes!

I always like to judge  customer service not on them getting it 100% right, but how they react when it goes wrong.

I agree. Every now and then PCB manufacturers mess things up and what counts is how they make it right. Over the years I have received several PCBs which where made wrong by various suppliers. People make mistakes so switching to a different supplier isn't the solution.

BTW in my experience the boards from Seeedstudio are just as good as the ones I get from Eurocircuits. The price difference is in the cost of labour.

to be honest, I have been using Seeed for some time now. first time to get a issue with the PCBs

[Ice-Tea]

Aside from drill diameter, did you check their anular ring requirements?

Their minimum is 4 mil, I designed for 8 mil

And how did you interpret this? This is how it should be:

If you added 8 mil to the 0.3mm for the final via diameter the you effectively have selected 4 mil AR. I noticed seed has options for 4 5 and 6 mil so if you had 4 mil and selected 6 from the options perhaps this could be the reason...

[Asim]

Aside from drill diameter, did you check their anular ring requirements?

Their minimum is 4 mil, I designed for 8 mil

And how did you interpret this? This is how it should be:

If you added 8 mil to the 0.3mm for the final via diameter the you effectively have selected 4 mil AR. I noticed seed has options for 4 5 and 6 mil so if you had 4 mil and selected 6 from the options perhaps this could be the reason...

It is available in their specification page http://support.seeedstudio.com/knowledgebase/articles/447362-fusion-pcb-specification

they mention 4 mil for the annular ring.
 I believe what you read is for Min Tracking / Spacing which is not applicable for annular ring? 

below are my via specs:

pad size=8 mm
hole size=4 mm

[Niklas]

If it is just a 2 layer board, then it would be very obvious if the layer to layer alignment and drill positioning are off. I guess that either the plating and/or drilling process are more likely to cause failure on the PCBs you received. Drilling small holes fast and trying to maximize the service life of the drill bit can cause uneven walls inside the holes. A worn out drill bit can wobble and also cause burr from the fiberglass to extend into the hole. During the following plating process, this burr can result in uneven thickness of the deposited plating material. Something to look out for if the PCBAs are failing after some time with thermal cycling.

If you have a microscope, fine grit sandpaper and some patience then you can make a cross section cut of a via hole to check the plating thickness. Typically the thickness should be approx 20 um or about half the thickness of the copper on the outer layers (18 um copper foil + 20 um plating) 

[salbayeng]

Class 1 boards don't require any visual inspection, but should have less than 3 voids in any via.
see http://www.circuitnet.com/experts/86649.html

I always specify 20mil (0.5mm) vias with 10mil annulus when using cheap PCB manufacturers.
And 100% electrical test.
Even then you can see gaps in the vias occasionally.

If you need quality PCB's then specify class2 with a reputable supplier.