Author Topic: PCB short between layers (Read 4965 times)

Franck · « **on:** February 19, 2018, 01:23:52 pm »

Hello,

I just got a strange PCB failure:
I have a 130 Ohm resistance between a track on the top layer and the power plane just under it. (I will call it a short, even though it is 130 Ohm)
This is a 4 layer PCB, freshly populated, has never been used and the track is only 40mm long (easy to inspect).

At first, I thought it would be a short between the via and the power plane, but after removing the components and cutting the track before the vias, the short is still there.
I visually check the track under microscope and it does not touch any other tracks.
The only thing I can imagine is that the short is underneath the track...

I was wondering if any of you had already had this issue and what could be the manufacturing process that caused it.

Thanks,
Franck.

Ice-Tea · « **Reply #1 on:** February 19, 2018, 01:27:30 pm »

Never. And 130 ohm isn't much of a short. You can always try to blow it out or locate it by means of electroluminescence by attaching a high current source and ramping it up...

Armadillo · « **Reply #2 on:** February 19, 2018, 01:34:27 pm »

Hi, I like taiwan.

130 ohms don't appear to be a short.
Why? Even you cut the top tracks, the vias will lead to other connections unseen to you.

If you have a schematic, you can track the vias. But I don't think you would be asking if you have one, right?

Maybe you can picture the board, so that we can have a look at it.

Franck · « **Reply #3 on:** February 19, 2018, 01:41:15 pm »

Ice-Tea,
Fixing the board won't be an issue, my main worry is to know what part of the PCB manufacturing process is at fault.
I have another board with similar behaviour, and I want to get back to the PCB supplier with a bit of knowledge first.

Armadillo,
Yeah, Taiwan is great, it's why I moved there from France

!
There is no more via along the track, I designed the PCB, so I am sure about it

!
I'll get a picture in my next post.

Franck.

Armadillo · « **Reply #4 on:** February 19, 2018, 01:45:14 pm »

Hi, the PCB libraries, is it from the public domain? "FREE".
Even from sparkfun, there are mistakes.
Have you checked or verified those?

Franck · « **Reply #5 on:** February 19, 2018, 01:54:44 pm »

Here is a picture of the track.
Quality is not very good, the 2 cuts before the vias are a lot cleaner than what it looks.
As you can see, the track is short and there is hidden via.

F.

Armadillo · « **Reply #6 on:** February 19, 2018, 02:04:49 pm »

Quote from: Franck on February 19, 2018, 01:54:44 pm

Here is a picture of the track.
Quality is not very good, the 2 cuts before the vias are a lot cleaner than what it looks.
As you can see, the track is short and there is hidden via.

F.

I only suggest that you verified from the name in the netlist for this track, how many connection. 2? or more.
But nevertheless;
Is this a completely empty board? Then ask manufacturer of PCB for refund surely. A short will read zero, you argue to them.

Franck · « **Reply #7 on:** February 19, 2018, 02:13:59 pm »

Yes, only 2 connections on the netlist.

Franck.

Armadillo · « **Reply #8 on:** February 19, 2018, 02:34:59 pm »

Before you bang the manufacturer.

how about a reverse side of the same picture you shown and indicating the same vias / track.

Franck · « **Reply #9 on:** February 19, 2018, 02:53:01 pm »

You won't see the track on the other side, this is a 4 layers PCB.
But I can show you the view from the PCB editor. The shorting track is the blue one in the picture.

Franck.

Armadillo · « **Reply #10 on:** February 19, 2018, 03:14:45 pm »

See attached.

Edit: Including the white outline.
Edit: Oops, I think the white ones are the grid. But hey, Franck, those pads should be snap to grid or half grid. I think you need to tidy those.

Monkeh · « **Reply #11 on:** February 19, 2018, 03:35:43 pm »

Why should they snap to his arbitrary grid when they're a very specific footprint for a connector?

And seeing as he visibily cut the trace before it gets anywhere near that connector, what does it have to do with anything?

Your faulty line runs very, very close to a via - any possibility it's shorted to that and you're reading through a component fitted on the other side?

sleemanj · « **Reply #12 on:** February 19, 2018, 09:33:33 pm »

Divide and conquer the track to narrow down the location of the flaw. I do notice that the track below has what appears to be a protrusion going towads the problem track.

Franck · « **Reply #13 on:** February 20, 2018, 11:17:06 am »

Armadillo,
The short is on the part of the track which is blue (bottom layer).
The track has been cut before the via, the bottom layer is insolated and the short is on this part of the track only.
The outline you have marked on the pictures are keepouts. They are not part of the copper.

Monkeh,
Usually, shorts are on the via or next to it, but this time the issue is on a straight line...
See attached pic.

sleemanj,
Yep, I did divide, but I don't think I conquer!
I reduce to few 5 millimetres of track and I still don't see the short.
See attached pic.

This confirms what I thought: the short (or low impedance if you prefer) is under the track.
As I say in my first posts, my problem is not to fixe the board but to understand the issue.
If it was a short between two tracks, I wouldn't bother. But that's the first time I see a low impedance between two layers. And it looks like I have a 2nd board with the same issue...
As anybody seen this issue before?

Franck.

Armadillo · « **Reply #14 on:** February 20, 2018, 11:37:23 am »

You still don't quite get it.
If you put outlines, labels, lettering, keepouts etc. onto the wrong layers, it become copper track or copper lettering, or copper keepouts, not silk screened.
Seeing the color, I would say, it became a inner layer shorting copper keepouts.

In this case, all the boards will exhibit the same problems.
It will read all kind of shorts or ohms.

You should inspect your gerber file using gerber viewer of each layer and you will understand it.
Before you send board for manufacturing, always check by using gerber viewer.
Edit: In the Gerber viewer, you can see whether the keepouts are in the layer or not. If in the layer, its copper. You can't see that with your PCB program.

You will also need to correct your libraries and put into "Franck verified lib folder".

if still ....dian dian..... you can't understand.... I can't help you then....

Armadillo · « **Reply #15 on:** February 20, 2018, 12:04:00 pm »

It is quite simple.

Print the same portion of each layer of the 4 layers board, from the Gerber viewer program to ascertain there are no shorts.

We like to see it.

If so, you can challenge the manufacturer.

Franck · « **Reply #16 on:** February 20, 2018, 12:09:33 pm »

Armadillo,
Thanks for teaching me how keepout works, but after 20 years of PCB design on Altium, I am pretty sure I got it right...
As other members pointed it out to you, the problem is not near the connector as the tracks have been cut before the connector.
Also, we have over 400 of those boards fully tested and running for several years so the problem won't be in the design.

Franck.

Armadillo · « **Reply #17 on:** February 20, 2018, 12:13:37 pm »

Quote from: Franck on February 20, 2018, 12:09:33 pm

Armadillo,
Thanks for teaching me how keepout works, but after 20 years of PCB design on Altium, I am pretty sure I got it right...
As other members pointed it out to you, the problem is not near the connector as the tracks have been cut before the connector.
Also, we have over 400 of those boards fully tested and running for several years so the problem won't be in the design.

Franck.

Thanks. 20 years is a long time and a working relationship with the manufacturer.
Have you sought for their advice?

Edit: Pity that 20 years....

sleemanj · « **Reply #18 on:** February 20, 2018, 01:07:24 pm »

I take it that the short section in the last photo exhibits the problem but the rest of the trace does not.

I would scrape down either side of that short section with a knife just to be sure it's not something on the surface under the soldermask which is evading your sight.

Then I might hqving exhausted options pass a goodly amount of current through the problem and see what gets hot (depemding on current, what starts smoking!)

And finally break out some HCl and H202 and etch away the copper of that trace to see what is under it.

Franck · « **Reply #19 on:** February 20, 2018, 01:34:23 pm »

sleemanj,
Yes, only the 5mm left is the problem, the rest of the track is fine.
I went directly to remove this last section of copper (didn't see your post before).
I first run the scalpel on the edges of the track in case there was any copper shorting there, nope.
I then went onto removing slowly the track under microscope to see if there was some copper protruding from the middle layer.
I saw some shiny bits after removing the copper, but it is difficult to say if it belonged to the track or to the middle layer...

Anyway, now my track is gone, so there is no short left!
I guess I'll never know for sure, but everything points to a short between layers.

Thanks for your help.
Franck.

eeviking · « **Reply #20 on:** February 20, 2018, 05:48:19 pm »

Sounds like contamination of the FR-4 with something conductive.
I have seen something like this before, measuring around 50ohm from an unconnected track to the plane below on a 4 layer board.
Take it up with your pcb supplier or find a new supplier.

Franck · « **Reply #21 on:** February 21, 2018, 03:20:00 am »

eeviking,
Thanks for the feedback, I had never heard of this issue before, started to think I was missing something...
Wanted to be sure before I get back to the supplier.

Franck.

Armadillo · « **Reply #22 on:** February 21, 2018, 05:06:40 am »

400 over boards fully tested and in use and you are still not sure how to get back to the supplier?

?
I would just send the board back and ask them for a reason for such manufacturing quality lapses instead of you finding a reason for them.

IconicPCB · « **Reply #23 on:** February 21, 2018, 05:31:36 am »

Franck,

It is quite possible to see such a fault especially in a multilayer board.
A lot depends on quality of prepreg used in pressing the board.
Sometimes if only a single sheet of prepreg is used You can end up with internal shorts , shorts which have been created by pressing process through prepreg.

Prepreg can also create another similar issue... during hole wall activation process if prepreg is of inferior performance you can get activation chemistry leaking into the layup and creating a copper short within the PCB.

I just fell foul of an aged piece of prepreg. I was pressing a panel last week and ended up with a bunch of shorts internal to a four layer board.

Closer inspection ( microscope ) revealed copper leakage into the layup.

Needles to say a bunch of prepreg got tossed out into the bin and a new batch should be here tomorrow.
I went as far as sectioning and potting the faulty area in order to inspect the formation of shorting intrusion.

Franck · « **Reply #24 on:** February 22, 2018, 02:24:18 pm »

Hi IonicPCB,

What you describe is exactly what I wanted to know when starting this post. Thanks, it makes a lot more sense now.
I got in touch with the manufacturer but didn't get much feedback yet.

Blueskull,
yeah, over 2 layers a consider the e-test mandatory and those boards were meant to be tested...
But again, that's not the first time I see tested boards being delivered with shorts. (Once, I even got a board with CNC slots in the middle of the tracks, they had taken the silkscreen as a mechanical layer!)
What got me with this one is that the short is between 2 layers, never seen that before.
I won't test the 1kV breakdown voltage on those boards, they have components on it...
I may do the test on the next bare PCB I receive.

Thanks,
Franck.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: PCB short between layers (Read 4965 times)

Share me