PCB error (EAGLE 5.1 and BatchPCB toolchain)

Greetings,

I’ve just completed construction of a project using a

PCB fabbed by BatchPCB. Unfortunately there was a

via to pad short that took a bit of detective work to

find. Apart from being under the body of a component

it was so thin that a 10x eye loupe was required to

see it. A brush with a sharp knife took care of it!

Has anyone else seen PCB defects like this?

The board was the first I’ve done with EAGLE 5.0,

which I upgraded to EAGLE 5.1 during the project.

I see that EAGLE 5.2 has been released, in part

to address a known bug that causes random shorts

in the Gerber output files (!!?).

In my case I have build two identical PCBs, and

as luck would have it the first I powered up had the

short, the other does not. I can’t find any defects

in the EAGLE files, and strongly suspect this is a

random PCB fab process error (dust, dirt, or

abrasive scratch on the tooling is my likely guess).

This is the first and only BatchPCB defect I’ve found

in over a dozen board designs, but it does come on the

heels of the recent EAGLE bug fix. Which is very good

for a “no touch” fab service.

Comments Welcome!

Most suppliers offer board testing as an option. I don’t bother with simple boards, but I’d always use it for anything complex. It wouldn’t catch a short caused by a bug affecting the Gerbers, though.

Leon

As a rule I get all PCBs with internal layers electrically tested. I guess I’ve always figured it’d be more likely for a board with internal layers to have shorts - and also internal shorts would just be so much harder to track down and fix.

It’s surprising to me that PCBs aren’t electrically tested as part of the standard process. If their process is bad enough that occasionally problems like this happen - it just makes sense to me that they’d do it for free.

Oh well.

I thought it was common knowledge that 2 layer boards on the BatchPCB service aren’t electrically tested, only the 4 layer ones, which helps makes the price what it is.

Peter,

From the sound of it, I’d say it’s a production defect based on:

  1. It was small enough to need a loupe, which sounds like it would have been thin enough to trip the BatchPCB bot if it was in the gerber, and

  2. I’m under the understanding that you used the same gerber for two boards. One worked, one didn’t, so it couldn’t be the gerber.

Still, it does speak well for BatchPCB/GP - I don’t recall hearing of any other complaints about board quality (or I’ve missed them).

NleahciM:
As a rule I get all PCBs with internal layers electrically tested.

Sorry, forgot to state that this is only a two layer

PCB (except the physical short was under the body

of another component after assembly).

NleahciM:
It’s surprising to me that PCBs aren’t electrically tested as part of the standard process. If their process is bad enough that occasionally problems like this happen - it just makes sense to me that they’d do it for free.

The two layer BatchPCB service is “no touch” (i.e. no

testing beyond visual inspection or possibly test coupons

on the panel borders for alignment and etch depth).

Electrical testing requires a “Bed of Nails” fixture custom

made for each design and more costly than the actual

boards!

The BatchPCB four layer process includes electrical testing

and costs quite a bit more ($8/sq vs $2.5/sq IIRC).

With internal layers they have to deal with opens as well

as shorts, so electrical testing is both continuity and isolation

from all points to all points.

Electrical testing these days is quite cheap, they use a flying probe.

Leon

leon_heller:
Electrical testing these days is quite cheap, they use a flying probe.

Leon

Yup yup - hence why it seems logical that it'd be a part of the standard process.

But then again - that may just be me yearning for a simpler time when people actually stood behind the products they sold…