8/13/2008

It seemed like a good idea at the time…

Famous last words.

I was so excited to finally be using professionally-made PCBs. I spent a week researching the subject – how to make Gerber files, best layout practices (applied when possible, heh,) DRC (Design Rule Check) expectations and, most importantly, customer opinions.

I found this article to be quite helpful, and followed the links therein. After checking around I decided to go with Gold Phoenix. They had great reviews and are utilized by Sparkfun Electronics. I used the min/max tolerances stated here on Sparkfun’s BatchPCB site. The restrictions listed on the Gold Phoenix site are actually even closer tolerances, see #22 for “Outer Layer Line Width/Spacing”. I sent my order to Shane in Toronto, who was actually quite helpful and told me when my Gerbers were out of spec/missing dimensions. My Trace spacing was (accidentally – IMPORTANT NOTE TO EAGLE USERS: when you copy and paste your .brd file to make a large panel the DRCs do not tag along for the ride, you will need to re-enter them) 8mil for pours – tight, but in spec.

When the PCBs arrived – apparently, they had been delayed due to a holiday, the dangers of international commerce – I looked them over and thought they looked good. They had even made more than I had requested (as in the article on Near Future Laboratory.) Then I started building.

ABSOLUTE NIGHTMARE. So far, 60% of the Plague Bearer boards have been bad, and the two Parasite boards I attempted to build were bad as well. The types of anomalies varied as well. As you can see in this image:
>
the etch was incomplete and left “short hairs” (my words) that are a pain to find. Each one is in a different place, some out in the open as in the image (and the easiest to find), while others were hidden under the silkscreening. You can also see a spot that has been over-etched. These errors all looked like a fiber had been laying across the board during etching.

The first few were discovered after I had built a couple Plague Bearers and found them working erratically. Then came the good old joy of troubleshooting! I’ve always had the approach of 1) Assume I screwed up or spaced out and forgot something, then 2) Verify components are good, and 3) Check the PCB. Which meant the errors took a while to find.

The short hairs are easy enough to fix once they’ve been found. Just a quick gouge with my homemade scraper/gouging tool and the boards were good to go. Now that I know what’s going on, I just check the boards with my continuity tester before assembly and fix the shorts. Not so simple with the Parasites.

>

As you can see in the picture, there is no pour due to component density. Therefore, these boards had the OPPOSITE problem: over-etching. It took me two boards and 14 hours a day for six days to find these bastards; they were all buried under silkscreening. The Parasite would work intermittently, flexing the board a little would make it stop working, or sometimes the unit would only work when warm – I found this one after I tested one, thought it good, and then went to bed. I checked it again in the morning and it didn’t work. What a mess.

Following all this, I’m forced to rescind the availability of the Parasites until I can get some reliable boards; these ones are just too much trouble to check. I’ll put some good news up in a few days, I’ve got a lot of catching up to do.

~flight

posted by flight at 3:57 PM