Wednesday, 9 December 2015

Guitar RGB fretboard PCBs

We've got our PCB designed and although we're 99.9% certain it's done right, we've had too many instances of ordering factory-made PCBs, only to discover a fault (missing ground connections are classic) or think of something that we should have added to the original design - sometimes even just an hour or so after "pulling the trigger" on an order from China.

So we're always keen to make up at least one, complete, finished, homebrew PCB of our own design before getting multiples ordered (because it always works out cheaper to order ten or so PCBs to spread the cost of the carriage, than just one or two).

Our guitar PCBs are quite long, thin boards. There's too much "black" in them to use press-n-peel (any design with a lot of fill in it causes the press-n-peel to slide around when all the toner is hot and melted) so we sprayed some copper boards with car paint, and used the "laser etching" method

Normally we use white spirit to wipe down the PCBs after "laser etching" but before dunking in the ferric chloride to etch the exposed copper. Doing so removes the thin layer of vapour that sometimes settles after the laser has finished etching. There was no white spirit to hand, but we found that regular household cleaner worked just as well.

It took us three pieces of 160mm x 100mm "eurocard" sized copper clad board to create the full PCB

(in our final design, we're only actually going up to the 17th fret with the LEDs, but the PCB will be the same size as the entire length of the guitar neck, to make it easier to get consistent thickness when the board is cast)

Now the BB laser didn't exactly do a great job of separating some of the tracks that we'd drawn closer together (particularly in some of the higher frets, where things start to get a little bit cramped and squashed together). So we had to tidy up some of the tracks, using a craft knife and the pointed end of our multimeter, to ensure that there were no little bits of copper creating unwanted bridges between tracks.

Because we've had trouble with running tracks under these fiddly RGB LEDs in the past, we thought it prudent to use some solder resist on these boards. Our solder resist paint is UV curable, so some LEDs from an earlier project were just the job

Maybe our UV LEDs were a little fierce, but some of the solder resist over the pads has since cured rock hard. The idea is to cover the pads with a transparency with a black mask over each of the pads - the UV light shouldn't penetrate these areas, leaving the paint wet, while the rest of it cures.
Once done, the idea is to wash off the un-cured paint and leave the copper of the pads exposed.

In another wak-wak-oops moment, we managed to cure the solder resist - but somehow leave a lot of the pads covered with now-cured resist paint too! We had to scrape away sections of solder resist before soldering the LEDs on:

After creating three separate sections, debugging proved particularly difficult. It took a lot of work to get each section to work in its entirely - all the way along the length of the whole section. Which meant that when we combined the three sections and the lights only made it half-way up the first section, we couldn't be sure where the problem lay!

While each of these sections lights up entirely when plugged into the controller individually, when daisy-chained together, only half of the first board would light up. Weird huh?

Could it be that the three sections were not joined together properly? Or that there's a fault or a short on the data line somewhere? Or maybe a bad or dry solder joint where the resist hasn't been cleared off the pads properly? A fault on the first section would be enough to stop every subsequent LED working - so when the lights on sections two and three aren't working, it's actually quite difficult to know where to start!

After a lot of soldering, re-soldering, unsoldering and replacing multiple LEDs, we figured that the PCBs were simply not of good enough quality to be reliable for testing.

Only one thing for it - the one thing we said we'd never do after so many botched PCBs in the past - get some ordered, and just hope they work when they get here!

Before pulling the trigger on the order, we checked and double-checked our PCB layout with the 25.5" scale fingerboard.

Although it's a tiny discrepency, by the time we reach the 21st fret, the fret markings on the PCB are about 1.5mm off where they should be. It's not the end of the world - those fret lines are only guides for placing the LEDs. And, looking at the design, they'd still be ok - each LED is still slightly "north" of where the fretwire will go across the fingerboard. But it might still be worth bunching things up a little bit before committing to getting a dozen or more full-sized (and relatively expensive) PCBs manufactured!

Although close, the fret wire will not actually obscure any part of the LEDs, even if we left our final design as drawn. Still, better safe than sorry - this design had to undergo three or four revisions before we were happy to  send it out.

A few clicks later and we've got the PCBs on order. No doubt, Steve will be smiling to himself - he suggested we do this about three weeks ago! While we're waiting for them to arrive, we'll get on with making a few jigs to help populate and solder them...