Slashdot

Several "oh bugger" moments.

Firstly, RS got part of my order wrong, and sent me 8MHz crystals instead of 4MHz ones. I can't be bothered sending them back though. For fun, I thought I'd try swapping the 4MHz crystal on the breadboard with an 8MHz one and see what happened. I hooked up the oscilloscope to check it was resonating (it did). To my surprise, the breadboarded computer mostly worked at twice its rated clock speed (the Z80 and PIO are old parts datecoded from 1985, rated at 4MHz). The PIO and Z80 worked fine. Some of the peripherals didn't though. The CTC worked a bit, the counter didn't work at all, and the ADC didn't work at all. However, I suspect this has more to do with parasitic inductance and capacitance inherent in the plug-in breadboard and rats nest of wiring than the chips themselves.

Aside from the wrong crystals 'oh bugger', there were several today when I began to fabricate the PCB that I've designed for the Z80 computer (named ALIAC-2. ALIAC-1 is the breadboard system, and a number of modules that can be plugged into breadboard).

Oh Bugger Moment 0. I don't think I cleaned off the copper clad board well enough, and when I had got them down to toner on copper, I could see a few broken tracks. These I filled in with permanent marker. The three PCBs I've made so far haven't suffered this problem. But the permanent marker did its job.

Oh Bugger Moment 1. I tried to iron on the silkscreen layer but the toner didn't stick very well. A bit of a disaster really.

Oh Bugger Moment 2. After etching, I discovered I had under etched the board. Not majorly, but such that a few tracks were bridged. This came back to haunt me later.

Oh Bugger Moment 3. Soldering the vias (this is a double sided board), I do by soldering in a wire, clipping it off close to the board on the other side, then soldering the other side. What I want to know is this: when you are actually *trying* to bridge two copper areas, it's almost impossible, but when you're trying *not* to bridge two tracks together, it's almost inevitable that you do. So I melted the solder again on the one that bridged, and used the solder sucker to get rid of the problem solder. This also sucked the track off the board! So I remade the via by removing the lifted piece of track, and soldering in a longer bit of wire. Then I found the track was still conducting with its neighbour. After scraping away at it for ages, it was still doing so. I found the cause eventually - the underetching had left a pad on the 40-pin ribbon header shorting against it. Fixing this one via took the best part of an hour!

I then found even more shorted tracks. So I had to carefully go through and scrape out the errant bits of copper. Grrr.

Oh Bugger Moment 4. I bought IC sockets that stand off the PCB a little way, with the intention of getting the soldering iron tip underneath to solder the top side where necessary. This proved to be impossible. The best substitute I found was to put the tip of the iron on the component side of the socket until the pin had heated up enough that I could run solder in underneath. So much for saving time by making IC pins double as vias! It also melts the plastic of the chip socket slightly, making it look less tidy than it could be. I could always solder the chips directly to the board (and perhaps I will when I make the second ALIAC-2 board), but until I've actually tested that the board layout works, I don't want to risk soldering in the more expensive chips!

So far, just soldering the vias has taken longer to do than preparing the board, ironing on the toner, washing the board off, etching the board, tinning the board and drilling it.

I have to come up with a solution for this. Vias take way longer to solder than components, too - since you have to solder both sides, cut off the wire, and they have a nasty habit of sticking to the soldering iron and coming away as you remove it. I'll end up just making single side boards at this rate, it's cheesing me off so much.

My thoughts on this are to either drill them with a 1mm drill (perhaps increasing their size slightly), putting solder fragments through the holes, and baking the whole thing in the oven at 220 celcius. In theory, the solder should melt and the surface tension should cause it to then adhere to the pads. I might re-make one of my earlier smaller boards (perhaps the LCD interface, as it needs a design fix) to test this out.

Other than the smorgasbord of annoying problems I've had with this board today, there are some positives. Tinning the board is definitely worth it - the board looks much nicer, and is easier to solder.

But if the board actually WORKS after all of this, I will probably fall out of my chair . . . this is by far the worst quality board fabrication I've done so far!