What a wonderful thing a week of leave is, especially if you have nothing planned, except your hobby.
I deliberately left myself plenty of time for just mucking around. I set myself three project goals, with some option “stretch” goals.
- Work out what was wrong with the floppy drive on the Archimedes.
- Decap the Amiga 600
- Print a 3D case for the Creativision Megacart
I got the Archimedes on the bench pretty much Friday evening and it stayed there for most of the week. The first thing to do was to completely strip it down to just the motherboard, the PSU and the CPU riser. That way I could have a look at what was going on with all the signalling.
I must put a big thank you to the StarDot forums, who, as always, were invaluable in their assistance.
Breaking out the logic probe and the schematics, there seemed to be 3 prime candidates that could be the cause of the fault.
- IC47, a 1772-02 Floppy drive controller chip. This was a very commonly used chip for this era. Also used in the BBC systems among others.
- IC38, a 74LS05 hex inverter. This is used as a simple inverter buffer. The prime candidate!
- IC29, a 74HC574 flip flop. Not quite sure what this is doing, but it ties into several key drive pins.
I started with IC47, and initially I thought this was the most likely candidate. I could see data going in but no data coming out. Turns out I was looking at the wrong pin! Oh well, the spare I have ordered will be useful somewhere else I am sure 😀
Next I moved onto IC38. Now these fail all the time, which is why I probably should have checked it first. The only reason I didn’t was a gut instinct. The logic probe showed this chip was happily inverting just as it was supposed to.
This left IC29. Now I don’t really understand how this chip is supposed to work, but simply looking at activity, I could see data coming in on the input side, but nothing coming out of the output side. Everything was locked. The Drive Select pins were all being driven continuously and the Motor pin wasn’t being driven at all. This could explain the lack of any activity.
Having a likely suspect, I carefully desoldered the chip, using my desoldering station to remove the majority of the solder, and my brand new hot air rework station to carefully loosen the remaining solder. Popping it out, I stuck it in my TL866 II+ and ran a logic analysis over the chip which promptly failed! That seems to vindicate my suspicions. Unfortunately it’s not a chip I can buy from any of the local suppliers, so I have one on order. In the meantime I’ve stuck a nice machined pin socket in there to make the next step easier when the part does arrive.
Amiga 600 decap
The Amiga 600 still had 4 capacitors I’d been unable to remove due to the limited room around the remaining capacitors. All four were relatively “flush” up against an existing component. I had attempted to remove those components, without much success, so I was at the point where I needed newer tools, and finally I had a hot air rework station.
As this was a new tool, I practiced on an old PC motherboard until I was comfortable with my work. I certainly managed to “pop” a capacitor that turned out to be through hole, not surface mount, but that board had loads of bent pins so I didn’t mourn it much. It did vindicate my basic approach. Slow air, 380 degrees for leaded solder, lots of polyimide tape and aluminium foil insulation around the temperature sensitive components near where you are working, and slowly slowly does it.
I gently heated the general area, trying to heat the board evenly as to not warp it, and at the same time raise the temperature of the components. Eventually they came free with only a light pressure on the tweezers I was using to gently lift the components with.
With this technique I was able to remove the remaining four capacitors. Replacements are on order. Soldering these components back in? That’s a problem for “Future me”. I suspect a delicate tip on the iron and, again, loads of polyimide tape and aluminium foil.
3D Case for the Megacart
This one was mostly me in front of a computer with OpenSCAD and lots of trying and trying again. The Madrigal Design forums had several people assisting here, with a BIG shout-out to Tom over there who managed to print a cartridge before I’d even had a chance to get mine on the print bed. It allowed me to rapidly iterate two new versions based on his feedback. I’m already contemplating yet another iteration, to add a way of securing things a little more robustly.
Once I had the designs in OpenSCAD, I was able to export as STL, then feed them to my slicer and begin printing.
These are a little too big for my resin printer, so I need to print one part on its shortest edge and another on it’s side. This means the prints take 8 hours and 5 hours respectively. If I had a bigger print bed I could print them directly on their faces, which would drop the print time down to much closer to under two hours.
(If it’s not obvious, the slowest part of printing with a resin printer is resetting the z axis. Anything that is larger in the z axis takes longer, so you usually try and align your models to be largest in the x and y axis)
The quality of the top was… OK. I think I need to get better at leaning my models. There’s some smearing that I think is improperly cleaned resin that rehardened.
Unfortunately the base failed completely. I need to review my print settings and reevaluate to see what went wrong. I don’t think it bonded properly to the print bed.
The moral of today’s episode seems to be – You don’t need additional tools until you do, and then you wonder how exactly you lived without them beforehand!
Now I have hot air, I think the next tool I will want to get will be a proper bench power supply with adjustable Amperage, Voltage and probably two voltages at once. That may be something for next year, though.