Unbreaking things

So having thoroughly broken both my Atari 800xl and my CreatiVision, I decided to spend some of my long weekend fixing them.

Atari 800xl

I started working on the Atari 800 first. This was actually several days work in the evenings as it got out of hand, as the Atari forums people got involved with helping me determine exactly what was going on with the RAM extender.

The first thing I did, and the thing I suspect most contributed to the stability of the system, was to remove the ROM tower. It had the original rom and 3 Machine Code Monitors. I’m never going to do anything other than play games and run demos on this system, so out comes the ROM tower. Unfortunately that didn’t help as much as I’d hoped and it became obvious why pretty quickly. The socket was bent so badly from having the tower plugged into it that the original ROM just popped out. The tower used square pins to connect, so had pushed it out of shape. It was also a point of weakness as the whole tower board acted as a lever.

My desoldering station made quick work of the pins without lifting a single trace and I dropped in a nice turned pin socket in it’s place. The old socket disintegrated as it came out so I’m quite confident about my “root cause analysis” here.

It booted up fine, which is always a good sign, but I wasn’t happy with the RAM expansion sitting like a dirty spider web in the middle of the system. By this point, people were getting involved with working out what was going on so I desoldered the board completely and started tracing and documenting the board.

The board was held in with double sided tape which took a long time to clean off both the motherboard and the RAM board. Once out, I tried desoldering the chips to assist in the tracing, but the board had been constructed in a peculiar fashion, with all the components soldered in from the top side of the board, making easy removal with my desoldering gun basically impossible, and after accidentally removing a trace I gave up and started tracing visually and with a Multimeter.

After getting it most of the way there (I have two more links that I need to correct) I decided it was time to put it back into the Atari. I started by carefully cutting some insulation sheet (Clear plastic sheet designed for use in binding documents). I then fixed up the capacitors on the back of the board. Two of them had failed as I cleaned up the board so I grabbed some similarly specced ceramics and dropped them across the rails in the same place. Once done I attached the insulation sheet with double sided adhesive foam, attached in the corners so as to not impinge on any of the underside components or soldered points.

I decided to make the rejoining of all the wires as clean as possible, so starting with the innermost wires, and routing and trimming as I went, I rerouted the board back in. Once it was all in, I cable tied some of the bunches of wires, applied shrink wrap to add mechanical stability to the “enable/disable” switch and even taped some wires down with polyimide tape.

Thankfully it fired up and passed all tests. It seems stable and I lost about an hour to playing an old classic game.

I’m still waiting for a replacement screen for my SDrive Max, as my current one is marginally incompatible and shrinks the image and mirrors it, which is not pretty 🙂
It’s functional, but not something I’d want to use for any length of time.

CreatiVision AKA the Dick Smith Wizzard

Next I needed to work on the Dick Smith Wizzard (Henceforth DSW for simplicity) and while I fixed one issue, I promptly introduced another.

The first fix I almost tripped over by accident. The DSW has been exhibiting an issue of distorted screen output and a tap is usually enough to fix it. I had stripped the machine down to bare and was checking it when I realised the place I needed to tap was the on/off button. It was corroded, which, considering the system’s age, should not have been a surprise. I sprayed some Deoxit in the switch, and worked it back and forth a bunch of times. It’s been rock solid since… right up to the point where I broke it completely with my next “fix”.

So this system uses a 7805 regulator to generate the 5v rail, and I’ve had success replacing them with 78E5s in some machines, which are embedded switchmode drop in replacements. Noting the very large heatsink attached to both the 7805 and the 7812 regulators (The latter being for 12v) I thought I’d swap out the 7805 so the system doesn’t run as hot. I grabbed out a 78E5, carefully desoldered the 7805 and… it wouldn’t fit! The legs on the 7805 had been bent into an angle to allow it to clear the 7812.

Gah!

No problem. I just extended the legs of the 78E5 so it was raised over the top of the 7812 and gave enough clearance. I used some pins I’d got in an electronic grab baggie and extended it up and above. I made sure there was no contact with the heatsink or the 7812 and it all looked good.

Except the system wouldn’t power up. It wouldn’t even begin generating video.

With a sinking feeling I measured to see if there was a hot chip. Nope. Next I started probing voltages around the 4116 RAM chips (These use +5, -5 and 12v, and thus are a good spot to check for problems). Sure enough the +5 rail was sitting down around 2.5v. Odd. I checked on the underside at the output of the 78E5 and it was delivering 5v so it was slumping across the board.

I was quite confused by this point, and in desperation, I soldered back in the original 7805, making sure I put in the mica insulator and fired it up. Worked first go. I can only assume that the 78E5 can’t generate enough current to support all the chips at full voltage and was thus causing the slump.

Oh Well.

I refitted everything and it’s been fine since.

Conveniently Clockmeister had also sent me a new version of the ROM image he developed for the MultiCart, that fills the last 5 banks, so I decided to write that out to the ROM chip in my cart.
I have burned ROMs in the past, but not with much success, so I approached this with some trepidation. I removed the ROM from the cartridge, peeled off the silver sticker over the UV window, (Which turned out to be a “void if removed” sticker. Thankfully I was able to remove the residue with some Orange Power goo remover) and stuck it in my UV eraser for a 4 minute erase cycle.

Once done I simply dropped it in my TL866 II+, told it which chip I had, told it to ignore the internal chip check (Mine is not exactly the same chip as they list, but it’s usually safe to pick something of the same family) and burned it across. After a few minutes, it was done, and verified OK so that was hopeful! Back in the cartridge, and back in the DSW and it powered up first go. Yay!

I covered up the UV Erase window with a little square of aluminium foil covered with black electrical tape. Not as “permanent as the “void” sticker, but I can at least safely and cleanly remove this.

I’m still working on a replacement keyboard solution for the DSW. I finally managed to locate a suitable key that, while expensive, won’t entirely break the bank. I’ll probably order one to begin with and, once I’m happy, order a bag of 100. I’ll also need to find some way to get keycaps to go with the switches.

This is going to be a VERY expensive keyboard 😀

Apple II Joystick

I now found myself with some spare time. I figured the time had come to make a joystick for my Apple II systems, learning from everything I’d picked up making the BBC Master Joystick.

This wasn’t actually a terribly difficult project, but it had one gotcha.

I was basing it on the design by Quinn of Blondihacks, which is an excellent design. I added a variable trimpot to allow a small amount of fine tuning, and I used an “off the shelf” box to contain it, the same as the one I used on the BBC Joystick.

Parts were:

1. Joystick – It’s cheaper to buy it mounted on a board than it is to buy it as a standalone.
2. DE9 M connector. (I accidently got DE9 F because every other system uses this not the male)
3. A suitable case. I always use this case for these kind of joysticks, as it’s an affordable case. I wish it didn’t have the battery holder as I just end up cutting into that.
4. A DE9 Hood. I had one lying around. Actually I had 2, which was lucky because I ruined the first one.
5. 6 x 100nF and 2 x 10nF ceramic caps. (I had some spare which was lucky as I missed 2 when I bought them)
6. 2 x 1MΩ trimpots. I add these in series with one of the 100nF capacitors to allow me to tune the capacitance.
7. 2 x 470Ω resistors. (Forgot these. Luckily I had some in my stash)
8. Some 6 strand cable. (I used leftovers. I think the cable is actually off a Polycom microphone extension. It’s really nice 8 core, untwisted in rainbow colours)
9. Some experimenters board. I had some left over from the BBC Joystick.
10. A DPDT switch. I had one I’d bought because they fit as drop in replacements for C64 power switches, so I’d bought lots.
11. 2 x momentary buttons. I bought the wrong ones (Too deep for the case) so I scrounged some from my stash.
12. Various standoffs, fasteners, cable ties and other things to make it all look neat.

Assembly went fairly well, but took a few hours.

I started by drilling a hole in the top of the case for the joystick to poke through. Last time I mounted the joystick then made a hole. This turned out to be a bad idea as drilling a hole that big introduces a lot of variability, so this time I added the hole first.

I then positioned the joystick by sticking it down with some blu tack, sliding it to the “right” location, then drilling through holes into the base for the mounting screws.

I drilled a hole for the cable exit and for the DPDT switch. This switch is wired so that it flips the left and right fire buttons, just in case you need them flipped.

Out came the experimenters board and I simply soldered up everything. I crimped on a dupont connector for the connection to the joystick, just in case I had to make changes.

I had to drill through the integrated battery compartment to fit the pushbuttons.

Once I had it all assembled, I tested it, only to discover that X and Y were both inverted. Simple to fix, I just rotated the joystick 180° and screwed it back in. Oddly I’d initially had it this way round, but rotated it because I thought I might have flipped it when mounting it.

Once that was done, I did some important testing.

All in all, a productive long weekend.