Sunday 28 October 2018

Atari 400 Composite Video Mod

This is an old post, preserved for reference.
The products and services mentioned within are no longer available.

This is the first of a series of posts on upgrading some Atari 400 machines.
This first one is in a bit of a sorry state, so is going to need a good clean up before we're finished. The Atari 400 only came with an RF output, so the first job was testing it was working.
OK, so far so good. The PRINT FRE(0) test shows how much RAM is installed, here 13596 shows it is a standard 16K machine (more on that later).
Inside you can see there is quite a lot going on, on the right is the power input / RF output board, with the RF modulator on a separate daughter board.
In the centre is a large diecast aluminium box (like an Atari 2600), and inside that is a motherboard and two plug in cards.
One card is the CPU, the other 16K of RAM. The brother of this machine, the Atari 800 had five card slots, usually populated with one CPU, one ROM, and three 16K RAM cards, giving it 48K of RAM.
The first step here is to carry a composite video output modification. The modulator board removed, and reveals more of the board below. It looks like it was designed for a larger modulator module, similar to that seen here in an Atari 7800.
The 400 power board is quite complicated affair. At first glance, it should just take in the 9V AC power, rectify and smooth it, and then feed the 7805 to get 5V. However, it also needs 12V and -5V for the 4116 RAM chips, so has a complicated arrangement of charge pump voltage doubling diodes and capacitors to raise (and invert) the supply voltage to generate the other rails.
At the back of the board, there is a channel switch. Not sure why it's fitted on the PAL machine as it's not connected to anything. It does look like an ideal place for the composite video connection, but the heatsink is in the way.
I am planning use or the 'Deluxe Composite Video Mod' from The Future Was 8 bit, so the socket should allow the cable to be unplugged for storage.
The metalwork around it acts as a heatsink to two voltage regulators, a 7805 and a 78M12.
I could have cut a section out, but I thought I would take the opportunity to remove the need for the heatsink and install some switching regulators.
I wired it into the 5 pin header that used to connect up to the modulator board. It is important with mods like this to keep the wiring short. The cables are supplied quite long for cases where that is required, but in general, cut them as short as you can, and keep them away from anything noisy.
The composite mod cable plugs in neatly at the back of the case when it's reassembled.
The ground, 5V and composite video signals where there already. I wired the white wire for the audio. That's where it gets a little complicated. The Atari has three audio sources. The first drives the built in speaker - you don't get speakers like that in computers these days. This is mainly used for things like the loud beep when you switch on with start pressed to load a tape.
That is not connected to the TV output, so will be ignored. The second is the normal audio output, and the third is the audio from the cassette deck. The Atari uses stereo tapes. The left track contains the data, and the right channel on the tape sometimes has loading music, diagloue or silence (or if the tape duplicate company has cheaped out and used mono equipment, data).
None of this makes it to the RF output board, the two sources are mixed and then added to the combined video signal that feeds the modulator. To get around this, I have made use of one of the spare pins on the 5 way header, this connects to a track that goes to a space on the board where a vertical phono connector would have been installed for the RF cable, on boards with the larger modulator module that didn't have it's own output connector.
I fitted a 0.1" pin header in the footprint of the phono connector, and will tap the audio and connect it there, to allow the boards to be separateid in future as this is quite a modular system. (N.B., note the top quality Nichicon and Nppon Chemi-Con capatitors, these all measured well in spec and will likely still be superior to a many modern replacements, so there is absolutely no need to replace them, and I will shout at anyone who does)
And yes, that is an elastic band. The microswitch is used to detect if the cartridge cover is open, and turns off the power if it is. This was for testing without any of the metalwork.
The next step is to split the audio and video, and it's is remarkably similar to the 2600 and other Atari systems of that era. You may recognise the large red inductor from those systems. As with them, that and the associated audio subcarrier oscillator transistor below are removed.
Also removed is a capacitor (C183) that joins the audio and video sections. The white wire taps off the audio and will connect up the the output / power board.
The wires can run out a channel in the metal casing. you may just be able to see it at the bottom of this photo.
That then plugs into the pin header on the power board. Note the resistor below, R202, is also removed as this was apparently used to bypass the 5V regulator. This is no longer required with the switching regulator, and it just gets hot, so I removed it.
The microswitch I was holding down during testing has a plastic piece attached which disconnects the power when the lid is open (and the cartridge could be installed or removed).
When the lid is closed, power can be applied.
With no cartridge installed the Atari 400 brings up the Memo Pad.
This allows you to type in messages, which will be lost when the power is removed, so it's sort of pointless. On later systems this would be replaced by the built in self test.
To use BASIC, you need to install a BASIC cartridge.
But soon you get sick of that and install a different cartridge.
The 400 can should be able to play the entire range of cartridges for the 8-bit Atari range (not sure of any that it cannot).
A lot of which are very good ports of classic games.
However, when it comes to tapes, it is limited by it's 16K RAM, and a lot of cassette games will not load.
I've been testing a lot with a few games I found that did load in 16K. This is Desmond's Dungeon.
In the next part, I'll see if I can address that and give the 400 a RAM upgrade.

Sunday 14 October 2018

Toshiba HX-10D Japanese MSX

This is an old post, preserved for reference.
The products and services mentioned within are no longer available.

Following on from my previous Toshiba HX-10 repair, here is an HX-10D.
This is a Japanese NTSC-J model, but otherwise very similar at first glance to the HX-10. Other than the fancy red colourscheme, I was expecting it to be the same machine inside.
They are both 64K MSX 1 machines, but the HX-10D, being a Japanese machine, also has a different keyboard.
The HX-10D has an extra key below return to enable / disable the additional character set, and an LED alongside to show the current state.
I had wondered what the recess was in the HX-10 moulding.
Turning on this mode allows the entry of additional characters. I hope I didn't write anything inappropriate. (the first command is the change the colours as I don't seem to be able to take decent pictures of white on blue screens).
Before I could take such a picture, I had to work out the external connections. As with the HX-10, this has two photo plugs for audio and video out, but I wasn't sure which was which.
Luckily it is marked inside.
I had expected the board to be the same, just without the PAL module, and with a different ROM and video chip, but the whole board was different.
The board was wider and laid out differently than the HX-10 shown below.
Even the chip count had changed, the large high pin pitch custom chip in the HX-10 isn't there, and instead there in an 8255, and maybe a few more TTL chips. I'm guessing it is the same Yamaha YM2149 chip under the heatsink, but I didn't lift it to check.
The other main difference is the lack of the parallel connector on the side, instead the tape connector is there.
This one again is being fired up to test out the new Rodman game by Misfit. MSX is one of the 8 supported systems, and I wanted to check a PAL and an NTSC system.
And away we go. Do you know, with all this testing, I'm getting better at this.
Rodman is on pre-order now at The Future Was 8 bit.

Sunday 7 October 2018

Atari XE System (XEGS) Repair

This is an old post, preserved for reference.
The products and services mentioned within are no longer available.

Following on from last weeks post on the Atari 5200, here we have it's replacement (sort of), the Atari XE System, usually called the XE Games System or XEGS.
This is really what the 5200 should have been. It is almost an Atari 65XE, without the keyboard, but because it has the Start, Select, Option and Reset buttons (in glorious 1980s pastel shades), it is compatible with a lot of Atari 8 bit software.
The fifth button is the power button. I had presumed this was some sort of soft power switch, a bit like the 5200.
The XEGS has most of the ports of the 65XE, including the peripheral port, so programs can be loaded from cassette or disk (it has been suggested that the XEGS was a way for Atari to sell off their excess stock of 1050 disk drives).
The power connector is the same as the rest of the range, the 600XL, 800XL, 65XE and 130XE, with which this machine shares a great deal of parts.
Unlike the others, the XEGS has phono connectors for AV out, and normally, a third phono connector for the TV out. However, this one is different. The plastic has been cut into a larger circle, but the metalwork is still only the size of the original phono plug.
Through that hole, I can see part of a DIN connector. Hmm. alarm bells are ringing here.
The unit is not working, all you get on the screen (via the video jack) is a black screen with a white line down the side. Time to have a look inside.
What lies beneath the usual Atari metal shield? In my line of work, the best result would be a load of dust, a board which hasn't been touched since it was assembled 30 years ago.
The worse case, is a board with evidence of a lot of work having been done. Guess which one we have here?
The first thing I looked at was the non-standard DIN socket. It's clear there was a modulator there, but it looks like they may have had pads on the PCB for an AV socket maybe?
Ah no. This is not a factory fit option. It appears to be wired up as S-Video, chrominance and luminance and audio.
Elsewhere, it seems someone has had a good go at fixing the problem.
Lots of chips have been socketed and possibly replaced. The POKEY is intact, but the CPU, Freddie and ANTIC chips have been removed and socketed.
The ROM and MMU are likewise desoldered and socketed, but at least the PIA and GTIA are intact.
Underneath, there is lots of evidence of this, flux residue, some scratching, and some white areas where the board shows signs of having been overheated (most likely during desoldering rather than in operation).
This is why I don't like boards that have been got at, because I now can't trust any of these solder joints. The problem may well be due to track damage, or missing through hole plating, or even the naff pins on this type of socket, particularly when used with desoldered chips. That capacitor also looks like it has been caught with the iron and the covering damaged.
My usual first tests are power, clock and reset. Let's see if they are all present. Power is good, around 5.0V on all the chips I tested. Clock into the CPU is good, but there is no clock out of the CPU. This output is buffered by a 74LS08. Oh look. Someone's already replaced that, with a rather dodgy looking chip.
I tried a known good CPU, and did get a clock pulse, and there was now a red screen.
OK, so that's a good start. One bad part identified. But still lots to do. The databus wasn't wiggling as it should. I checked for reset, and found two things.
Firstly, the reset button was missing, and secondly, the 555 timer which generates the reset pulse had already been replaced. Here we go again.
There was no reset pulse on power on, and even shorting out the pads where the switch was didn't generate one.
With a new 555, we now have a reset pulse. Who solders a bad 555 back onto a board? Has someone taken parts from this one to fix another Atari and put the assorted broken bits back on this one. Several broken Atari's maybe?
I also fashioned a replacement reset switch out of a right angled tact switch, which came out to about the right height. So now we have power, clock and reset working, but still a red screen.
The RAM was socketed (this time in nicer turned pin sockets). I check the pins for continuity, and those were OK, but I'm really not happy about the other sockets, time for them to go.
With the board cleared, I could see a bit of damage beneath, but everything tested through OK.
With new sockets on, things were looking a bit better, but with a set of known working chips, there was still a red screen.
I tried a Star Raiders cartridge, useful for running up on almost dead systems, and it did start.
Or at least it went as far as the title screen. I could select the mission type, but whenever I tried to start the game, it would do nothing. I thought this might point to memory corruption, so I tried new RAM, but that made no difference.
The one chip I didn't have a replacemnt for was the MMU. This is actually a GAL, so I programmed up a 16V8 with the appropriate equations and tried that. Nope, no change. I also tried the 74LS138 decoder in case that was causing the conflicts, but no.
There was still conflict on the databus. All that was left on there were the 6520 PIA and the GTIA. Neither of which had been originally touched.
Well, what do you know about that. the 6520 was bad as well. With that replaced with a W65C21N, the system booted up to the Missle Command screen. But it wouldn't start. Pressing reset brought up self test, but that wouldn't start either. Still problems?
Well, not exactly. Rather stupidly, I hadn't checked the actual switches. I had assumed since the reset switch had been removed, it was faulty, therefore the others would be OK. Wrong. Two of the others were also bad. That may explain why I couldn't start Star Raiders before. Always check the simple things first. I managed to get it running, but couldn't see the results as red and green boxes would look the same in monochrome.
I thought this might be something to do with the mods for the DIN video socket, so I removed the links to chroma and luma, and wired it as a standard composite video and audio output socket, so I could use the same cable as I use on the other Atari (and Commodore) computers. I traced the chroma signal from the outputs of the GTIA, and it looked good, but it stopped after a buffer transistor.
Ah, that would explain it. I fitted a new 2N3904 transistor and we had colour again, and the ROM and RAM tests all passed.
I was already sick of shorting the contacts to press start, so, looked for some replacement switches.
These ones were close enough, so I replaced all four.
The power button was OK, as it turned out it was a mechanical on off switch, rather than a soft power circuit as I had expected.
This was all looking good. I could now start and play the built in missile command game.
Since I had wired up the 5 pin DIN socket as something useful, I thought I would try to use it. It is currently fouled by the metalwork.
I marked out the section that was in the way.
A bit of Dremelling later, and that is clear.
The hole in the plastic case was already suitable enlarged.
With that all back together, time for some testing. The XEGS can play cartridges from the 400/800 and later XL and XE computers. How about a bit of Centipede?
As it has the XEGS has the peripheral port, it can also load from cassette. [grabs a Rodman tape from the pile]
As with many games, that is perfectly playable as it only needs the start / select / option buttons and the joystick. The XEGS has two standard 9 way D joystick ports, at the same 45 degree angle as the cartridge slot on the top.
On the other side, at an equally jaunty angle, is a 15 way D connector marked 'keyboard'.
Into that you can plug in the optional keyboard. As you can see here, mine came from a different system, one that spent a great deal more time soaking up those 80's UV rays.
With that installed, you now have pretty much a 65XE. It boots into BASIC and away you go. The keyboard is infact the same as the 65XE
You can see at the top where the start / select / option and reset buttons would normally be fitted.
Underneath is a controller board containing the mux and demux chips which address the switch matrix.
So, that's the Atari XEGS. Back up an running.
In the end it turned out it was only a bad CPU. Oh and a bad 555. And a bad 6520. Oh, and that damaged cap. One broken transistor leg, two bad switches, and one missing, and let's not even mention the 5 pin DIN connector.
Ah well, I am going to have to do some more testing, there are two versions of Rodman on the Atari 8 bit, one in full colour
and one in monochrome, but with higher resolution graphics.
Atari 8-bit is one of the 8 systems on the 3 tape special edition of Rodman, now available to pre-order from The Future Was 8 bit.