Converting a PAL TRS Colour Computer to NTSC

When I got my CoCoSDC, I was surprised how many of the games out there relied on NTSC Artefact colour to generate either some or all colours. It was close to 80%, and even higher in the more commercial games.

After a while I decided I wanted an NTSC system. As I had 3 CoCos, it seemed silly to buy yet another system. Instead I decided to see what was involved with converting from NTSC to PAL.

Asking around, no-one seemed to have already done a conversion, so I went to the schematics and started comparing.

Initially, the most obvious change was there were two new chips on the PAL system. To wit: IC3, the TCC1000 and IC4, the MC1372.

Looking at these, the MC1372 was a video encoder chip, and I knew I wanted to keep that one as I was going to tap Composite shortly after that chip did it’s conversion. (More on that later)

Researching the TCC1000 was difficult. There doesn’t appear to be any official documentation on it, except in relation to the CoCo, and even that was vague. All I could find out was it was involved in converting the video signal from NTSC resolution to PAL resolution. I believe it was involved in generating extra lines for the larger PAL line count.

At this point I dropped a comment in the TRS 80 Color Computer channel on Facebook, with a suggestion that I could just try bypassing the TCC1000. There was a lot of interest (Hi Everyone!), no one could see any immediate reasons not to try it, so I began.

First step was to add a Composite out, as there was no way I was going to use RF. I desoldered the RF Encoder box, built a single transistor video amp with bits I had lying around, and tested.

I was getting Composite video out OK so it was time to move forward.

I desoldered the TCC1000, and dropped in a turned pin socket. I accidently also desoldered IC13, so I tested to make sure I hadn’t broken anything. It was still fine.

Meanwhile, Clockmeister reached out to me to say he’d already tried this mod, and had hit problems with picture stability that he thought may be related to timing. The NTSC ran at a different clock frequency to the PAL system, so that was something to look at before I went any further.

The NTSC system has a single clock of 14.31818MHz, whereas the PAL system has two clocks, one near the MC1372 at 4.43618MHz and a master clock of 14.23753MHz.

I went down a bit of a dead end at this point, but a productive one. I had it in my head that the MC1372 chip was running on the wrong clock frequency, and by going over the schematics, I somehow decided it needed a different frequency, so I swapped the crystal for a 3.5795MHz crystal. I was half right but all wrong.

I couldnt source a 14.31818MHz crystal locally, but luckily I was going to Artifactory, and when I checked in their giant bin of parts, I managed to score four of the correct crystal. (I plan to return the unused one for other travellers. I only grabbed as many as I did because they were all clearly used, which makes them a little more likely to be DOA)

Many Crystals

I desoldered both crystals, and carefully replaced them with the new crystals.

I then decided it was time to start working on the bypass. I got a second turned pin socket and built up an adapter with some bypass wire. Looking at the schematic for the PAL CoCo, the TCC1000 has a whole bunch of paired inputs/outputs, so naively I started bridging across the pairs.

Pin 21(ØA IN) to Pin 10 (ØA OUT), Pin 5(Y IN) to Pin 6(Y OUT), Pin 7(ØB IN) to Pin 8(ØB OUT), Pin 19 (GTCLKOUT) to Pin 2(CLKIN) and Pin 25(HSIN) to Pin 17 (H-SYNC)

Line and lines and lines and LINES! What do they mean, Edward?

(There’s also a factory mod wire from the H-Sync Pin to Pin 40 on the MC6822, so I patched that back in to the jumper)

Once I had that all together, I thought it was time to fire it up.

Amazingly, nothing exploded, and I got a picture. It was extremely shimmery, with the whole picture showing waving around the horizontal plane. The colours were also incorrect.

This is also doing a side to side shuffle

Undeterred, I loaded up Donkey King. The solid start page was a giant shimmery mess, and when I got into the game, all the artefact coloured objects were exhibiting a rainbow effect.

Rainbow!

But I got a picture! This was further than I expected to get on my first try.

Ciaran, over on the CoCo FB group, pointed out that the NTSC CoCo only had the one crystal, and it was all keyed together. I realised he was correct, and I carefully lifted pins 1 and 2 on the MC1372, and fed the clock directly from Pin 19 on the TCC1000 socket. This means that secondary crystal? It’s totally bypassed.

Bodges and patches oh my!

Plugging in again and… I have artefact colour! The backgrounds are still all wrong but Artefact colours look good.

Woo! More progress!

Once again, Ciaran pointed me to the next step. In his words:

so i’m not sure why all colours would be off in normal alpha/sg/colour graphics modes. that sounds like a thing to fix 😉

but probably the next thing you need to bear in mind is that NTSC cocos have a 555 timer configured to pull øA low during the colourburst – triggered by GM0 being high (i think – it’s for the otherwise mono high resolution graphics modes) and hsync, but only lasting for as long as the colourburst. this is needed to adjust the phase of the colourburst so that the alternating high/low luminance signal is interpreted as red/blue colour instead of green/purple

definitely consult the schematic for this!

OK. Yes, looking at the schematic for the NTSC CoCo, there’s a whole additional circuit in there. Thankfully everything was available locally and I grabbed out a little experimenter’s board and started building up the circuit.

I’d also managed to find some video of what is going inside the RF Modulator on the NTSC model, and interestingly, it has an MC1372 inside the can. The pins were pretty much directly connected to the pins on the side of the RF Modulator, which gave me more confidence I was on the right track.

It didn’t take me long to get it all assembled. It would have been faster if I’d noticed that the Capacitors I had on hand were not, in fact, 22nF capacitors, but actually 22pF. Oh well, I was going to Artifactory again, and once again, the giant stash of parts delivered.

These little experimenter’s boards are brilliant for this kind of work

Now I had it all assembled, it was time to look at mounting it, and connecting all the wires. I ran out of colours for my patch wires, so I ended up putting stripes on one of the yellow wires to differentiate it.

To raise the board high enough, I grabbed some cheap knockoff construction bricks we’d got for free from a supermarket promotion, and added some double sided tape.

As I started to add the board I realised I was going to do “something” with IC13 after all. This appears to be an Op Amp and I think it may be an inverting type.

The output of the transistor on the mod board was going to ØA, which promptly goes through Pin 12 of IC13. The NTSC board doesn’t have that Op Amp. I tried bypassing the op Amp by removing it and jumpering from Pin 12 to Pin 1 of the socket and…

Green desktop.

Ooh. Let’s try Artefact Colour!

Blue Kong. So close, yet so far.

Nope. OK. Back to installing the Mod.

Mod in place.

Patching the wires directly to the various locations, I carefully fired it up and…

More green. So green.

Well, that’s good, right? Let’s try Donkey King.

Brown King. Blue girders.

Oh wow! I have it right at last!

So the steps, for those following at home are:

  • Change the primary crystal for a 14.318181MHz crystal
  • Install the Bypass in the position of IC 3. This consists of Pin 21(ØA IN) to Pin 10 (ØA OUT), Pin 5(Y IN) to Pin 6(Y OUT), Pin 7(ØB IN) to Pin 8(ØB OUT), Pin 19 (GTCLKOUT) to Pin 2(CLKIN) and Pin 25(HSIN) to Pin 17 (H-SYNC). You’ll need to patch a couple of these to additional signals.
  • Install a shunt in the position of IC 13. This goes from Pin 1 to Pin 12.
  • Remove IC 4 (MC1372), bend out pins 1 and 2. Solder Pin 2 to the Bypass on IC3. It needs to connect to the bridge between Pin 2 and Pin 19.
  • Build the Mod board. This consists of:
    • A 555 Timer.
    • 2 x 22nF ceramic capacitors
    • 1 x 1nF ceramic capacitor
    • 1 x 9.1KΩ resistor
    • 1 x 4.7KΩ resistor
    • 1 x 2.2KΩ resistor
    • 1 x 1KΩ resistor
    • 1 x 2SC945 transistor (I substituted a 2N3904)
    • 1 x 1KF20-04 gemanium signal diode (I substituted a 1N60P)
  • Wire it up as follows:
    • Yellow connects to Pin 30 on IC3 (MC6847)
    • Blue connects to Pin 39 on IC3 (MC6847)
    • Yellow/Black connects to the shunt wire between Pin 1 and pin 12 on the IC 13 socket.
    • Green connects to the Bypass on IC 3, between Pin 25 and 17.
    • +5V and GND can be tapped from wherever you find convenient. I tapped the empty pins on the Bypass. (Pin 15 for GND, Pin 28 for 5V)
  • Add some sort of Video Out mod. I did a simple 1 transistor video amp, tapping out from the leftmost pin where the RF Modulator used to go.
Another shot.

So that’s it! Hope this helps someone else.


Posted

in

by

Comments

2 responses to “Converting a PAL TRS Colour Computer to NTSC”

  1. Dirk Avatar
    Dirk

    Which model CoCo were you using. I have a PAL CoCo 1 and have not found a TCC1000 in it 🙁

    1. Cheshy Chesh Avatar
      Cheshy Chesh

      Mine are CoCo 2 units. 64K. I know the video circuit is quite different between the CoCo 1 and the CoCo 2.

Leave a Reply

Your email address will not be published. Required fields are marked *