Tuesday, 27 August 2019

Punch-Out! pcb repair

This board was mostly working but had a messed up background and no sprites on the bottom screen. The top screen was perfect apart from the missing logo.

Most times powering on the board it also had what looked like an interference issue.

As you can see (just about) the background is now fixed, this was a bad 74LS244 @ 3K on the back / small pcb.

The rest of the issues were on the middle / graphics pcb, first replacing a bad 74LS273 @ 1N brought back glass joe and the main logo.

I also found two 74LS04s with floating outputs @ 1J and 4E replacing these fixed the 'interference' issue.

Replacing two more 74LS273s @ 7H and 8J brought back the player (green guy).

This just left one small intermittent graphics issue with glass joe and the logo flickering and having black lines through them every so often.

I noticed some previous work on the board, someone had replaced the four 74F161s @ 1F,2F,3F and 4F with 74S163s. The one at 1F was rusty and when removing it from the socket one of its pins fell off, I soldered on a lead from a resistor just so I could test it and sure enough it failed.

I replaced all the 74S163s with proper F161s, they seemed to be working okay but I have a bunch of the correct F161 anyway.

In the same area of the board three 74F283s had been replaced with standard 74LS283s, I removed these and installed the correct F (fast) chips to finish off the repair.

I hadn't worked on a Punch-Out pcb before, this is the basic adapter I knocked up. It uses a Mike's Arcade Nintendo jamma adapter to invert the colours and amplify the audio. I basically copied what Adam @onecircuit came up with for his setup. :)

The switches are to flip between top and bottom monitors and the other switches between the two audio channels.

Saturday, 1 June 2019

Sega self test mistakes

Thought I'd note this quickly as it has caused some confusion on a couple recent board repairs.

Outrun - IC75 in self test is actually IC57.

GP Rider / Afterburner - IC30 in self test is actually IC37 (315-5248).

Friday, 3 May 2019

Missile Command - 4164 RAM upgrade

Purists look away.

Missile Command originally uses 8 x 4116 rams, and 4116 rams are terrible. They don't just die (a lot) they often take out other parts when they die. It's best to replace them with 4164s which run a lot cooler and are far more reliable, they also don't require +12v and -5v like 4116s.

It's already common practice to replace them on Williams boards but doesn't seem to be so common on Atari boards, not yet anyway.

Since there's no tried and tested mod for Missile Command I spent some time looking at the board to see the cleanest and most reversible way of doing the modification. You can mod the ram itself but this looks awful and not practical for some collectors to replace them where as anyone can drop in a stock chip.

Here's how I did it.

The cuts need to line up with the via on the back of the board where the 12v and -5v come through, if you hold the board up to light you can see it a lot clearer and mark where you need to cut with a sharpie.

Saturday, 23 March 2019

Outrun pcb repair

I did this repair a few days ago now and forgot to take certain pictures but I'll document what I can. This Outrun was watchdogging (resetting) and had a smeared red and blue image on screen which I forgot to photograph.

Firing it up with the extremely helpful test roms and a known working video board I could see a lot of bad rams on the CPU board, the watchdog was kicking in before the test would finish but I could see all I needed to for now. IC55 was failing every so often too so I replaced them all apart from IC114 and IC130 which are made by Fujitsu and hold up very well (unlike their TTL).

Swapping back to the original video board I could just about see that both palette rams (IC92 and IC95) were failing. With these replaced I got this image.

This was caused by two bad roms opr-10267 and opr-10268 if I remember correctly.

With these replaced the image was perfect apart from there being no road on the left side (again I forgot to photograph this).

The road issue was fixed by replacing a bad rom on the top board opr-10186 at location IC47.

This fixed everything graphically and the controls worked too.

Now the problem was the lack of sound, nothing completely quiet. The clock signal for the Z80 was stuck. Looking at the schematics I see the last thing I want to see.

That the clock is created by the 315-5218 surface mount custom chip. I tried a little experiment and created a 4mhz clock from the 16mhz clock using a 74LS74 and fed it into the Z80 to see if there was any life left in the custom but unfortunately there wasn't.

Here's the custom removed and pads cleaned up ready for the donor.

The new custom fitted, originally from a System X board.

Still no sound whatsoever but the clock was back and the Z80 was a lot more active. I decided to hook up the Fluke 9010a to the sound cpu (Z80) and see if the ram was good. First thing I do is a bus test and it reported A15 stuck high, this was caused by the main sound eprom epr-10187 which is interesting as the eprom programmer had no issue with it.

With the rom replaced there was still no sound, the fluke reported that the ram was fine so I moved onto the PCM eproms and rams. The two rams failed out of circuit testing and were replaced and the maskrom at IC70 failed and was replaced too.

This finally brought back the PCM sounds that were slightly too fast and getting cut off early but finally I could hear something.

I used a sound probe on the op-amp TL084 (IC125) and nothing was going in, this traced back to the YM2151 and sure enough nothing was coming out of it. I took a replacement from a System X donor board and all sounds were back and everything is working perfectly.

Repair complete.

Tuesday, 19 March 2019

Vectrex transformer repair

I've been after a Vectrex forever but the right deal never seemed to come along plus I didn't like the idea of having a vector tube sent through the post. A local one finally dropped in my lap and with a nice selection of games too.

The only problem was it was completely dead, checking the easy stuff first the fuse in the plug and the one inside were both fine. I could tell right away from the missing and mixture of screws that someone had already been inside before but luckily they seemed to give up when they saw the fuse was fine and they didn't damage anything.

Checking the output of the transformer and it was completely dead, a bit of research online and this is quite a common issue. I had a look for a replacement but the one maplin used to sell which is known to be an ideal replacement they no longer stock. I found another on ebay that was a good match but rated for 4amps where as the original is rated for 3amps (or possibly 2?) I was worried it would be too big and require some case modification that I'd prefer to avoid.

While continuing my search I found this very informative video here https://www.youtube.com/watch?v=zIqCVLveSys&t=1322s which shows how to repair the original transformer. It seemed that my transformer might be fine and the only issue is a bad thermal fuse hidden in the windings. You don't have to dig too far into the transformer to bypass or replace it either.

Here's my teardown, bypass of the fuse and rebuild.


The glue is all dried up so it was very easy to pry open, I dug into the insulation snipped the two wires for the fuse twisted them together and soldered them. Then a small bit of heat shrink sleeving (not necessary).

I sealed it all back up and fitted / soldered it back into the vectrex it fired up first time with no other issues.

Thinking about it afterwards I didn't have to tear into the transformer as far as I did, with the metal cover removed from the 18v side the two wires for the fuse are visible at the edge, I could have just soldered a wire or old capacitor leg across them or if I wanted to be fancy a new thermal fuse.

The thermal fuse really isn't needed though, there's already two fuses on the 240v side.

So there you have it, repair your old transformer don't replace it and a big thank you to Hugo B. on youtube. I don't think anyone knew about the hidden fuse until his video, I certainly didn't. :)

Tuesday, 26 February 2019

Block Block to Pang conversion

Block Block was Capcoms answer to Arkanoid and it's an okay version at best, a much better game on the same hardware is Pang / Buster Bros. This conversion is pretty straight forward but you will need to make a jamma adapter even if you have one of these.

This is the official Block Block jamma adaptor but it won't work with Pang as the controls are different.

Here's the pinout

This is how mine turned out, I left out 21,22,Z and a since eight +5v wires is a bit overkill.

Now on to the conversion itself, mine had been stripped of eproms and all markings on the board suggests non jedec type 27C301 eproms would be needed but actually four need to be the standard 27C010 type.

So 4x 27C301 4x 27C010 and 1x 27C256 you can obviously blank and reuse the original eproms if your board is still complete.

There's two options as far as which version of Pang to run you can either use the decrypted version from Arcadehacker posted here or you can use his method to reprogram the original Kabuki cpu that requires an Arduino Uno and a battery to be fitted to the board. I went for the easy first option but may try the other option at a later date to see how Super Pang handles on Block Block hardware (it should work fine).

With the no battery / decrypted roms option you still use the Kabuki cpu but you need to either have 28 sticking out of the socket or remove the resistor at R40 (which is what I did).

Decrypted roms
pang6.bin - 27C256 location 14F
pang7.bin - 27C301 location 15F

Now you need the mame rom pang.zip for the rest of the conversion. Mame usually has useful file names with locations but not in the case of pang.

pang09.bin  - 27C010 location 8H
bb3.bin        - 27C010 location 9H
pang11.bin   - 27C010 location 8J
bb5.bin         - 27C010 location 9J

bb10.bin       -27C301 location 2J
bb9.bin         -27C301 location 3J

bb1.bin         -27C301 location 2D

 Conversion complete.

Tuesday, 19 February 2019

Air Buster pcb repair

This repair has been going on for quite a while, first the custom for the I/O near the edge connector was missing (snapped off) when I got the board and the custom CALC1 chip was also bad. I bought a working Gals Panic off ebay for both these parts.

I also had to replace several missing and bad maskroms with eproms, this got the board running but it had a very annoying intermittent issue with the background flickering and disappearing.

Half the cloud is missing, it would also effect the buildings later in the level.

Probing around with my logic probe while the fault was visible I was eventually able to hear the issue. Thanks to my logic probe having a audio / beeper feature, I confirmed on the scope I was getting an invalid output from one of the 74LS352s.

The bad 352 removed and waiting for a new one. Any other TTL on this board I have plenty of spares but not a single 352 so I had to order some and wait.

 With the new 352 fitted the background is restored, repair complete.