Ghosts n' Goblins - Repair Log

[biglouie]

I have a buddy that lives in Argentina, so for fun we set about finding some arcade boards.
We found 10 and shipped them. A saga involving his mother taking them to Italy between lock downs, driving them into France, several wrong customs paperwork SNAFUs and DHL using the box as a Piñata. Not important, but a box of boards with chip shop scraps layering the bottom arrived.

I have started to learn how to fix boards with the help of some far cleverer people than I. I am massively grateful to those people for passing on their experience and being so generous with their time and patience!

Board one is Ghosts n' Goblins. I am fully aware that this is not a good board to learn on for ALL the reasons. I've put it away a few times, but I feel like I am learning the board and switching to other schematics once you get deeper in feels a little painful.

Above after dust down and replacement of Caps

On first visual inspection, the board is in reasonable shape, brushed off the dust, nothing really corroded on ICs and wotnot. Many Fujitsu ICs, which from what I've watched/read, these are often problematic. Nothing much has been socketed prior, no jumpers on the board.

I've replaced 4 or 5 Ceramic caps that sit next to the ICs, they presumably smooth things out and reduce noise etc... 104s and 101s. The most damaged part of the board is corroded Tantalums on the edge of the board. 4 or 5 have gone and left corroded pins in the through holes. I haven't been able to persuade them out for now. I need to order those and revisit.

I checked the 5v and 12v circuits, to see if I get resistance. Seems ok and not open. Classic Capcom pinout, used the adapter. Plugged it in, changed the power on the switcher to give me 5v on the board. It was 5v on the switcher, 4.3v on the board, so I juiced it. Checked 5v on an IC far far away from the edge.

STATUS - bit of garbled colourful "stuff" on screen

I pulled the EPROMs, few tested bad pins, cleaned pins with Fibreglass pen, got a few working, pulled the contents, checked against MAME using Hamster. It seems to be GnG US World 1, but it gives me plenty of options on the ones it does find. Some of the BINs don't check out at all and find nothing, so I assume they are bad. I have some 128 and 256 EPROMs that programme at 12.5v so use those, because what I did not realise (because I have never done anything like this before) is that the new TL688ii plus Advanced Universal Programmer I bought, won't burn these older EPROMs, they want 21v to burn.

Burnt EPROMS, all checked out with GnG US World 1. Old ones below, new ones on the board.

STATUS - bit of garbled colourful "stuff" on screen

I removed the bottom board (it runs to title screen without the graphics board b I understand) and went to look at CPU. It's a 68B09.

Logic probe out.. We're closing in on limits of my current knowledge.. Had a quick look at the MC68B09 data sheet HERE

Watchdog (Pin 37) shows HI, no pulse. Which I believe is good, as RESET has a line over it in schematics, which I understand from the data sheet means that it is "Active LO". If it's LO it's active. Mine is HI, so not resetting.

Next I thought I'd check the clock Pin 38 & Pin39 are both clock related. I don't really understand this bit yet, I know the this variant has an internal clock, I had a quick scan of the MC6809 Cookbook . I have discovered I am too stupid to understand it all. Pin 38 EXTAL pulsing. 39 XTAL - Not.

I think I have clock, but not really 100% sure, as the internal clock of the CPU thing I did not understand. I had a quick probe around the rest of the pins while I'm here. Address Outputs Pins 8-22, all pulsing apart from Pin 10 & 12.

So I thought I'd go and look at the external clock/crystal on the board. It hangs out over near 3N, top right of the board. Maybe thats not working.

12000 crystal, all standard from what I have gleaned, pic above shows after I replaced the LS04 (spoilers)

Nothing pulsing on the crystal. Nothing pulsing on the resistors up there at R43-R45, one side of R45 is dead, everything stuck LO. Decided to check the LS04 at 3N as its part of the circuit. The biggest frustration I have at the moment, is that I don't really know what I am meant to be seeing.

No pulsing. Piggy backed an LS04, nothing pulsing. I had a 12,000 crystal, so I piggy backed that (dunno if that is a thing). Turned off and on, things started pulsing at Pins 1 & 2 and Pins 9 & 8. Removed the piggy back crystal, things still pulse. So assume it's the LS04 and nothing to do with crystal.?! Maybe it's temperamental.

Pulled the LS04, new socket, popped in the new one. Pulsing.


STATUS - thick White Lines on screen

Tested the LS04 in my Logic programmer & tester (TL688ii will test 74xx Logic chips, which I didn't know when i bought it), and it came back as GOOD. So I popped it back into the board, just to see. No pulsing. So the tester found the LS04 automatically, tested it and found it was good, but it doesn't work in the board.

So LS04 at 3N is now giving me some pulsing and in turn providing clock to an LS74 at 3K and elsewhere.

Working my way back through the schematic from here now, after some advice from afore mentioned experts. Lots of GnG repair logs talk about 2K, 1K, 3L and 2L being bad due to Fujitsu LS161. I check, lots of dead stuff here, all floating with no HI or LO or pulse, so I replace all 4 of these LS161, they seem to be counters for the Horizontal and Vertical data. 2 check out as bad in the tester.

Now they all give me data and pulsing at Pins 11,12,13,14 and the screen goes from bars to block, white screen, which I feel is progress!

I like all the counter outputs and pulsing and things seem healthier. I read that the white screen is the CPU unable to run, so maybe it isn't getting data. That info was just one post in a very long thread though and it didn't seem to get any acknowledgement...

STATUS - Block white on screen

The thick dark line is just the refresh obv.

So if it's not resetting watchdog and clock pulsing and a few ICs have been found dead and been replaced and now show outputs, I thought I'd go look at some Address lines Pins 8 through 23 on the CPU again. If you recall, everything was pulsing, apart from 10 and 12 which were stuck HI.

Now EVERYTHING is pulsing. Which I think is good..Welcome back Pins 10 and 12. Always been my favourite MC68B09 pins.




Had a little feel around the board, nothing is hot, no specific boiling chips. Don't know what this means, thought I would mention it.

Now I have pulsing on the Vertical and Horizontal counters (tips of blue triangles below), I take advice and start working back through from here, output lines from each. Red circles. I think thats where we go now!

RED CIRCLES

2J - An LS10. 3 inputs on Pins 3,4,5 into an NAND gate.

All pulsing - HI, HI/LO, LO - Output on pin 6 is HI

Correct right? Anything but 3 HI is HI

4L - An LS08. 2 inputs on Pins 9 & 10, into an AND gate.

LO pulsing, HI pulsing, into Pin8 output which is LO pulsing.

Correct right? 0 + 1 = 0

So thats where I am to date... Along way to go, but I didn't have a logic probe or a copy of Don Lancasters TTL Cookbook last week so I've come a long way!

Any suggestions or advice more than welcome.

biglouie2021-03-25 14:16:54

 

[yoganuggy]

Looks like you are having fun.

A very tricky aspect is removing the LS chips from the board cleanly. It's very easy to damage a trace if the desoldering gun is too hot or held to the pins for too long or the gun nozzle hole is too big (IMHO it really needs to be 0.8mm and most guns are 1.0+ mm).

Check the continuity on the pins to the traces of all chips you have lifted and confirm they are good. Keep in mind there may be more than one trace leading from a pin.

 

[Lurch666]

A point about your lack of clock on pin 39 of the CPU.

If you look at the schematic it's tied to GND so you wouldn't see any activity on it.

 

[biglouie]

A point about your lack of clock on pin 39 of the CPU.

If you look at the schematic it's tied to GND so you wouldn't see any activity on it.

Perfect I see that now, thank you. So is XTAL the internal clock for the 68B09? It is indeed, LO for me.

And EXTAL the clock over at XL on the board?

 

[biglouie]

Yep, having lots of fun, some frustration.

The desolder gun takes them pretty cleanly, I add new solder and then desolder with the gun. Then it's a tidy up job with some braid so the new sockets can pop in. I'm buzz testing each socket once they go in, I dont enjoy the jeopardy of that AT ALL..

 

[Lurch666]

A point about your lack of clock on pin 39 of the CPU.

If you look at the schematic it's tied to GND so you wouldn't see any activity on it.

Perfect I see that now, thank you. So is XTAL the internal clock for the 68B09? It is indeed, LO for me.

And EXTAL the clock over at XL on the board?

EXTAL would stand for external crystal.

 

[Lurch666]

If a boardset is not running then you need to verify the program roms (already done),the ram and the associated chips.

So the program roms are at 8N,9N,10N,12N and 13N and I'm assuming the CPU ram is the 62647 at 2C.

The thing with ram is you can see activity on the address and data lines but the chip can still be bad.

Unless you have a way to check in circuit replacing it with a known good chip is the only way to be sure it's working properly.

If that makes no difference then look at the other chips on the same page as the CPU in the schematics.

Even with the graphics not showing properly you should be able to discern if the game is actually running. Have you got the sound connected because sometimes that can show if a game is running blind.

Although I have never repaired a G&G so I could be wrong.

 

[biglouie]

Magic understood re EXTAL.

 

[biglouie]

Confusion Questions:

RAM @ 2C - Schematic calls it a 6264

The chip looks like this, with no reference to 6264 on the top.

I'm curious, why? Is it not complicated enough without random naming? Or am I missing something?

M5M5165P - Datasheet

I cant see mention of 6264 in there..

 

[Lurch666]

You will find with ram there are several different numbers but they are a match for the type of ram.

http://pcbjunkie.net/index.php/resources/ram-info-and-cross-reference-page/

 

[biglouie]

Got it, thank you.

 

[Shinydaz]

I've just repaired one, it was showing static garbage, desoldered the ram @7h to fix it.

 

[biglouie]

I have been through the 6264 RAM with the probe, all pulses. But noted Lurch666 comment, " that doesn't mean its good". I have ordered some spare RAM.

My next possible find... I was following back from the 68B09 from Pin34. A little tricky, this is the top of the page of the schematic.. It brings me to 7K an LS08, 5K an LS04 and 9L an LS32

This is where I try to use my logic, but my understanding fails me...

Here is the schematic for the area I looking at. 7K Pin1 is DEAD. 5K Pin1 is DEAD. 9L Pin 10 is LO.

The gate top right, its a NOR gate that's inverting inputs on 1 and 2?

BUT in the TTL Cookbook, a 7408 is an AND gate

So I don't get why schematic shows 7408, but NAND gates..

And if Pin1 is dead on the 7408, then Pin 3 does what? In this situation, I'm getting HI & LO and Pulse going crazy, super fast...

I have piggy backed the LS04 at 5K, and Pin 1 still dead. so I think I need to replace 7K LS08. I don't have one.

Don't understand the AND and NAND thing above..Schematic shows NAND gates but TTL Cookbook shows AND for LS08.

Confused from Kent..

 

[biglouie]

Thanks for mentioning that ShinyDaz. M58725
Nothing floating for me there at 7H

12 lo - Correct

18 lo - Correct

20 lo - Correct

21 hi - Looks to be Active Lo WE (don't know what that is) Goes to 3H Pin 4 - HI - Correct

24 hi - Correct

Rest pulsing.

 

[Lurch666]

Could be the incorrect symbol on the schematics.

Mistakes do happen and you have to assume it's the correct sort of chip in the PCB or the game would have never worked properly.

What do you mean when you say the pin is dead?

I'm assuming it means it's floating.

That signal comes from pin 8 of the 04 at 12D on the video board (page 1/9 of the schematics) and should not be floating.

Check the output pin 8 of the 04 at 12D and if you see a signal it might mean the connectors between the PCBs are bad. You would have to trace the signal to the point it goes missing to verify that.

 

[John Bennett]

That's crashed - CPU isn't running. ROM/RAM/CPU/Address controller would be the area to look. GnG is a pretty good board to learn as it's pretty-much custom IC free and well documented, it's just worth a bit when in good nick (original stickers, no obvious repairs etc).

Quite often there's one failure on a board, not twenty, so do proceed with caution - one false move and you'll set yourself back weeks. It's mildly concerning that the screen is now white rather than the original random data (that you get when RAM hasn't been initialised at boot). Many things might not be active (like sprites) as the game isn't doing anything with them when the CPU is sitting in a crumpled heap.

Use a digital oscilloscope to look at signals if you can. Logic probes can give a quick overview, but don't give the full picture.

 

[biglouie]

As you say, go with whats in the board.

 

[biglouie]

Thanks John. Appreciate the advice.

So you think its actually gone backwards and the initial screen was closer to a fix than now? Well thats depressing.

I need a system of steps to get the CPU running. I am only running the top board for now. Top at least see if I can get away from white screen.

 

[John Bennett]

I don't know if it's worse or not, but one thing to remember is that it's a one-way system with a lot of games - CPU tells the graphics and sound sections what to do, but it's got no idea they're following what is put in the RAM/Registers on the board for each section.
So if you have made an error in the graphics section, it's unlikely to make the game less likely to run code.

The 'correct' way (easy to say with access to scopes and logic analysers) is to look at the address bus on the main CPU and see what it's doing. Often folk get lucky and the CPU isn't executing or it's a bad program ROM/RAM.

I very much doubt it'll work with one PCB - you'll need the lot. CPU might not care what the image on the screen looks like, but the program might try to read back from the video RAM (or use it as extra main RAM).

I've done a quick (top-of-my-head, no guarantee) annotation of page 1. The CPU RAM is on the page after that.

First question - is there life on A0,1,2? Don't expect them all to be doing stuff if it's stuck in a program loop, but is the CPU alive?

 

[favouredson]

......pic above shows after I replaced the LS04......

The IC in the pic is not an LS, it's an F. If you removed an LS you really should have replaced it with an LS.

Yes, they do the same job but they are different and this might cause issues.