Topic: Wardner PCB - anyone got one and a meter?

[system11]

Hi - I need to find where pin 1 of rom 17 goes - it's the bottom left one if looking at the broad from the edge connector. Pin 1 of roms 18,19,20 are linked, not sure if 17 should be too - on mine it was hard wired to +5v which is probably wrong.

[Womble]

Nope that is totally normal, will explain why a bit later, running late for work at the mo.

Update - OK have checked on my wardner board and it is normal for pin 1 of rom 17 to be tied to the 5V rail, ROM 17 is half the capacity of the others therefore it needs one fewer address lines to control it, pin 1 is the one that gets the chop so its not needed. Its tied to 5V purely because it might as well go to either 5V or ground, leaving it not connected anywhere would mean they could never use the higher capacity ROM in that socket, and as prices of components do silly things over time its possible that the lower capacity eprom could at some point be more expensive than the higher capacity ones if they were discontinued. Tieing the line high means they could use a half empty ROM at location 17 as long as the correct half contained the data that had filled the smaller eprom completely. Eproms were very very expensive when these boards were new so it would not have been the case of saving a few cents per chip, it was more likely to be saving few tens of dollars per chip.

What does your board do? Anything like this???

[system11]

Well the first time I tested it, it said something like BAD SRAM, or SRAM ERROR. I didn't actually touch it at all, and it switched to doing nothing whatsoever. I did notice the amp cap was very hot and touching it brought up the famous Flying Shark interference line problem, so I replaced the cap which solved that issue, but it's still not working.

Thanks for the confirmation. I'll post here if I get any life out of it. No idea which SRAM it was upset about, I'll just have to probe around and see what's going on. A couple of address lines are definitely doing nothing, again I've no idea if that's normal.

I'm starting to really dislike these Taito boards, had to diagnose a stuck bus tonight, ended up having to snip legs on about 8 ICs, and remove anything socketed just to find the culprit. At one point I even cut a track on purpose to isolate half of the things on the bus and rule them out in one go. Take a look at custom L2 and see how many places pin 3 goes... I counted 17 I think.

[Womble]

BAD SRAM is far more common than bad logic but its far from guaranteed. Take a look at what the CS and OE pins are doing, that will tell you if the chip should be outputting anything on the data lines. Not sure if I have any customs on my board, am pretty sure its a bootleg, will check tonight tho.

[system11]

Oops - I should add, when I was griping about Taito boards it was a Flying Shark - very similar to Wardner.

[Womble]

Its also worth keeping an eye out for Fujistu TTL chips, they are much more likely to die than other brands for some reason. Some folk in the arcade realm claim the F stands for something other than Fujistu considering their failure rate. Had a bad Fujistu LS245 on a New Zealand Story Taito board a few weeks back, it was missing 2 outputs.

[Womble]

Take a look at custom L2 and see how many places pin 3 goes... I counted 17 I think.

Yeh my board is a bootleg so no customs, you need to pull up the datasheets for all the chips that line goes to and find which one is an output pin, the output might be the custom but you have to assume the best and look for others.

[system11]

Take a look at custom L2 and see how many places pin 3 goes... I counted 17 I think.

Yeh my board is a bootleg so no customs, you need to pull up the datasheets for all the chips that line goes to and find which one is an output pin, the output might be the custom but you have to assume the best and look for others.

Oh, that's where it gets most fun - they're all tri-states with directional control apart from maybe two

[Womble]

Oh, that's where it gets most fun - they're all tri-states with directional control apart from maybe two

Yeh I hate that, you start out hoping to narrow it down and the list of options just fans out with the circuit.

If I were you I would desodler the SRAM and test that off board if you are able, well over half the boards I have worked on have been fine except for faulty SRAM. If you find any that have silent data pins when others are busy I would suspect them. Its also handy to poke a scope at them, often you will see very very messed up output signals on bad SRAM.

[system11]

Well I got back to where I started at least - despite me being pretty careful with these old boards, one of the tracks on the CPU ram broke under a chip. So, now I'm back at 'CRAM ERROR' - yes, now I look at it more closely, it says CRAM, just about - the letters aren't completely drawn, as if a colour was missing. One of the RAM out as it looked faulty from a piggyback test, but when I hold a new one in place over the legs which are still sticking out, it goes back to the same problem. So... I'm going to assume CRAM really does mean colour ram, and use my basket case flying shark to do some diagnosis tomorrow, theres more than one bank of ram doing colour on those. That FS runs but the sprite layer is all over the place, jailbark flickering corrupt lines, sometimes covering the whole screen. And that's /after/ a lot of repairs..

Unfortunately I'm kind of stuck actually putting new chips in right now, because the desoldering iron on my station has broken.

[system11]

Hey Womble, could you do me a favour please? Try interrupting the address lines on the character RAM (I believe this is the pair near the sound CPU), and also on the colour RAM (which is on the far right side near an array of 74LS240s and resistors), and then see what it says while booting?

I'm guessing yours is actually getting one stage further than mine. I've found if I tamper with the CPU RAM lines, I get 'SRAM ERROR' in white, so that RAM is logically OK. If I tamper with any other bank I still just get 'CRAM ERROR' - not sure if C means Colour or Character.

I tried using a Flying Shark to simulate this, but unfortunately the same tests just result in it not booting instead of error messages.

[Womble]

I was concerned that as my board is a bootleg that it may not look much like an original, but actually the layout is rather similar, minus the daughterboards that probably replicate the customs anyway. Anyway, I have gone round the board interrupting the RAM on boot, even the 6264s under the daughterboard and I never get CRAM errors.

http://www.klov.com/images/10/1061188821.jpg

On the main CPU SRAMs (2xTMM2016) I get either FRAM or SRAM errors. (by the white sticker above the jamma connector on photo linked above)

On the colour SRAMs (2xTMM2016 near a bank of ls240s and the dipswitches) (Above the "L" on the bottom edge of board) I get LRAM errors.

On the character SRAMs (2xTMM2016s on the far side of the board - in column B one has a yellow dot on it in photo ) I get FRAM errors

On some RAM chips I cant get an error at all (a bank of 5 6264s and 1 2016 in column A) all even tho they mangle the gfx when interrupted once its booted, clearly these are not tested at boot. I also get a BACK RAM XX error if I poke a set of 6264s thats barely visible under the main daughter board(lower 4 in column J on photo). There is another pair or 2016s that do not cause errors either also hidden under a DB (upper 2 in column J)

Are you sure your errors are CRAM and not LRAM, its hard to tell on mine coz the text is multicoloured half the time.

The HW Error message I was getting was due to bad joints on the daughterboard riser pins, so on a real board it would suggest a bad custom chip I guess.

[system11]

Yeah I fired up Wardner in MAME, my letters don't entirely appear, but it can only be CRAM or GRAM. Quite frustrating really.

[system11]

If you can believe this, CRAM ERROR is actually the sound RAM. I've replaced it and no change but I was able to simulate the failure in the MAME debugger. The game does error in the split second I see strange activity on the sound RAM too so it's definitely that.

Unfortunately I can't see anything wrong with the pathways from there to the CPU, maybe something else is polluting a data line.

Wish I had a logic analyzer now.