Neo Geo AES Work Ram Error

[ebichu]

Hi there!
I got a NG AES here which is showing a solid red screen after powering on without a game - Work Ram Error.
After checking the traces I found some damaged ones, caused by some kind of lquid I bet.

Here you go (click to enlarge):



After checking the traces I figured out that the left trace (2 arrows showing on it) are not broken - the signal is still going through the line. The upper right arrow point is getting a signal too but the lower right one not. I used some lacquer wire (don't know if it's correctly translated ) to bridge the connection from the underside.

My problem here is, that I don't get a 'beeep' from my multimeter when I check the connection. Also I don't know exactly where the line have to go.

Is there any shematic that shows me where I have to measure and where the signal should be on? Like which signal should be going from A to B, X to Y - you get it I think.

Thanks for your help!

 

[channelmaniac]

There is no schematic to show that.

Use q-tips (cotton swabs) and alcohol to clean up the area as best as you can. You might have to use a wooden toothpick or something else to scrape the crud off.

Try to redo the solder on the lower right one as it may just be corrosion in the plated thru hole. You're going to have to use a bright light to try and follow the traces on the board.

RJ

 

[ebichu]

I already cleaned everything with alcohol (98% alc) and used a small piece of wire to bridge the lower right point from the underside of the board because the corrosion messed up the hole at the upperside. It's just curious that some direct connections won't beep while some others are beeping - signal is there.

 

[channelmaniac]

If you can't get a good contact on the solder with the meter probe because of the corrosion on that plate thru then you won't get a beep.

Sometimes the corrosion is so bad you have to scrape the green coating off the trace and put the meter probe on the trace to do the continuity test.

 

[ebichu]

Nah, there are also lines which are 100% good and they don't beep either.
Problem is that I can't see where the trace with the corrosion is going to.

Look at that picture:



Left red trace is bad as you can see, right red trace is good and untouched. But also the good right trace isn't beeping. So how should I manage to check them if not all of them are beeping?

Maybe someone knows where this left red marked trace has to go to?

 

[channelmaniac]

Turn the board over!

Look under that chip for a thru-hole where the trace may go to a different place on the board.

RJ

 

[ebichu]

Negative.. On the underside you won't see a damn thing because of the work ram chip on the board. There aren't any holes so I guess, that the trace is going the way that I marked red. I will have another look under the microscope tomorrow at work and if no one knows where the line has to go, I will desolder the chip and take a look.

 

[Kyuusaku]

Don't mess with traces, just test continuity to the CPU and from one chip to the other on the pins. The only pins you might have to find are four control lines for upper/lower read/write strobes (/OE and /WE) that are decoded by the C# chip.

 

[ebichu]

But how should I test the continuity if some pins wont beep? I can't be sure then if they got contact or not. Like the picrute abbove: one corrupt and one good trace. Both don't give me the 'beep' sound of my multimeter :/

 

[channelmaniac]

Yeah... I don't think you are following the traces correctly.

Going off the top of my head:

Check continuity between the SRAM chips. Pins 1-10, 14, 21, 23-26, and 28 should connect together.

Pin 27 is R/W, pin 22 is OE*, and pin 20 is CE*, these three pins may or may not be connected together between the chips.

You show that one trace going to pin 26, which I'd have a hard time believing.

The data lines 11-13 and 15-19 don't connect together between the 2 chips.

RJ

 

[ebichu]

It was just a guess, since I don't see where the trace is going to :P

I measured all lines you wrote.
They all got contact. From the three separate traces you mention only 20 is connected together. 27 and 22 aren't.

So..?

 

[channelmaniac]

See if any of those traces up there that go under the chip go to the separate lines...

 

[channelmaniac]

OK,

Let me make it even easier for you...

Work RAM connects directly to the CPU. Trace out all the address and data lines from the work RAM to the CPU. Make sure you have continuity for those.

If you don't, fix it.

If you do, then trace the OE*, CE*, and R/W lines back to their source. If you have continuity there then the RAM is bad and needs replacing.

Do you have a logic probe? If so, use the damn thing!

One of the Enable lines may be tied to ground, I don't remember. If it is, then you can ignore that one. The other one and the R/W line should have activity when the system is trying to read the chip. If they don't, then look for bad traces! If they do, then replace the bad chip(s)!

Sigh.

 

[Kyuusaku]

If you aren't getting a reading, you need to press your probe a little harder against the pins and/or scrape the pin a little with the probe. From the picture they don't look bad at all.

The /CEs are tied together because they're both always enabled. Even if the RAMs have continuity between each other, you still need to verify that all the address and data lines are OK to the CPU. If they are, only then do you want to test continuity for /OE and /WE back to the C# chip because it will more of a pain. If you have a later board with the PRO-C1 chip, you can check the schematic thread for the exact pins to save yourself time following the traces.

 

[ebichu]

Ahh, logic probe it's called. I own a Multimeter -> http://www1.istockphoto.com/file_th.../2/istockphoto_2274062_digital_multimeter.jpg
I think it's the same

I will follow them tomorrow at work and check, of all traces are connected to the cpu.

@Kyuusaku: Where can I find that thread? Thanks so far guys, I'll see tomorrow if I can fix it

 

[Kyuusaku]

You don't need a logic probe for this, though it would make finding the bad pin very easy. Your multimeter's continuity test is good enough.

Don't follow the traces! Use pinout information available on the web:

http://www.elecfans.com/article/UploadPic/2008-4/200841172657788.gif
http://mc68000.com/68000dip.gif

PS: RAM A[14:0] connects to CPU A[15:1]

The schematics are at:
http://www.neo-geo.com/forums/showthread.php?t=203495

 

[MKL]

C2

1 68000-12 (43)
2 68000-12 (41)
3 68000-12 (36)
4 68000-12 (35)
5 68000-12 (34)
6 68000-12 (33)
7 68000-12 (32)
8 68000-12 (31)
9 68000-12 (30)
10 68000-12 (29)
11 68000-12 (5)
12 68000-12 (4)
13 68000-12 (3)
14 GND
15 68000-12 (2)
16 68000-12 (1)
17 68000-12 (64)
18 68000-12 (63)
19 68000-12 (62)
20 GND
21 68000-12 (39)
22 NEO-C1 (92)
23 68000-12 (40)
24 68000-12 (38)
25 68000-12 (37)
26 68000-12 (42)
27 NEO-C1 (93)
28 VCC

D2

1 68000-12 (43)
2 68000-12 (41)
3 68000-12 (36)
4 68000-12 (35)
5 68000-12 (34)
6 68000-12 (33)
7 68000-12 (32)
8 68000-12 (31)
9 68000-12 (30)
10 68000-12 (29)
11 68000-12 (61)
12 68000-12 (60)
13 68000-12 (59)
14 GND
15 68000-12 (58)
16 68000-12 (57)
17 68000-12 (56)
18 68000-12 (55)
19 68000-12 (54)
20 GND
21 68000-12 (39)
22 NEO-C1 (6)
23 68000-12 (40)
24 68000-12 (38)
25 68000-12 (37)
26 68000-12 (42)
27 NEO-C1 (7)
28 VCC



The green via is on the trace that connects NEO-B1 (86) to LSPC2 (106).

 

[ebichu]

Mornin'
Woah thanks Kyuusaku and MKL!

@MKL: Green Line seems fine, got connectivity. But the blue one doesn't - I will fix this and hope that this is the solution.

Will check all other traces anyways and will report here when I'm back home from work.

 

[ebichu]

Okay, back home with fixed lines which got all connectivity now but with a new problem. Now the ng doesn't give me a solid red screen, it now flickers red with a permanent 'click click click click....'. See the video here (maybe turn the volume a bit up) -> http://www.myvideo.de/watch/7278701/Neo_Geo_Work_Ram_Error

Any ideas?

 

[Kyuusaku]

A new problem could mean you don't actually have a WRAM error... I think it's the data bus. Was the BIOS replaced?

Look at the schematics from the other thread and verify that all data lines have continuity.

 

[ebichu]

But before cleaning the traces off the corrosion the screen was solid red.
And because of the broken traces from the work ram, I thought it was the work ram error
Now that I fixed all the lines, it flickers like the video shows.
I don't get what you mean with the shematic from the other thread. All I see there is a MVS PDF - no AES..? And I'm not that good in reading them.. :P

The bios is in it's original state. Nothing changed there.

 

[Hewitson]

That "click click click" is the sound of the watchdog timer constantly resetting the machine because the CPU has crashed.

The BIOS could well be the problem, a good start would be to check continuity between the BIOS and the CPU. Check without a cart inserted, so it rules out any connectivity issues with the cartridge slot.

 

[channelmaniac]

Sigh.

Make sure you didn't short across the data or address pins.

 

[Hewitson]

Good point. Can you show us a picture of your repairs?

 

[Nightmare Tony]

As a rule, when you have corrosion on any of the feedthroughs, there are several, not just one. You may need to remove chips to get to their underside to bridge their wires. I would use a desolderer and pull out as much solder from the hole and run the wire directly through the feedthough, then scrape off the trace on both board sides then solder them.

The solid red screen meant no clocking tot he video, I think. At least with the watchdog firing, the cpu is now TRYING to run.

A small trick which may help, get the 68000 and rAM pinouts. One of the RAMs will have d0-d7 to the 68000 and the other will have d8-d15. So you can infer 16 of the lines. Also, the address lines are also directly connected to the 68000.