MV-1 turns on, now has the click of death

[HeavyMachineGoob]

This MV-1 was working absolutely perfectly at one point, I'd feel cheated to just junk this system or use it for parts. I was just turning it on and off to see if it would get past the green screen anyways, then all of a sudden it started clicking again. I wasn't power cycling fast or anything, I'd wait maybe 10 - 20 seconds and then switch it off and on. Could something have died on the PCB in that time? Perhaps I should get a logic probe now...

 

[Xian Xi]

I think this is one of those back burner projects. Come back to it once you gain more knowledge with repairs. I did that before and it's funny when you go back to old repairs and it turns out to be something really simple.

 

[HeavyMachineGoob]

Yesterday I poked around the PCB, looking for bent pins touching on the solder side, bridged LSPC, QFPs, SRAM, BIOS and CPU pins and didn't come up with much of anything. Although I have a question about the 68k...

Are any of the data lines supposed to be "touching"? By that I mean getting a resistance of zero in the continuity test. I found a pair of pins (4 in total) adjacent on the 68k that make contact between each other according to my multimeter and I'm positive they are datalines and no ground or anything.

 

[Xian Xi]

What are the pin numbers?

 

[HeavyMachineGoob]

Pins 12 and 13 have a resistance of 0.01. Pins 13 and 14 are 4.69. Pins 16 and 17 are 7.00. Pins 17 and 18 are 9.16. Pins 22 and 23 are 9.39.

 

[Nightmare Tony]

ok, so if you are getting the clicking, the watchdog is firing which is why it stays in the first screen setup. That means the data and address busses have issues. From the corrosion pictures, look at the CPU and battery RAMS as well. You may have multiple traces out due to corrosion.

 

[HeavyMachineGoob]

I've already tested BIOS > CPU > WRAM connections using one of MKL's pinout lists and it's all good. I already fixed one line between the BIOS and CPU, which also caused a disconnection between the BIOS and WRAM. The MV-1 was working for some time, but while I was testing the PCB, it started clicking again.

 

[mjs256]

Pins 12 and 13 have a resistance of 0.01. Pins 13 and 14 are 4.69. Pins 16 and 17 are 7.00. Pins 17 and 18 are 9.16. Pins 22 and 23 are 9.39.

Pins 16, 17, and 18 should be GND, /Halt, and /Reset. They should not be connected to each other, because that means the CPU would constantly be told to reset and/or enter a HALT state. That might be the problem. Are you sure those resistances are in Ohms (not kOhms)? I don't know enough about the 68000 CPU to comment on those other pins you mentioned.

 

[HeavyMachineGoob]

Physically speaking, the CPU is very sound. No apparent issues with the legs, joints and traces. If there was a problem with the CPU, it'd most likely caused elsewhere, but I don't know where because the rest of the PCB appears to be fine. The LSPC has no more bridged pins and the WRAM chips seem to be fine.

When I tested those resistance numbers, I had my meter set to 20k. I don't specifically know if the numbers were in Ohms or Kohms.

If 16, 17 and 18 are for sure supposed to be separated and given specific connections (like voltage for halt and reset), do you think it'd be easiest to cut the traces on the PCB and manually adding wiring in? Or should the proper approach of hunting down the source of the problem be taken? I can't for the life of me find a pinout for the 68HC000...

 

[mjs256]

If you had your meter set to 20k then those resistances would have been kOhms and 7.00 kOhms (which is 7000 Ohms) and 9.16 kOhms (which is 9160 Ohms) is a very reasonable value that does not indicate a problem. Do not cut the traces. There is definitely not a short there. (I just measured the same pins on a 2-slot and got 5.14 and 6.34 kOhms).
I am intrigued by how you found it working when you started up a game and then after a while graphics glitches started. This suggests that the board warming up is causing the problems. This could be thermal expansion causing some cold joints to lose contact or it could be a bad chip that starts malfunctioning when it warms up. It might sound crazy, but if you put the board in the fridge for 10 minutes to cool off and then try starting it up while it is cold that might help to confirm it is a thermal issue. At times I have even used an ice cube inside a plastic bag to cool individual chips to find which one is thermally sensitive. Just watch out for condensation (water build up due to the moisture coming from warmer air depositing on a colder surface - you see it on soda cans all the time).

 

[smkdan]

I can't for the life of me find a pinout for the 68HC000...

Any pinout for a DIP 68k is fine, like this one.

There is a set of 4 resistors in an inline package next to the 68k with something like "4.7K x4" written near to it. You were measuring close to 4.7K (4.69) or close to double (9.39) the value of of them which is normal given the paths. A bunch of 68k pins are tied to vcc through that resistor package. In the 68k pinout I linked they're IPL2,BGACK,BR,BERR,HALT,RESET. You already measured BR -> VCC (the one close to 4.7K) as being good. If *don't* get the reistance to VCC then the 68k won't work properly.

 

[HeavyMachineGoob]

If you had your meter set to 20k then those resistances would have been kOhms and 7.00 kOhms (which is 7000 Ohms) and 9.16 kOhms (which is 9160 Ohms) is a very reasonable value that does not indicate a problem. Do not cut the traces. There is definitely not a short there. (I just measured the same pins on a 2-slot and got 5.14 and 6.34 kOhms).

I wouldn't do something rash without a good reason for it (except for the LSPC *facepalm*, it's just the severe graphic glitching in-game was really bothering me. It's all good now though!).

I am intrigued by how you found it working when you started up a game and then after a while graphics glitches started. This suggests that the board warming up is causing the problems. This could be thermal expansion causing some cold joints to lose contact or it could be a bad chip that starts malfunctioning when it warms up.

I should probably mention something here so you have a clearer idea of what's going on. When the MV-1 was working and suddenly had graphic errors in-game (looked like missing lines and pixels), I took it all apart and cleaned all the riser card and cart slot connections. Without the cart guide, I tested again and the glitches went away, but after a while the graphic errors appeared again and I thought it was because there was no cart guide to keep everything steady. I went to re-assemble the system and had the very dumb idea of cleaning up the bad side of the LSPC with some flux, which leads to the VRAM error. I cleaned it all up the following day, MVS board was working but was stuck on green. After a while of testing, it started clicking again. And here we are...

It might sound crazy, but if you put the board in the fridge for 10 minutes to cool off and then try starting it up while it is cold that might help to confirm it is a thermal issue. At times I have even used an ice cube inside a plastic bag to cool individual chips to find which one is thermally sensitive. Just watch out for condensation (water build up due to the moisture coming from warmer air depositing on a colder surface - you see it on soda cans all the time).

If the MV-1 was working enough to play a game, I'd try doing that, but it's still clicking and heat doesn't seem to matter. I've powered it up after it sat in it's box for a night and still no dice.

Any pinout for a DIP 68k is fine, like this one.

Thanks! That helps a lot.

There is a set of 4 resistors in an inline package next to the 68k with something like "4.7K x4" written near to it. You were measuring close to 4.7K (4.69) or close to double (9.39) the value of of them which is normal given the paths. A bunch of 68k pins are tied to vcc through that resistor package. In the 68k pinout I linked they're IPL2,BGACK,BR,BERR,HALT,RESET. You already measured BR -> VCC (the one close to 4.7K) as being good. If *don't* get the reistance to VCC then the 68k won't work properly.

Thanks for explaining all that. I think I'm barking up the wrong tree in this case, but I suppose it was a good idea anyway. The fault could pretty much be anywhere at this point right?

Going off of what mjs256 said about heat sensitive IC/traces, could some part of the PCB have given out (broken/worn out) while I was testing it? I'm wondering if a different trace isn't making a connection anymore... One that I haven't tested already. I've gone around most of the battery affected area since the return of the clicking, perhaps I should verify everything else again...

 

[mjs256]

Going off of what mjs256 said about heat sensitive IC/traces, could some part of the PCB have given out (broken/worn out) while I was testing it? I'm wondering if a different trace isn't making a connection anymore... One that I haven't tested already. I've gone around most of the battery affected area since the return of the clicking, perhaps I should verify everything else again...

Yes, when you were testing it, the board warmed up and expanded and when it is tunred off it cools down and shrinks a bit. This would not be a big deal except for the fact that different materials (like the IC casing and motherboard) expand and contract at different rates. This is quantified with something called the coefficient of thermal expansion (or CTE for short). For example, you turn on the MVS motherboard and it heats up by say 10 degrees F. This makes the plastic IC casing (CTE = 0.00004 per degree F) want to expand by about 0.04%, however, the fiberglass board underneath (CTE = 0.00001 per degree F) only wants to expand by about 0.01%. This may not seem like much, but the fact that they want to expand/contract different amounts cuases thermal stress which can make a cold solder joint pop and lose contact.
Something that makes these sorts of things very hard to track down is that when you test for continuity with a multimeter you press the probe tip either near or right on the solder joint, and the force you apply to the probe can press the broken joint together and (temporarily) restore contact.
You can minimize this by pressing the probe tip very gently on the very base of the pin where it comes out of the chip casing. Better yet, you can get some tiny probe attachments like these:

http://www.radioshack.com/product/index.jsp?productId=11390149

They will grab the pin and make good contact without applying much external force on it.

A different approach would be to turn it on and carefully press on individual pins to make temporary contact and see if the board comes to life or at least does something different. Be sure to use something that does not conduct like plastic (nothing metal!) to avoid accidentally shorting any two pins when pressing on them (trust me - you WILL cause a short if you try to do it with something metal on a QFP IC). I like the first way better because you may have more than one cold joint and temporarily fixing only one may not make a detectable difference.

 

[HeavyMachineGoob]

I tried your "stress test pins with plastic thing" idea, going around the different QFPs and other SMT chips and usually I get nothing. However a couple times when I was going over either the bad side of the LSPC or the RAMs, the board stopped resetting and just hanged on that first garbled image. What does that mean?

Also, for this other odd thing that occured, I need to explain something. On my main CRT this MV-1 usually doesn't appear onscreen when it's clicking, but on a Plasma display it always appears. I think it's because the picture is usually distorted in some way and isn't completely there, sorta like how VHS tapes can roll on the screen. Anyway, sometimes when I'm press on the bad side of the LSPC, the picture will appear onscreen and won't be distorted or "pulled" horizontally like in most cases. However, when this board was working, it would always appear onscreen after that first garbled bit (like when showing the green screen).

Are bad connection(s) on the LSPC known for causing the click of death?

 

[smkdan]

The constant resetting includes the video h/w being reset every time and I think that's all that's stopping your CRT from syncing with your board. If the resetting stopped when you pressed on something then I think you either a) temporarily broke the reset line from watchdog IC to other components or b) the 68k is writing it successfully due to a certain trace being temporarily fixed. Since the LSPC is involved, maybe one of the few 68k traces needed from it were fixed for a moment there. When the resetting stopped did it also stop the audio clicking or did the video just stabilise with the clicking still playing?

Something to try: do whatever you did to stop the clicking and, while still doing that, power cycle the board and see if it goes anywhere even if it's only for a short time.

Are bad connection(s) on the LSPC known for causing the click of death?

There's a bunch of 68k pins to LSPC that are needed for the system to get anywhere (IPL0,IPL1,some address/data) but I only have pinouts on the A2 with nothing on the A0. I just know that they lead somewhere on the chip.

 

[HeavyMachineGoob]

The constant resetting includes the video h/w being reset every time and I think that's all that's stopping your CRT from syncing with your board. If the resetting stopped when you pressed on something then I think you either a) temporarily broke the reset line from watchdog IC to other components or b) the 68k is writing it successfully due to a certain trace being temporarily fixed. Since the LSPC is involved, maybe one of the few 68k traces needed from it were fixed for a moment there.

Sometimes pressing on the pins or in that area stops the click, sometimes I press there and nothing happens and sometimes the MV-1 stops clicking all on it's own...

By the way, where is the watchdog IC? Is it a discrete component or is it bundled with another, larger IC?

When the resetting stopped did it also stop the audio clicking or did the video just stabilise with the clicking still playing?

The clicking stopped. Although sometimes the picture stabilizes while the clicking goes on. When the clicking stops though, the picture doesn't jump anymore. It becomes a still and static image on the screen, as if it were frozen.

Something to try: do whatever you did to stop the clicking and, while still doing that, power cycle the board and see if it goes anywhere even if it's only for a short time.

There's a bunch of 68k pins to LSPC that are needed for the system to get anywhere (IPL0,IPL1,some address/data) but I only have pinouts on the A2 with nothing on the A0. I just know that they lead somewhere on the chip.

Is there a chance those 68k pins go to the same pins, but in different physical places? Also, could you list all the 68k pins that are connected to the LSPC?

Also, I tried going over MKL's WRAM to CPU pinout and everything is connected, but on the C2, half of the data pins (between 11 and 19 I think) don't go to the pins listed in the pinout (5 - 1, 64 - etc). Instead they go to the pins starting between 61 and 54 I think. Is that of any significance or just how the MV-1 does things?

 

[smkdan]

Watchdog is in one of the custom chips. I'd guess the B0 first but only because the Bx chip had the watchdog in the next revision. The way to tell for sure is to see what chip has the 68k HALT line connected to it.

Is there a chance those 68k pins go to the same pins, but in different physical places?

The pin numbers will be different but the function must be the same. The packages are different so it can't be a 1:1 match either way. 68k pins A1~A3,D0~D15,IPL0/IPL1 should all connect to it. There's some other important ones that don't go straight to the 68k but from another custom instead depending on what chip does the decoding.

half of the data pins (between 11 and 19 I think) don't go to the pins listed in the pinout (5 - 1, 64 - etc). Instead they go to the pins starting between 61 and 54 I think. Is that of any significance or just how the MV-1 does things?

That's normal and it's the same for all systems. I mentioned the pattern here. 68k has a 16bit data bus (D0~D15) and each 8bit RAM is connected to half of it. It's the same thing for most RAMs on your board. All are 8bit and grouped in pairs to various 16bit buses except for the Z80 one.

 

[HeavyMachineGoob]

I reflowed some of the LSPC pins on the "bad" side (the side facing the RAM chips). I'm fairly positive they are all good and there's no problems left... The clicking is still present. I've tried pressing in a few places and I haven't been able to get the clicking to stop like before. I'll try leaving it on for a few hours and see what that does.

 

[HeavyMachineGoob]

Can someone else here with an LSPC-A0 tell me where 68k signals IPL0 and IPL1 go to? I've found where A1 - A3 and D0 - D15 are located on the A0, but not IPL0 and 1. They both lead to through-holes underneath the LSPC-A0, so it's pretty much impossible to visually see where they go. I've gone all around the chip and no dice.

EDIT: I also stumbled across this while searching through old threads:

Thanks everyone!

I'm hoping that one day one of the mods will need a system fixed and I can bargain with them to get a custom title of Mr Neo Fix-it.



Yes.

If the box is "stuck in watchdog", in other words, constantly resetting itself then you have to look at what it takes to boot: The CPU needs to read the BIOS ROM and initialize the Work RAM.

If it cannot do that then the watchdog timer expires and clobbers the CPU over the head via the reset line to try and restart it. If it cannot successfully restart the CPU then it hits it again... and again... and you have a stuck in watchdog problem.

Check for bad traces on the address, data, and control lines between the CPU, work RAM, and the BIOS ROM.

If they are all good then replace the surface mount work RAM.

RJ

I've tested all the traces between the CPU, BIOS and Work RAM and they all check out fine (including the one fix I made). The sudden click of death occurring during operation could certainly be explained by Work RAM passing away, traces don't go bad that quickly. What do you guys think? Is it worth buying some 32k SRAM and replacing one or both of the WRAM chips?

EDIT x2:

I just finished trying out a somewhat different BIOS connection pinout from an early AES revision (with an LSPC-A0).

* = Backup RAM

1----5V
2----GND
3----68000-12 (54), LSPC (23), NEO B0 (60)
4----68000-12 (55), LSPC (22), NEO B0 (59)
5----68000-12 (56), LSPC (18), NEO B0 (58)
6----68000-12 (57), LSPC (17), NEO B0 (57)
7----68000-12 (58), LSPC (16), NEO B0 (56)
8----68000-12 (59), LSPC (15), NEO B0 (55)
9----68000-12 (60), LSPC (14), NEO B0 (54)
10---68000-12 (61), LSPC (13), NEO B0 (53)
11---GND
12---68000-12 (62), LSPC (12), 43256* (19)
13---68000-12 (63), LSPC (11), 43256* (18)
14---68000-12 (64), LSPC (10), 43256* (17)
15---68000-12 (01), LSPC (9), 43256* (16)
16---68000-12 (02), LSPC (8), 43256* (15)
17---68000-12 (03), LSPC (7), 43256* (13)
18---68000-12 (04), LSPC (6), 43256* (12)
19---68000-12 (05), LSPC (5), 43256* (11)
20---LS08 (11)
21---68000-12 (29), NEO C0 (60), 43256 (10)
22---68000-12 (30), NEO C0 (61), 43256 (09)
23---68000-12 (31), NEO C0 (62), 43256 (08)
24---68000-12 (32), NEO C0 (63), 43256 (07)
25---68000-12 (33), NEO B0 (92), 43256 (06)
26---68000-12 (34), NEO B0 (93), 43256 (05)
27---68000-12 (35), NEO B0 (94), 43256 (04)
28---68000-12 (36), NEO B0 (95), 43256 (03)
29---68000-12 (37), NEO B0 (96), 43256 (25)
30---GND
31---68000-12 (38), NEO B0 (97), 43256 (24)
32---68000-12 (39), NEO B0 (98), 43256 (21)
33---68000-12 (40), NEO B0 (99), 43256 (23)
34---68000-12 (41), LS05 (01) (R9)
35---68000-12 (42), LS05 (03) (R9)
36---68000-12 (43), LS05 (05) (R9)
37---68000-12 (44), LS05 (11) (Q9)
38---NC
39---5V
40---5V

Everything on that list connects properly. A couple things like BIOS pin 20 goes to pin 59 of the NEO-G0 chip and pins 34 - 37 go to the 68k, instead of where the pinout says.

That pinout testing, plus lots of testing of the LSPC-A0 for bad connections, I conclude this board is probably 99% working, just like when it was actually booting and playing carts. I've exhausted pretty much everything I can find between this thread and old ones here on NG.com, as well as on Channelmaniac's repair logs. Bad Work RAM seems to be my last solid lead and it's a logically sound source of the sudden clicking. I'd just like an expert opinion on this before I try replacing the Work RAM.

On a similar note, I should probably buy a logic probe to go along with some 32k SRAM chips, so I can see which of the two Work RAM chips are dead.

 

[HeavyMachineGoob]

Bought a couple SRAM chips off arcadecomponents.com and a logic probe elsewhere. I'm gonna have to be conservative with that SRAM, may need it for this 4-slot MVS... Shouldn't go replacing the WRAM with no verification.

Also found this:

http://www.neo-geo.com/forums/showthread.php?102676-MV-1-quot-Work-RAM-Error-quot

Not sure what to do now... The logic probe will be a step in the right direction though, I can check chips for activity now.

 

[smkdan]

Can someone else here with an LSPC-A0 tell me where 68k signals IPL0 and IPL1 go to? I've found where A1 - A3 and D0 - D15 are located on the A0, but not IPL0 and 1. They both lead to through-holes underneath the LSPC-A0, so it's pretty much impossible to visually see where they go. I've gone all around the chip and no dice.

I can't help here but do you have the pin numbers to share? It'll be a good addition to the wiki. The old AES pinout has some overlap but it's still incomplete.

There's lots more to potentially check (RAM/ROM enables mentioned earlier) but a probe would help out a lot in tracing them or ruling them out. If you feel the time is better spent replacing the RAM than tracing out the board, which I know is slow and tedious then go ahead but I wouldn't personally replace anything until you've probed it. If you don't know what/where to probe then just post here when you receive it and we'll help you out.

 

[HeavyMachineGoob]

Here you go:

68k > LSPC-A0

5----LSPC-A0 (5)
4----LSPC-A0 (6)
3----LSPC-A0 (7)
2----LSPC-A0 (8)
1----LSPC-A0 (9)
64---LSPC-A0 (10)
63---LSPC-A0 (11)
62---LSPC-A0 (12)
61---LSPC-A0 (13)
60---LSPC-A0 (14)
59---LSPC-A0 (15)
58---LSPC-A0 (16)
57---LSPC-A0 (17)
56---LSPC-A0 (18)
55---LSPC-A0 (22)
54---LSPC-A0 (23)

I'll wait for my logic probe to show up before doing more tests.

 

[HeavyMachineGoob]

I now have a logic probe. I haven't used it much but it seems simple enough to operate. Where would a good place to start checking for errors be?

 

[smkdan]

Major ones to start with are:
-#15 on 68k
-#20 on BIOS
-#22 and #27 on both WRAMs
-#22, #27 and #20 on both BRAMs (system might not get far enough for these to be visible but their states are worth knowing)
They should all be pulsing but #20 on BRAM may appear dead depending on how far it gets during startup.

 

[HeavyMachineGoob]

Pin 15 on the 68k and Pin 20 on the BIOS are pulsing, but none of the specified pins on the WRAM and BRAM chips are pulsing. I tried a few other pins on the RAM chips and they do pulse. It's just the WE/OE pins that don't have anything. Are they outputs or inputs? Am I looking at trace repairing or bad RAM?