Mrhide
Legendary
Extreme CPS2 Battery Damage Repair - ROUND 3!
- Thread starter Arthrimus
- Start date
Thanks a lot for the work you've done. This documentation is so useful for repairs.
While I have your attention, I have another B board that is exhibiting behavior that I'm struggling to explain. It boots consistently to a green screen. I've probed the graphics addresses and data and I get normal looking activity there, but the code rom, Q-sound roms, and sound sample roms have all of their addresses held constantly high and there is no activity whatsoever. I've probed the clock signals at DL-1525, DL-1827, and DL-1727. They are all present and looking normal. I've also probed all of the enable pins and they are all high, which matches a working B board that I'm using for comparison. Do you have any idea what could potentially be holding up the code rom, Q-sound rom and audio rom addresses? Thanks in advance!
If you have nothing on the code rom it maybe means that nothing is transmitted through the DL-1727 (MIF) which is main purpose is to manage the main bus. And because sound is derived from the main bus, it's kinda the same thing. So I would check:
- All the main enable lines: ~RESETI (HIGH), ~RESETI-OUT (HIGH), ~A-BOARD-CONNECTED (LOW), ENABLE-ADDR-BUS (HIGH). It should be okay since you said you have clock in 1525.
- Check for activity at the address lines of 1525 (or at least make sure it's not pulled all HIGH)
- Check the activity / state of the BUS-CONTROL bus (as input to the 1727)
That's where my gut feeling is. The other idea I have would be to check for activity on the Z80-BUS signal (there is a back & forth between the DL-1525 and Z80, if they cant talk properly everyone is frozen).
Thanks for the info. I have a strange new development with this one. I tried running this B board with Razoola's suicide test rom and a blank key written to it and it consistently boots in that scenario. This is a -4 board, and I have read that they can struggle with key writing, so this made me think that it might be an issue with my key writing setup, so I programmed an arduino nano with a single game key writing program configured for my Street Fighter Alpha 2 test roms and soldered it directly to the CN2 connector with a 75ohm resistor on the CLK signal per Artimeo's notes on his github. Now it inconsistently boots the street fighter alpha 2 roms. When it boots successfully it has graphical issues, no surprises there, but when it doesn't boot successfully it just shows a completely garbage screen with occasional flickering pixels. '
At this point I'm just going to focus on tracking down the graphics issues and if I can't get reliable key writing working I'll just set this board up with a decrypted rom set.
Last edited:
Time for round 2. This B board was one of 3 that I acquired from another forum member a few days ago. It didn't have the kind of severe pin damage on the customs that the first one did, but it was definitely in need of some work.
My first course of action on this one was to clean the corrosion off of the DL-1927, PAL 10H, and all of the LS245s near CN1 and CN2. To do this I have started using a dremel with a wire brush attachment. It is really good at removing corrosion, but is still gentile enough that with light pressure I am not damaging any legs or pads. Once the corrosion was removed and the legs were inspected, I decided to go ahead and try powering the board up to see what would happen.
Unsurprisingly it had issues, but at least it boots reliably. I began with a close visual inspection of the most corroded areas, which were mostly around the pin31-60 range of the DL-1927.
Several pads were completely missing, but none of the legs were damaged so I went ahead and reflowed them, and repaired the obviously missing pads. After that I tested it again, and unfortunately I forgot to snap a picture of it, but the background layer was covered in $$$ now. It did appear to be generating more graphics correctly though, but the sprites for the characters were super blocky, like they were 1/4 or 1/8th their normal resolution. Poking around the DL-1927 with my oscilloscope, I found that pin 67 (A01-ADDR19) was floating, so I poked around it's trace until I got to a via. The via was dead, no conductivity between the two sides of the board. I repaired the via by running a wire through it and soldering it to the traces on either side of the board, but after that pin 67 was still floating. It turns out that the trace was also broken underneath the DL-1927, so I ran a bodge wire from the pin to a spot further up the trace.
After all of this the background was no longer filled with dollar signs, and was being rendered properly. The sprites for the characters were still super blocky. All of the address and data lines for the graphics roms were showing good activity, so I started looking at the graphics bus. That's when I found that 10 of the graphics bus pins on the DL-1927 were floating. It turns out that almost all of the vias below the DL-1927 were dead, so I went about repairing each one in the way I described earlier. There were also 2 dead vias on the LS245 at position 10L. I repaired all of those and here's what the board looks like now.
Test time!
Total victory!
I have to admit that this one was much easier than the first one. I think I spent a grand total of two hours working on it. Having the knowledge of where to look and understanding which chips are responsible for what helped a lot in pinpointing the problem this time. It also helps that the damage on this one was far less severe.
Now that this one is up and running I can start looking at the -4 board that has been giving me trouble further upthread. Hopefully I can nail that one down and provide a nice detailed writeup for it as well.
My first course of action on this one was to clean the corrosion off of the DL-1927, PAL 10H, and all of the LS245s near CN1 and CN2. To do this I have started using a dremel with a wire brush attachment. It is really good at removing corrosion, but is still gentile enough that with light pressure I am not damaging any legs or pads. Once the corrosion was removed and the legs were inspected, I decided to go ahead and try powering the board up to see what would happen.
Unsurprisingly it had issues, but at least it boots reliably. I began with a close visual inspection of the most corroded areas, which were mostly around the pin31-60 range of the DL-1927.
Several pads were completely missing, but none of the legs were damaged so I went ahead and reflowed them, and repaired the obviously missing pads. After that I tested it again, and unfortunately I forgot to snap a picture of it, but the background layer was covered in $$$ now. It did appear to be generating more graphics correctly though, but the sprites for the characters were super blocky, like they were 1/4 or 1/8th their normal resolution. Poking around the DL-1927 with my oscilloscope, I found that pin 67 (A01-ADDR19) was floating, so I poked around it's trace until I got to a via. The via was dead, no conductivity between the two sides of the board. I repaired the via by running a wire through it and soldering it to the traces on either side of the board, but after that pin 67 was still floating. It turns out that the trace was also broken underneath the DL-1927, so I ran a bodge wire from the pin to a spot further up the trace.
After all of this the background was no longer filled with dollar signs, and was being rendered properly. The sprites for the characters were still super blocky. All of the address and data lines for the graphics roms were showing good activity, so I started looking at the graphics bus. That's when I found that 10 of the graphics bus pins on the DL-1927 were floating. It turns out that almost all of the vias below the DL-1927 were dead, so I went about repairing each one in the way I described earlier. There were also 2 dead vias on the LS245 at position 10L. I repaired all of those and here's what the board looks like now.
Test time!
Total victory!
I have to admit that this one was much easier than the first one. I think I spent a grand total of two hours working on it. Having the knowledge of where to look and understanding which chips are responsible for what helped a lot in pinpointing the problem this time. It also helps that the damage on this one was far less severe.
Now that this one is up and running I can start looking at the -4 board that has been giving me trouble further upthread. Hopefully I can nail that one down and provide a nice detailed writeup for it as well.
Last edited:
Anselmo
Champion
Damn dude, your skills are top notch. I have a few boards I can donate to your cause if it means you can save some cps2 from the trash!
santa978
Professional
Wow you are a god 
You want to try fixing my CPS3 motherboard?
You want to try fixing my CPS3 motherboard?
And One
Student
Very shortly you may want to start offering your services for repair and earn some good side hustle money 
. You truly have some serious repair skills. Keep up the good work and keep posting your repairs, I find them great knowledgeable reads. Thumbs-up 
!
. You truly have some serious repair skills. Keep up the good work and keep posting your repairs, I find them great knowledgeable reads. Thumbs-up !
A
Apocalypse
Truth is, those repairs take hours, and even on the minimum income it's often not worth it unfortunately.Very shortly you may want to start offering your services for repair and earn some good side hustle money
Also people with the skills have probably higher expectations than getting the minimum wage for their labor.
It can be fun and challenging when doing it for yourself, you don't have the pressure to deliver in a timely maner, etc.
Doing it for others can quickly gets boring.
+ there's people who take the piss and advantage of those doing repairs reasonably. I've heard of people ordering broken boards on eBay and having them sent directly to their "repairer" unannounced.
It's ok to use these services when needed, but treating them like a sweatshop will quickly burn repairers out of their passion and time.
It's ok to use these services when needed, but treating them like a sweatshop will quickly burn repairers out of their passion and time.
A
Apocalypse
You know, I'm not buying anything on auction sites anymore because... I kept being overbid by $1 by other people who send me a message 1h after they won the auction asking for a repair...
That REALLY burns me out, result is I only repair boards for ME and CLOSE friends now.
I've seen this happen is it is really, truly bull-shit.
I can understand if there's a hard to find grail game and you find an auction for a non-working one and discuss it with your repairer before you bid. But seeing people just blindly buy up this junk and then using the good nature of these repair guys as basically cheap labor. This is why broken boards are nearly full price and repairers are dropping out of the business.
Hammy
Enlightened
kikaso
Enlightened
Even seemingly simple jobs like capacitor replacement can be labor intensive. I’ve recently seen modders / repair folk showing photos of Turbo Duos with leaky caps and how much work is involved just with cleaning and testing. Three hours for cleaning a leaking Duo isn’t uncommon apparently.
I have tried to pick up the skills necessary to do light repair / maintenance work primarily to save on time and money but also not to burn out the folks pushing the envelope on what can be salvaged.
I have tried to pick up the skills necessary to do light repair / maintenance work primarily to save on time and money but also not to burn out the folks pushing the envelope on what can be salvaged.
Don't worry guys I have no intention of starting to offer repair services. I don't have the time for that, and I'd have to charge too much to be worthwhile for anyone. This is purely for my own benefit. Now, with that said...
Would you believe it's time for round 3?
This board is the one I was talking about in post #24. As a preamble, I did some work to this board before I posted about it earlier, so let me explain what has been done so far. The most obvious damage was to the LS245s at 10L and 10M. The input side of both were corroded beyond salvageability. The pads and legs were completely gone, so I pulled good LS245s from a donor board and replaced them. Since half of the pads were gone, I had to run new traces for all of the missing pads. I had also found a few broken traces on the graphics bus at DL-1927 and ran new traces to repair them. That's about all I had done at the time.
When last we left off this board was booting occasionally with garbled graphics, but most of the time it simply boots to a garbage screen with no audio. When I'm struggling to pinpoint the source of an intermittent problem my instinct is to apply pressure to various components and see if I can cause any changes to the behavior of the board. I'm basically looking for cold solder joints or other sources of poor conductivity. While pressing on components, I found that If I applied pressure to PAL BGSB2 while the board is booting it would boot successfully every time. On this board all 3 of the PALs were socketed, and the socket for PAL BGSB2 was disgusting. I decided to remove the sockets for PAL BGSB1 and BGSB2 and solder the PALs directly to the PCB. Most of the B boards I have worked with do not have these socketed anyway. After soldering these two PALs in place the board boots every single damn time. Excellent, the biggest mystery is solved. Now this is what I get.
So the first graphics issue I tried to solve was the missing and messed up background layer. Obviously some of the maskroms data were missing entirely, so I checked the CE lines on all of the roms and found that CE for roms 14 and 16 was floating. I tracked the problem back to dead vias for GFX-A23,A22 and A21 at the DL-1927. That solved the missing graphics that you can see above, but the background layer was still garbage. I found that pin 67 on the DL-1927 was floating which connects to pin A9 on the CN2 connector and is part of the graphics addresses coming from the A board to the DL-1927. This issue was caused be a dead via, which I repaired. This improved the background situation, but it still looked like there was some sort of layer priority issue going on. I probed around some more and found that pin 65 on the DL-1927 was also floating. This one is interesting because it's supposed to be connected directly to ground, but it's actually an address pin. I patched a wire over to a nearby ground pin and the background now looked perfect. The remaining issue was with sprites. This is what it looked like.
Poking around on the LS245s at position 10L and 10M I found that I missed a couple of broken traces going to CN2 when I replaced the LS245s due to severe corrosion and leg damage. I was distracted by the damage on the input side, and I completely forgot to check that the output pins of the LS245s were connected to CN2. Well it turns out that I had 3 broken traces between the LS245s and CN2. Now the sprite issues are gone! THIS BOARD IS FIXED!
View: https://www.youtube.com/watch?v=-NbUMA6nJck
Would you believe it's time for round 3?
This board is the one I was talking about in post #24. As a preamble, I did some work to this board before I posted about it earlier, so let me explain what has been done so far. The most obvious damage was to the LS245s at 10L and 10M. The input side of both were corroded beyond salvageability. The pads and legs were completely gone, so I pulled good LS245s from a donor board and replaced them. Since half of the pads were gone, I had to run new traces for all of the missing pads. I had also found a few broken traces on the graphics bus at DL-1927 and ran new traces to repair them. That's about all I had done at the time.
When last we left off this board was booting occasionally with garbled graphics, but most of the time it simply boots to a garbage screen with no audio. When I'm struggling to pinpoint the source of an intermittent problem my instinct is to apply pressure to various components and see if I can cause any changes to the behavior of the board. I'm basically looking for cold solder joints or other sources of poor conductivity. While pressing on components, I found that If I applied pressure to PAL BGSB2 while the board is booting it would boot successfully every time. On this board all 3 of the PALs were socketed, and the socket for PAL BGSB2 was disgusting. I decided to remove the sockets for PAL BGSB1 and BGSB2 and solder the PALs directly to the PCB. Most of the B boards I have worked with do not have these socketed anyway. After soldering these two PALs in place the board boots every single damn time. Excellent, the biggest mystery is solved. Now this is what I get.
So the first graphics issue I tried to solve was the missing and messed up background layer. Obviously some of the maskroms data were missing entirely, so I checked the CE lines on all of the roms and found that CE for roms 14 and 16 was floating. I tracked the problem back to dead vias for GFX-A23,A22 and A21 at the DL-1927. That solved the missing graphics that you can see above, but the background layer was still garbage. I found that pin 67 on the DL-1927 was floating which connects to pin A9 on the CN2 connector and is part of the graphics addresses coming from the A board to the DL-1927. This issue was caused be a dead via, which I repaired. This improved the background situation, but it still looked like there was some sort of layer priority issue going on. I probed around some more and found that pin 65 on the DL-1927 was also floating. This one is interesting because it's supposed to be connected directly to ground, but it's actually an address pin. I patched a wire over to a nearby ground pin and the background now looked perfect. The remaining issue was with sprites. This is what it looked like.
Poking around on the LS245s at position 10L and 10M I found that I missed a couple of broken traces going to CN2 when I replaced the LS245s due to severe corrosion and leg damage. I was distracted by the damage on the input side, and I completely forgot to check that the output pins of the LS245s were connected to CN2. Well it turns out that I had 3 broken traces between the LS245s and CN2. Now the sprite issues are gone! THIS BOARD IS FIXED!
Last edited:
A
Apocalypse
This topic gave me the motivation to finally start fixing a box of CPS2 boards today that in my mind were for my retirement 
B-board n°1: No boot => bad LS245 @10D
B-board n°2: No boot => bad LS245 @10A
B-board n°3: Background graphics perfect but sprites corrupted => bad DL-1927
B-board n°4: No boot => bad GBS2 PAL
B-board n°5: Background graphics perfect but sprites corrupted => bad DL-1927
B-board n°6: Sprites perfect but background graphics corrupted => bad DL-2027
A-board n°1: No sound => bad LS273 @7H
A-board n°2: Corrupted graphics => bad LS157 @3W
A-board n°3: Stuck inputs => bad DL-1123
A-board n°4: No sound => bad HC04 @14F
That's all for today.
On a side note, almost half of the problems were due to faulty custom chips.
B-board n°1: No boot => bad LS245 @10D
B-board n°2: No boot => bad LS245 @10A
B-board n°3: Background graphics perfect but sprites corrupted => bad DL-1927
B-board n°4: No boot => bad GBS2 PAL
B-board n°5: Background graphics perfect but sprites corrupted => bad DL-1927
B-board n°6: Sprites perfect but background graphics corrupted => bad DL-2027
A-board n°1: No sound => bad LS273 @7H
A-board n°2: Corrupted graphics => bad LS157 @3W
A-board n°3: Stuck inputs => bad DL-1123
A-board n°4: No sound => bad HC04 @14F
That's all for today.
On a side note, almost half of the problems were due to faulty custom chips.
A
Apocalypse
BTW I forgot to thank @Arthrimus for the motivation factor and @WydD for his reversing work and schematics.
Made those repairs a breeze, to the point I felt I didn't deserve the success.
Made those repairs a breeze, to the point I felt I didn't deserve the success.
kikaso
Enlightened
A
Apocalypse
Yes it's a bit sad to have a pile of faulty boards that are just a custom chip away from being perfectly working...
At least CPS2 isn't rare stuff.
At least CPS2 isn't rare stuff.
A
Apocalypse
Oh, I forgot to add you can actually diagnose the A-boards alone (without the B-board). Many of the TTL chips I found bad had incorrect outputs even in static, like LS04s with the input and output of the same gate showing the same signal level (when it should be inverted), same for the LS74s, with either floating outputs or Q and !Q signals at the same level, floating outputs on LS374s while OE was low, etc.
Last edited by a moderator: