SE/30 logic board replacement woes

[Tj7596]

Tl;Dr I need to know the bare minimum that needs to be populated on a SE/30 logic board for it to boot. I don't need it to be able to do anything except show me a question mark.

Hello everyone. I apologize for what surely will be a huge lack of etiquette on my part. It's been 10+ years since I have posted on here. I have a SE/30 that had bad caps. The LB was in working order prior to the recap, but after I was met with simasimac. The corrosion was not nearly as significant as I have seen on some, but I figured there must be damage under the surface. I did some troubleshooting with the old board but never got anything definitive. Eventually I came across Bolle's project and ordered some replacement logic boards. I fully populated the new board and am now met with the exact same thing. For the new board I used as many new components as possible. I even have tried a new set of GAL (I think GAL anyway) chips programmed from the files on Bolle's GitHub repository. Still no dice. I have replaced all of the UX8 chips as well. 166 and 253s and whatnot. Still simasimac. I have checked a ton of traces as well at this point. All of the ones from the glue, CPU and I believe UH7. The 20 pin PLCC. This one I have not replaced yet. I have tried multiple ROMS and RAM combinations to no avail. I have now started removing some chips and transferring parts back and forth. The new board now has the original oscillator crystals. Still nothing. No change whatsoever.

So, any help or suggestions anyone could give would be highly appreciated. I have diagnostic tools (Oscope, multimeter, but no logic probe). Here's hoping withe the community's help we can save one more 030 from the grave.

 

[mmjs]

Hi, I've gone through a pretty difficult rebuild recently (some details here: https://68kmla.org/bb/index.php?threads/se-30-32mhz-clock-and-uh7.46062/ and on Mastodon if you want to read a *very* long thread: https://mastodon.vladovince.com/@mejs/111008723163117281)

Here's what I found was necessary to boot to question mark (not counting the passive components):

- Y1 oscillator and programmable logic (UH7, UI6, UG7, UG6, UE7, UE6)
- 68030
- Glue
- video RAM and ROM
- LS383, LS166, F253
- F258 muxes
- RP4-9, RP1, UD1, UJ6
- ROM
- one bank of RAM
- RTC (UK4) was necessary on one of my boards, the other one could boot without it. One would result in sad Mac without it
- both VIA chips
- SWIM and SCSI chips
- sound chip UE10 and Y3 oscillator

I'm pretty sure this is it. You could get to a raster (but not the question mark) without a few of these like SCSI.

I found sockets useful before I knew 100% which of my components were good, but they did cause trouble later on (see this thread: https://68kmla.org/bb/index.php?threads/se-30-fpu-trouble.46718/).

Additionally, I found some replacement programmable logic to be very finicky. Some combinations just wouldn't boot even though they were correctly programmed. I suspect the reason was speed, but I didn't do enough testing beyond identifying this as a possible issue.

Good luck!

 

[Tj7596]

Hi, I've gone through a pretty difficult rebuild recently (some details here: https://68kmla.org/bb/index.php?threads/se-30-32mhz-clock-and-uh7.46062/ and on Mastodon if you want to read a *very* long thread: https://mastodon.vladovince.com/@mejs/111008723163117281)

Here's what I found was necessary to boot to question mark (not counting the passive components):

- Y1 oscillator and programmable logic (UH7, UI6, UG7, UG6, UE7, UE6)
- 68030
- Glue
- video RAM and ROM
- LS383, LS166, F253
- F258 muxes
- RP4-9, RP1, UD1, UJ6
- ROM
- one bank of RAM
- RTC (UK4) was necessary on one of my boards, the other one could boot without it. One would result in sad Mac without it
- both VIA chips
- SWIM and SCSI chips
- sound chip UE10 and Y3 oscillator

I'm pretty sure this is it. You could get to a raster (but not the question mark) without a few of these like SCSI.

I found sockets useful before I knew 100% which of my components were good, but they did cause trouble later on (see this thread: https://68kmla.org/bb/index.php?threads/se-30-fpu-trouble.46718/).

Additionally, I found some replacement programmable logic to be very finicky. Some combinations just wouldn't boot even though they were correctly programmed. I suspect the reason was speed, but I didn't do enough testing beyond identifying this as a possible issue.

Good luck!

That's awesome. I wasn't expecting the first reply to be so thorough haha. Thank you so much. I am gonna give this a shot. I have video now so I at least know all of that circuitry is working. I will update soon on the status. I have a power Mac G5 quad that I am trying to reverse engineer the LCS on as well right now so it's a mess haha. Too many projects lol.

 

[Tj7596]

That's awesome. I wasn't expecting the first reply to be so thorough haha. Thank you so much. I am gonna give this a shot. I have video now so I at least know all of that circuitry is working. I will update soon on the status. I have a power Mac G5 quad that I am trying to reverse engineer the LCS on as well right now so it's a mess haha. Too many projects lol.

Hi, I've gone through a pretty difficult rebuild recently (some details here: https://68kmla.org/bb/index.php?threads/se-30-32mhz-clock-and-uh7.46062/ and on Mastodon if you want to read a *very* long thread: https://mastodon.vladovince.com/@mejs/111008723163117281)

Here's what I found was necessary to boot to question mark (not counting the passive components):

- Y1 oscillator and programmable logic (UH7, UI6, UG7, UG6, UE7, UE6)
- 68030
- Glue
- video RAM and ROM
- LS383, LS166, F253
- F258 muxes
- RP4-9, RP1, UD1, UJ6
- ROM
- one bank of RAM
- RTC (UK4) was necessary on one of my boards, the other one could boot without it. One would result in sad Mac without it
- both VIA chips
- SWIM and SCSI chips
- sound chip UE10 and Y3 oscillator

I'm pretty sure this is it. You could get to a raster (but not the question mark) without a few of these like SCSI.

I found sockets useful before I knew 100% which of my components were good, but they did cause trouble later on (see this thread: https://68kmla.org/bb/index.php?threads/se-30-fpu-trouble.46718/).

Additionally, I found some replacement programmable logic to be very finicky. Some combinations just wouldn't boot even though they were correctly programmed. I suspect the reason was speed, but I didn't do enough testing beyond identifying this as a possible issue.

Good luck!

Follow up question to this. Is the FPU also necessary?

 

[mmjs]

Follow up question to this. Is the FPU also necessary?

Not necessary to get to question mark! I think the original ROM may fail to boot a system without an FPU, and custom ROMs like the Rominator II may boot but applications that require an FPU will fail with Error Type 90 (ie. no FPU). But you should still be able to get to floppy icon.

 

[Tj7596]

Not necessary to get to question mark! I think the original ROM may fail to boot a system without an FPU, and custom ROMs like the Rominator II may boot but applications that require an FPU will fail with Error Type 90 (ie. no FPU). But you should still be able to get to floppy icon.

Ok, so I am down to the lowest level I can be from your post. I still have a simasimac pattern but it is different. I am letting the board take a good long soak in alcohol to see if anything changes. I tried redoing UH7 since the pattern looked similar to yours but I got the same thing. I will do one more thing and then it's back to checking all the traces. Gonna swap that 166 chip for the old one first.

 

[mmjs]

Ok, so I am down to the lowest level I can be from your post. I still have a simasimac pattern but it is different. I am letting the board take a good long soak in alcohol to see if anything changes. I tried redoing UH7 since the pattern looked similar to yours but I got the same thing. I will do one more thing and then it's back to checking all the traces. Gonna swap that 166 chip for the old one first.

Feel free to share your pattern if you've got a pic, might provide some insight! Also, are you using the oscilloscope to check out your data and address lines? It might provide a lot of insight while you identify whether you have a ROM, RAM, memory mux or another bus related problem.

This @techknight video might also be useful to test whether your CPU walks the bus without RAM (check around minute 23):

Additionally, while the above mentioned components are needed to get to the floppy icon, as you can see in his video you could do without some chips to even get to a chime or a raster, in case one of those components is giving you trouble. Might be impractical if you're directly soldering, but could be useful.

 

[Tj7596]

Feel free to share your pattern if you've got a pic, might provide some insight! Also, are you using the oscilloscope to check out your data and address lines? It might provide a lot of insight while you identify whether you have a ROM, RAM, memory mux or another bus related problem.

This @techknight video might also be useful to test whether your CPU walks the bus without RAM (check around minute 23):

Additionally, while the above mentioned components are needed to get to the floppy icon, as you can see in his video you could do without some chips to even get to a chime or a raster, in case one of those components is giving you trouble. Might be impractical if you're directly soldering, but could be useful.

Haven't had a chance to do any diag yet but here is where I am with the board right now.

 

[Tj7596]

Okay so here's where I'm at now. I have checked all three oscillators on my oscilloscope. They are all at the frequencies they should be. I have removed both via chips, and the swim to try to see if the CPU would walk the bus. I am still not getting address bus activity. I have removed the glue, tested, didn't see any bus activity, then put the glue back on. I am at least back to having a simasimac pattern so I guess I didn't damage the glue. I attempted to swap ue8 with the old one to no avail. I am beginning to wonder if something is holding the reset line for the CPU. I'm going to start digging into the schematics and see if I can trace a few things down. I'll do a few more checks and after that I may have to start considering the possibility of having a component failure. I will keep you posted on the progress.

 

[mmjs]

Okay so here's where I'm at now. I have checked all three oscillators on my oscilloscope. They are all at the frequencies they should be. I have removed both via chips, and the swim to try to see if the CPU would walk the bus. I am still not getting address bus activity. I have removed the glue, tested, didn't see any bus activity, then put the glue back on. I am at least back to having a simasimac pattern so I guess I didn't damage the glue. I attempted to swap ue8 with the old one to no avail. I am beginning to wonder if something is holding the reset line for the CPU. I'm going to start digging into the schematics and see if I can trace a few things down. I'll do a few more checks and after that I may have to start considering the possibility of having a component failure. I will keep you posted on the progress.

Definitely sounds like you will have to start to check additional lines across the board to troubleshoot this. One note -- the GLUE is absolutely necessary for any kind of operation (as you've probably noticed with the screen). Without it, the 16Mhz clock will not even get to the video logic, so you won't even get video sync.

I also wanted to note for you that I made an error above with the list of necessary chips -- while you can get a chime and a raster, you will need the ADB chip as well to get to the floppy icon. Not super important right now since you're working about getting more immediate stuff done, but just FYI for later.