• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

SE/30 Boot Process - stuck at ~15% on progress bar. How to debug?

craig1410

Well-known member
Hi,

I'm in the process of repairing a battery bombed SE/30 and have managed to repair enough of the PCB traces and vias to get it booting up partially. I'm using a SCSI2SD device with a bootable 7.5.5 system installed and it gives me the startup chime, happy Mac, "Welcome to Macintosh" and then the progress bar screen appears and the progress bar moves a couple of times in quick succession but sticks at around 15% complete where it stops. At this point the mouse is still responsive so it hasn't hung up completely but it never progresses further.

I have tried booting from 6.0.8 on my SCSI2SD as well and it doesn't display a progress bar but instead it stops at what looks like an empty desktop with empty menu bar. Again mouse is responsive but nothing else happens.

I tried booting by floppy drive after cleaning the drive and finding some known-good 7.5.3 install disks but it seems to stop booting after a minute or so of reading from the drive.

Lastly, I have tried removing half my RAM SIMMS so that only bank A is in use and tried swapping the actual chips from bank B into bank A but that made no difference either.

Oh and I tried booting into safe mode and it confirmed extensions would not be loaded but that didn't help either.

Any ideas on how I can figure out which stage of the boot process is failing? Does the SE/30 support a verbose boot mode or can I monitor the boot process in some other way?

Many thanks,
Craig.
 

craig1410

Well-known member
I just wanted to add some more findings to the above.

I can successfully boot 6.0.3 and once booted I can pull down menus, use Finder and Control Panel etc without issues. I can also access the floppy drive and hard drive without issues. During the first boot on 6.0.3 (disk image prepared via Basilisk and SCSI2SD) I was asked for a name for my computer which I entered.

I then tried booting 6.0.5 and it reacts the same as 6.0.8 by showing an empty desktop background with empty menu bar. So something has changed between 6.0.3 and 6.0.5 (I don't have access to 6.0.4) which either avoids or triggers whatever fault remains on my board. Joe from JCM suggested it might be the RTC/PRAM chip but I have probed this with my shiny new Rigol Oscilloscope and I am getting the expected 1Hz signal on pin 1 and can also see what appear to be valid signals on the clock and data connections. While this doesn't completely guarantee that the PRAM is working, I'd say it's a strong indication. If I could replace the chip I would though so if anyone knows of a source then please let me know.

Lastly, I buzzed out the continuity of all the signals from the FPU to the data bus, address bus and various glue chip lines and they all seem to check out.

Anyone got any ideas? Does the SE/30 generate any sort of system log files during boot that I could pull back off the SD card and inspect? I looked but couldn't find any but maybe I'm looking in the wrong place. Is any debug output via the serial port during boot? I'm happy to clutch at any straws available at this point because I'm so close to having a working SE/30 I can taste it! 😋

Thanks.
 

elemenoh

Well-known member
IIRC @brayne recently ran into a similar problem and troubleshot it down to a bad FPU. He has a video covering how to disable the FPU in order to confirm the issue, but I'm not sure which it is. Maybe he'll remember :)
 

craig1410

Well-known member
IIRC @brayne recently ran into a similar problem and troubleshot it down to a bad FPU. He has a video covering how to disable the FPU in order to confirm the issue, but I'm not sure which it is. Maybe he'll remember :)
Thanks, this is something that came up on Joe's Computer Museum channel on YouTube where he was fixing an SE/30 and is what led me to buzz out all the FPU connections. You can disable the FPU by disconnecting the chip select line. I was reluctant to try that as it involves desoldering the pin on the quad flat pack chip and bending the pin away from the pad and tying it to 5V with a resistor. I can certainly give that a try but having monitored the chip select line with an oscilloscope I can see that it is asserted by 6.0.3 as well. But maybe 6.0.3 is more tolerant to whatever issue might be present than 6.0.5+.

I might try this out this evening - will update the thread if I do. Thanks.
 

craig1410

Well-known member
By the way, I just realised a short time after my last post that @brayne is Bruce from Branchus Creations on YouTube and he has a bunch of videos on SE/30 and other repairs. I looked through a few of these to try to find one that mentioned the FPU but couldn't find any. These are multi-hour live streams though so I might have missed it. Lots of good information on those videos though so I will try to watch them properly when I have some more time.

I did note some comments about how a wonky RTC chip can cause spurious issues including issues with booting from SCSI so I might try swapping the RTC from a working SE board into the SE/30 as they are they same. If that fixes it then Bruce mentioned the RTC chip can be sourced from here ( https://www.utsource.net/itm/p/11744375.html ) so I can get some new ones if needed.
 

craig1410

Well-known member
Another quick update:

I tried swapping over the RTC chip from my working SE but that did not help. I also tried disabling the FPU by desoldering the chip select pin and pulling it up to 5V but that just caused a system error on boot up.

I also received a ROM-inator Atom from BMOW since my original ROM was a bit corroded, and that allowed me to successfully boot into Mac OS 7.1 via the ROM Disk facility. Unfortunately I still couldn't boot into 7.5.5 even after applying the ResEdit hack to the system file as described on the BMOW website. It still sticks at around 15% on the loading progress bar. One thing that I wondered is if 7.5.5 can run on just 8MB of RAM? I know the system file is around 4MB so I thought it should work okay but the BMOW mentions that it needs 16MB so maybe that's an issue. I was planning to get at least 64MB of RAM so maybe that's worth a try.

What I was able to do was set up a full installation of 7.1.3 using Basilisk but initially this wouldn't boot until I replaced the system suitcase file in the system folder with the one from the ROM-inator ROM disk. I don't understand why these are different because the ROM Disk is supposed to be 7.1.3 as well. Even the size of the system files is different with the one from the ROM disk being about 80% of the size of the one installed with the full 7.1.3 installation.

Generally the SE/30 has been working well and has been powered on for many hours and has been doing a lot of disk transfers between logical devices on the SCSI2SD card etc. So I'm really puzzled as to what might be wrong with it. I even did some tests using MacBench and the FPU seems to be working just fine as is the CPU, RAM, graphics, SCSI etc.

All very puzzling and somewhat frustrating. If anyone has any ideas then please let me know.
 

cheesestraws

Well-known member
This is extremely strange.

Off-kilter weird suggestion: make sure that the A/ROSE extension isn't installed in your 7.5.x installation? I've had that cause weird boot-time issues on some hardware.
 

craig1410

Well-known member
This is extremely strange.

Off-kilter weird suggestion: make sure that the A/ROSE extension isn't installed in your 7.5.x installation? I've had that cause weird boot-time issues on some hardware.
Thanks for that suggestion but I can't see anything of that name in my list of extensions. I have also tried booting with extensions disabled (hold down shift during boot) and that yields the same result. By the way, this exact 7.5.5 image boots fine in Basilisk where it was originally prepared.

What I might try is creating a very minimal 7.5.5 or even a 7.5.3 installation and see if I can get that working. I have a set of 7.5.3 floppy installation disks and I did get as far as disk 6 before it threw an error the other night and aborted the installation. That might have been due to the ROM-inator compatibility issue though. I could try that again with the standard ROM installed I guess.

Anyway, it's 1am here in Scotland so I'm going to call it a night. Any other suggestions very welcome. :)
 

cheesestraws

Well-known member
Anyway, it's 1am here in Scotland so I'm going to call it a night. Any other suggestions very welcome. :)

Yup. I'm down in southern England. I got caught up trying to fix things so I'm too running late to bed...

If you fancy uploading your image I can try it out on an SE/30 down here in the next day or two. I have been meaning to open one up and pop a scsi2sd in it anyway...
 

brayne

Well-known member
You could perhaps try running Snooper when you successfully boot into System 6. It goes through a series of hardware tests, so it might give some clues as to the cause. For example, I had a Mac Classic that would boot into System 6, but stopped while loading System 7. It turned out that System 7 does a little self-test of the serial ports during startup, while System 6 didn't. When I ran Snooper under System 6, it crashed when it tried to test the serial ports, which helped to reveal the problem.
 

craig1410

Well-known member
You could perhaps try running Snooper when you successfully boot into System 6. It goes through a series of hardware tests, so it might give some clues as to the cause. For example, I had a Mac Classic that would boot into System 6, but stopped while loading System 7. It turned out that System 7 does a little self-test of the serial ports during startup, while System 6 didn't. When I ran Snooper under System 6, it crashed when it tried to test the serial ports, which helped to reveal the problem.
Hi Bruce, that's an excellent suggestion because I think the serial ports are just about the only subsystem I've not been able to test either directly or indirectly. I did do a search for diagnostics software previously but never found Snooper. I'll certainly give that a try this evening.

@cheesetraws, thanks for the offer to try out my image, I'll see how I get on with Snooper later and if I'm still having trouble I'll try uploading it somewhere that you can get at it. I actually tried reverting back to the "minimal" image which is available on the savagetaylor.com site here ( 7.5.5 ) and which I believe is legal to distribute since it is a minimal image. This initially failed to even get to the progress bar screen due to the need for the ResEdit hack mentioned on the BMOW site here but once I applied that hack, it again got stuck at the same ~15% progress mark. So you could try that image on your SE/30 if you wish.

Another interesting snippet - I am now able to boot up the 6.0.8 minimal image from savagetaylor.com although it complains about one of the extensions by saying "Responder was not installed... -1". Apart from that it boots to a usable desktop which it didn't do before. I'm not sure if this is due to the ROM-inator now being installed or if something else has changed. I might try popping the original ROM back in to see what happens later today as well.

Thanks again both - will update later. (y)
 

craig1410

Well-known member
I couldn't wait until this evening... 😅

Well whad'ya know - it locks up at "Modem Test" after passing all the previous tests. Initially I got a bad keyboard and bad RAM error when running Snooper in 6.0.8 but when run in 7.1 from the ROM Disk, it passes these tests. I did notice on Macintosh Garden that it said it needed System 7+ so I think the RAM and keyboard are probably fine and just a symptom of running it in 6.0.8.

So, I guess I need to figure out what's up with my modem port now. I've not checked the schematics yet but will trace everything through and see what I find. Hopefully just a bad trace somewhere or maybe a bad chip. Will let you all know!

Thanks @brayne, you've given me a valuable lead! (y)(y)

ps. I wouldn't be surprised if the 8530 serial controller is missing one of more data bus connections because I already had to repair a few of those due to battery corrosion and maybe the serial controller was branched off a corroded connection that wasn't reinstated by my repairs.
 
Last edited:

cheesestraws

Well-known member
You could perhaps try running Snooper when you successfully boot into System 6. It goes through a series of hardware tests, so it might give some clues as to the cause.

Today I learned! I always though the tool was probably also a gimmick because the NuBus card it came with looked like it didn't do anything. Now I know better :).

saying "Responder was not installed... -1".

This also points in the direction of the serial ports: Responder is part of the AppleTalk stack and defaults to using the serial ports for networking, so if Responder is trying to get the network ready and can't that may be further evidence in this direction.
 

craig1410

Well-known member
Quick update:

So I buzzed out the 8530 serial controller and found a broken PCB trace for A2 which I was confident must be the problem. A2 was one of the traces I had to repair between the CPU and ROM in order to get the machine to boot. Unfortunately after repairing the trace and powering the board back up it made no difference at all! 😞

So I continued buzzing out the remaining CPU/control side of the 8530 and everything seemed fine. I then turned my attention to the output side of the 8530 and I noticed that UE12 was getting a bit warm. Not hot, just warm. So I investigated further downstream and found that the two RCNET chips RP2 and RP3 both had some faults and were effectively shorting out some of the serial signals. The resistors seemed fine as I was seeing 50 ohms across the chip but if I measured resistance to ground I was seeing some pins at >2k ohms and others as low as 24 ohms. I presume it must be the capacitors shorting out.

So as a temporary fix while I wait for new chips to arrive, I snipped the input wires to the RCNET chips to relieve the load on UE12 and also UB12, UC12 and UD12. Again I was hopeful that this might move things along but alas it did not!

What I have also noticed is that the 8530 serial controller (UG12) also gets a bit warm (guessing 35-40C) although not to the point of being uncomfortable to touch. However, UG6 and UG7 which are part of the video system do get what I would say is a bit too hot (guessing 50-55C+) and are uncomfortable to touch for longer than a few seconds. I don't know if this is normal or not.

So, not sure what to make of it but I'll keep on persevering. It might be worth replacing the 8530 if I can find one and also UB12, UC12, UD12 and UE12 as they may have been damaged by the load from the damaged RCNET chips.

Time for bed... :sleep:
 

craig1410

Well-known member
Another quick update:

Since my last update I’ve not had a lot of time to work on the SE/30 but in the last few days I have replaced some of the chips including the 8530 serial controller (UG12) and one of the 26LS30 driver chips (UD12) which didn’t seem to be behaving correctly electrically and may have been previously damaged by the shorted RC network chips. I also replaced the RC network chips RP2 and RP3 themselves and literally buzzed out every single electrical connection related to the serial controller and drivers, marking each confirmed connection on the schematic using a highlighter pen until all traces were all tested.

Unfortunately I’m still no further forward and still see the same issues…

I am pretty confident that the 8530 serial controller is doing something because the voltages on the modem port outputs is changing when I do the modem port test in Snooper, but for whatever reason it’s still failing the test. I even tried making a rudimentary loopback connection on the port as it says in the test that a loopback plug needs to be attached. I just used some pieces of tinned copper wire and linked pins 1 & 2, 3 & 5 and 6 & 8.

I guess I should test out the 75175 chips (UB12 and UC12) and I have replacements for them if I need them, but I’m not holding my breath that the problem will be found here either.

In other news, I managed to pick up another SE/30 on ebay which was supposed to be “in full working order” but turned out to also be battery bombed and is probably in worse shape than the one I already had! I decided to just keep it though because it only cost £55 and has an ethernet card, 8MB of RAM, 1.44MB floppy drive, 40MB hard drive, working PSU, analogue board and tube etc. Anyway, I’ll post a new thread about that machine if I get anywhere with it. I suspect it’ll end up being for spares only though…

If anyone has any suggestions for anything else I could try to get the above board working then please let me know. I’ll post an update with any significant progress.🤞
 
I have an SE/30 with much the same symptoms that I had to set aside because I spent way too much time on it with little progress. https://68kmla.org/bb/index.php?threads/se-30-only-boots-from-system-6.36379/ Have you made any more progress? Your statement about system 7 looking at the serial ports gives me another place to look for broken traces and shorted filter networks. I'm working on a Mac IIx with trace rot right now but I'll go back to the SE/30 next. The SE/30 is the second of about 50 boards that I've repaired that would boot under system 6 but not system 7. Apparently this is fairly rare and most responses I got were convinced that it was a software issue even though I posted that the same hardware worked on all my other good boards.
 

craig1410

Well-known member
Hi,
I wish I could offer a solution but sadly I've not been able to figure this one out. I've replaced the 8530 SCC chip along with both SN75175 chips and one of the 26LS30 chips. The other 26LS30 seemed to be producing appropriate signals when I checked inputs and outputs using a digital oscilloscope. (Edit: I also replaced my resistor networks as mentioned earlier in this thread)

I've gone on now to rebuild another battery bombed SE/30 using a new PCB by @Bolle and that one is working perfectly so far. I've still got a few finishing touches to apply to that project and then I'll be circling back to my original SE/30. Most likely I'll end up using it as a donor for another new PCB but I'm really curious to know what's actually wrong with it. I suspect it's not much!

Note that the "Responder was not installed... -1" message I was getting in 6.0.8 seems to be a red herring because I see that on my new Bolle board as well and yet Snooper does not lock up when doing modem or printer port tests and I am able to boot into 7.5.5 no problem. I need to finish attaching serial port connectors and I need to create a serial loopback plug before I can perform a successful test.

I hope you find a solution because this has frustrated the heck out of me for many months! Please let me know if you find a solution.
 

craig1410

Well-known member
Just to take another variable out of the above equations, I can confirm that Snooper will work as long as you bridge pins 1 to 2 and 3 to 5 in both modem and printer ports when testing. No need to bridge any other pins. Pins 1 and 2 are the two pins in the bottom layer and pins 3 and 5 are the right and left pins in the middle layer. Quick ASCII diagram below (which gets formatting messed up when I save it...):
_____
/ 8 7 6 \
| 5 4 3 |
\ 2 1 /

I just used the leads from an axial resistor bent into a U shape and inserted into the ports from behind. Note that you'll also need to ensure that no software such as AppleTalk is trying to use the ports. I'd suggest disabling extensions on boot by holding down the shift key. Also I think Snooper might only work on System 7+. I was using the ROM embedded on my BMOW Rominator II ROM SIMM.

I've not got my original SE/30 working yet but I'll probably have one more attempt at fixing it now I at least know it should work with other variables accounted for.
 
Top