Were these machines always so painful? Maybe I'm misremembering, or maybe age hasn't been kind to them.

[The Finale of Seem]

I've taken time during the holiday break to get back to souping up and fixing up a few oldies, and holy moly has it been frustrating. I've been tweaking and tinkering with old Macs for 20+ years off and on, but I don't remember it being this bad, and I'm guessing that aging electrical components are the likely culprit. These include:

• Power Computing PowerTower Pro 250
• Power Macintosh 9600/200
• Daystar Genesis MP 800+
• Beige G3 233mhz

All of these have had problems, and all of them have fresh PRAM batteries. The PowerTower Pro was probably the least finicky, but that's also likely because I didn't try to push it too far and only went with OS 9 on it. Still, it exhibited a problem that every one of these has: I cannot get anything later than a Rage 128 working reliably in any of them. I've tried Radeon 7000s (flashed and Mac Edition), a Radeon Mac Edition, a Radeon 9200 Mac Edition, and a few flashed Radeon 9250s. The Radeon Mac Edition would work, but only intermittently. I might get five boots with no trouble, and then on the next reboot/power cycle it would completely crap out with no amount of CUDA resets or PRAM zaps helping. And even with the stock video card (ixMicro cards and old ATI Rage II or Mach64), sometimes when rebooting or power on I get no video at all, but it might work on a reboot, or it might work if I zap the PRAM (in which case I'll get the short blip of video before it restarts, then it works). Sometimes various cards would have the monitor power indicator blip on after POST (going from orange to blue to indicate it's getting some kind of signal), but then it would go back to dormant after a few seconds.

Expansion cards have been maddening. I know the six slot machines can be picky when it comes to upper/lower slots, but sometimes things are screwed no matter how I swap things around. The Power Tower Pro took USB, Firewire, an IDE card, a GPU, and a gigabit PCI card like a champ. I didn't have a lot of problems with that one, but I didn't mess around with it too much before setting it aside.

With all the others, I've found that USB cards that are explicitly compatible with Mac OS 8.6+ (and these were NIB old stock I picked up that had it very clearly spelled out on the box/manual and work with the stock USB extensions) still don't work in OS 9. At best, they'll be recognized by Apple System Profiler as USB cards, and I get a blip of power on a thumbdrive LED when plugging it in, but then it goes dark and I get nothing else. These work in a Quicksilver in both OS 9 and X.

In the 9600 and the Daystar I found that a few old Opti chipset cards with 2 external ports would work in OS 9, but strangely they don't work in the Beige G3. I tried a flashed SIL3112 card in the 9600. It seemed to work. I booted from a SATA DVD drive hooked up to it and it formatted and installed to an SSD on the other port. But it would always lock up at the Mac OS 9 loading screen when trying to boot from the SSD. An IDE card with a SATA adapter on the SSD worked fine, though. Oddly, the same flashed SATA card seems to work fine in the Beige G3. I tried a native Seritek 1S2 in the 9600. It was fully recognized by Apple System Profiler, but could not see any drives attached to it no matter what I did. Again, this one works perfectly in the Quicksilver. The various IDE cards I have laying around have helped to bridge this gap, thankfully.

I popped a 1ghz G3 upgrade into the 9600 and tried getting 10.4 running on it via XPostFacto. No luck no matter what I did. I could get it to boot up from either a DVD or Firewire drive (with the helper drive option) and install. The installation would take hours and the video was quite choppy, as if there was no acceleration whatsoever. It still installed, though, but would hang when booting at the same point late in the process, usually where launchd started sleeping and it might complain about the old ROM on the Rage 128 (no change when the Radeon ME worked). One time it made it far enough to repeat a message every so often about waiting for printing services, but I never got a full boot. I ran XPF on a bunch of stuff ages ago and didn't remember it being this troublesome. It always had quirks, for sure, but this guy is just being stubborn.

The Daystar was inexplicably worse. I tossed in a drive that already had a 9.2.2 install on it with a bunch of other goodies. It worked, but it did not like the MP extensions when I installed them from the updater, so I figured I should start low and work up. It didn't want to boot from a DVD drive hooked up to the IDE card, even though the 9600 and PTP had no issues. The CD-ROM drive in it wasn't too keen on burned CDs, but I did get it to boot from a SCSI DVD drive. I burned an ISO of the stock Daystar install disk (7.5.3) and was able to boot and install from that, but it was really screwy about booting from the IDE card and I had to install to an old 2GB SCSI drive. That worked, and I was able to run a few of the sample programs to test the CPUs after installing 7.5.3. All four worked. An OS 9 disk didn't want to update from 7.5.3, so I booted from 8.1 and it was able to upgrade rather than do a clean install. And that's where I had to stop, because for some reason the external DVD drive stopped working with the Daystar. Other SCSI drives worked, and no, it wasn't termination as this was on an external bus with an active terminator. The DVD drive works fine with other machines, but the Daystar decided it wasn't going to boot from it or see anything in it again. I was able to hook up a drive with the Daystar extensions update on it, which is a folder with two floppy images in it and Disk Copy 6.1.2...which refused to work. Open it from the external drive I used to get it over there (no USB, no working floppy drive...ugh)? I get an error saying that "Disk Copy" couldn't be opened because "Disk Copy" couldn't be found. Copy it to the internal drive? Now it's an error -199. Lovely. I set it aside at this point to save my sanity. Daystar lost their license and apparently their ability to provide support at some point, so I'm not even sure if I can get this thing working with 8.1+ and the MP extensions. Has anyone else fiddled with this?

The Beige G3 (Rev A ROM, sadly) is flaky as hell on the PCI bus. The SIL3112 card worked flawlessly in it, but nothing else did. Even the Opti USB cards that worked fine anywhere else didn't work in OS 9 here, even though ASP saw them properly. An ALI FW/USB combo card caused a bus error hard crash at startup. Other USB cards or Firewire cards caused an error type 11 at startup, but not always. It started getting awfully buggy when I tried using a SCSI drive to move some additional data onto it. OS 9 would install to the SSD on the SATA card no sweat, and it would boot up too, but trying to copy anything remotely substantial to it from anywhere else? It would lock up during the file transfer. Something less than 1MB might work, but something larger? File transfer stops, though the OS isn't locked up. Might as well be, though, because you can't stop the transfer and force quitting the Finder won't restart it. I'll try running from an IDE card or the internal IDE bus and see what happens. Is the Rev A ROM issue where the OS has to be on the first 8GB of the boot drive limitation only for the internal IDE bus, or is that for anything attached? I know there are some ROM patchers for the PCI bridge issues, but I seem to remember that only being for the 6500 and related boards?

I'm thinking that all of these might need a full refurb. Clean and recap the boards/PSUs and then look for any weak solder joints, screwed up traces, etc. I imagine poor electrical performance could screw things up like this, especially on the PCI bus when they can already be picky. I definitely don't remember Beige G3s being this bad. It's about the only explanation I have as I don't remember this much frustration back in the day. I'm probably forgetting some additional quirks here too. The past week kind of blurs together.

 

[wthww]

I'm thinking that all of these might need a full refurb.

These machines are certainly the right age! The other thing I run into often with PPC machines of this vintage is that they need nvram patches to do some of this and it's easy to displace them.

 

[Byrd]

That’s a lot o troubleshooting in one post - maybe suggest focusing on one at a time, deep cleaning of all checking caps voltages and testing each unit stripped down to basics and adding components one at a time. Also keep to standard software not hacks/parches/flashed cards see if you can get joy with stock parts first.

Contact cleaner is your friend as well.

 

[The Finale of Seem]

That’s a lot o troubleshooting in one post - maybe suggest focusing on one at a time, deep cleaning of all checking caps voltages and testing each unit stripped down to basics and adding components one at a time. Also keep to standard software not hacks/parches/flashed cards see if you can get joy with stock parts first.

Contact cleaner is your friend as well.

Yeah, pretty much. I tried paring it down to native Mac hardware and keeping it minimal, ie, the 9600 + Firmtek (and stock GPU), or 9600 + Mac Edition video cards only, but even that didn't work reliably. The Daystar was worse, even when using stock hardware as much as possible, and I know for certain that Beige G3s weren't always this finicky as I ran off of a hacked up one for a while in ages past. USB, Firewire, and a flashed Radeon 9100 PCI and I certainly didn't have this level of problems at the time.

Looks like I'll be getting into low level component repair soon enough. Oh goody.

 

[s_pupp]

A recent recap of my PM8600 fixed a malfunctioning PCI slot (and the built-in graphics accelerator), so a recap might indeed help.

I do recall, however, a lot of frustration back around the year 2000 getting PCI cards and Acard SCSI-IDE adapters to work without problems in an 8600, a 9500, and a 9600 that had Sonnet G4 CPU upgrades. I had freezes, data corruption, failure to boot, etc.

 

[croissantking]

The Beige G3 (Rev A ROM, sadly) is flaky as hell on the PCI bus. The SIL3112 card worked flawlessly in it, but nothing else did. Even the Opti USB cards that worked fine anywhere else didn't work in OS 9 here, even though ASP saw them properly. An ALI FW/USB combo card caused a bus error hard crash at startup. Other USB cards or Firewire cards caused an error type 11 at startup, but not always. It started getting awfully buggy when I tried using a SCSI drive to move some additional data onto it. OS 9 would install to the SSD on the SATA card no sweat, and it would boot up too, but trying to copy anything remotely substantial to it from anywhere else? It would lock up during the file transfer. Something less than 1MB might work, but something larger? File transfer stops, though the OS isn't locked up. Might as well be, though, because you can't stop the transfer and force quitting the Finder won't restart it. I'll try running from an IDE card or the internal IDE bus and see what happens. Is the Rev A ROM issue where the OS has to be on the first 8GB of the boot drive limitation only for the internal IDE bus, or is that for anything attached? I know there are some ROM patchers for the PCI bridge issues, but I seem to remember that only being for the 6500 and related boards?

Two things come to mind here:

- a poorly seated Personality card is often the cause of startup type 11 errors so definitely check that.

- the SATA card and SSD combo might not be working properly. Compatibility issues between the SSD and the PCI card most likely - the SiI3112 is an old chipset and there are compatibility issues with some newer drives.

Start with a traditional IDE drive on the built in bus and get to a point where things are stable. Then you can add things back in and see what’s introducing the problem.

The 8GB limitation applies to only to Mac OS X and only the built in ATA bus. If you’re running OS 9.x you don’t need to worry about this. The difference between Rev A and later ROMs has to do with slave IDE support on the latter.

 

[Byrd]

Looks like I'll be getting into low level component repair soon enough. Oh goody.

I don't think so, it appears you're trying to max the systems out all at once which leads to incompatibilities much like they acted when new. Presuming the systems were well stored, kept running periodically I'm yet to consider a recap on later PPC hardware - not saying it's not necessary, but not the main issue at play.

 

[Boctor]

Is the SiL3112 card in your G3 a newer, cheaper one with no specific branding? I found out the hard way that those have out-of-spec 3.3V VRMs on them. The card would cause my PM7200 to never see other cards and occasionally hang on boot, and it'd cause my G4 to not even display video for the same reason, despite chiming. It was as if it was the only thing on the PCI bus, nothing else would work.

An old Crucial MX300 275 works brilliantly for 9.2.2 and OS X, I'm still using that now. With normal, 3.5" HDDs, I would sometimes get the prohibitory symbol (single-user mode showed "still waiting for root device") or they'd spin down and never spin back up.

Since recaps have been rightfully mentioned already, I'd second that idea. I've had mid-90's Macs gracefully dry out with no electrolyte on the board, no odors when reflowing and walking the caps off the pads (I clean them up anyway to be safe) yet there'd be anomalies like built-in Ethernet never being visible... Until suddenly, the new caps made it come back to life.

 

[The Finale of Seem]

Two things come to mind here:

- a poorly seated Personality card is often the cause of startup type 11 errors so definitely check that.

- the SATA card and SSD combo might not be working properly. Compatibility issues between the SSD and the PCI card most likely - the SiI3112 is an old chipset and there are compatibility issues with some newer drives.

Start with a traditional IDE drive on the built in bus and get to a point where things are stable. Then you can add things back in and see what’s introducing the problem.

The 8GB limitation applies to only to Mac OS X and only the built in ATA bus. If you’re running OS 9.x you don’t need to worry about this. The difference between Rev A and later ROMs has to do with slave IDE support on the latter.

Pretty much. I had another crack at it today and finally found a combination that works. I reseated the personality card in the process, and I think Firewire + Gigabit + ATA66 card has fixed a lot of the flakiness I've seen. Even hooking the SSD up to the onboard IDE was creating problems. But this has finally fixed a lot of issues and it appears to be stable. I was seeing some really weird issues with certain applications refusing to start up for a variety of reasons, Stuffit errors, etc. All working now. Fingers crossed that it stays that way. I wanted to use this as a bridge machine to image some floppies for Macintosh Garden.

 

[The Finale of Seem]

I don't think so, it appears you're trying to max the systems out all at once which leads to incompatibilities much like they acted when new. Presuming the systems were well stored, kept running periodically I'm yet to consider a recap on later PPC hardware - not saying it's not necessary, but not the main issue at play.

I was still having problems even with very few extras. The Daystar was especially bad, even with only a few bits of period-appropriate hardware. Same with the 9600. There were things that didn't work even by themselves that should've, or that acted really odd and flaky, and a lot of this was newer than the machines, sure, but was still supposed to be explicitly compatible with them as well, like certain USB cards, native IDE/SATA cards, etc, and Mac Edition video cards. All of these would have worked when everything was new out of the box, but they don't work now. Something has aged poorly in the hardware.

 

[The Finale of Seem]

Is the SiL3112 card in your G3 a newer, cheaper one with no specific branding? I found out the hard way that those have out-of-spec 3.3V VRMs on them. The card would cause my PM7200 to never see other cards and occasionally hang on boot, and it'd cause my G4 to not even display video for the same reason, despite chiming. It was as if it was the only thing on the PCI bus, nothing else would work.

An old Crucial MX300 275 works brilliantly for 9.2.2 and OS X, I'm still using that now. With normal, 3.5" HDDs, I would sometimes get the prohibitory symbol (single-user mode showed "still waiting for root device") or they'd spin down and never spin back up.

Since recaps have been rightfully mentioned already, I'd second that idea. I've had mid-90's Macs gracefully dry out with no electrolyte on the board, no odors when reflowing and walking the caps off the pads (I clean them up anyway to be safe) yet there'd be anomalies like built-in Ethernet never being visible... Until suddenly, the new caps made it come back to life.

Yeah, it's a generic card. I figured it may not work or may have issues, but wanted to have a crack at it anyway. I thought the 3.3v issue was only with DA/QS machines? It doesn't allow my QS to boot as expected, but was fine in an MDD. The 9600 and the G3 both could see and use the card, but still had problems with it. Maybe the Mathey ROM will work with a few 3512 cards I have laying around. I saw dosdude1 posted an OS 9 only version, so I may give that a crack to see if it can be used.

 

[croissantking]

Even hooking the SSD up to the onboard IDE was creating problems.

This is expected, and generally doesn’t work properly.

You mentioned USB cards not working and I suspect it’s because you haven’t installed the USB Adapter Card Support package.

 

[The Finale of Seem]

This is expected, and generally doesn’t work properly.

You mentioned USB cards not working and I suspect it’s because you haven’t installed the USB Adapter Card Support package.

The odd thing is that it did work previously. I'd set this machine up with a few different bits and pieces a couple years ago and don't remember a lot of trouble with the SSD on the internal IDE bus. When I came back to it recently I found that the SSD was dead, so I had to start over.

I took a look and realized that the installer didn't install USB card support despite a card being present at installation. The Opti card is now working in the G3. Thanks for the tip. Strangely, the ALI FW+USB card still throws a bus error at the Mac OS splash screen, and the other two USB 2.0 cards with OS 9 support throw an error type 11 when the extensions load. I've already reseated the personality card, but the problem persists. The A/V card has the pads for a USB chip, so maybe I should look into that instead and save myself a slot...

 

[Ortho'sDeli]

2 things I would mention: my 9600 always needed a "running start" (power on, chime, power off, then on again quickly) to boot properly. Additionally, I've found that because the mobo (and therefore the CPU card) are mounted in a vertical orientation, the CPU slot and/or card has started sagging after 30 years and needs to be in just the right spot for it to boot.

 

[Unknown_K]

9600 slots seemed pretty deep and tight to me for the thick CPU to sag in any way.

I would be looking for weak PRAM batteries, marginal power supplies, or general capacitor issues that come with age.

Old computers, Cars, stereo systems etc. all seem to age badly if not fired up every so often.

 

[Juror22]

Old computers, Cars, stereo systems etc. all seem to age badly if not fired up every so often.

SO true!

 

[trs]

The easiest solution to your Genesis MP problem is, of course, to install BeOS R5.

Remember, that machine is basically just a quad-CPU 9500. I wouldn't install anything newer than 8.1 on it, maybe 8.6.

It's been decades since I had my hands on one of these, but I'd also see how things behave using a SCSI CD-ROM drive instead of DVD. Keep it simple.

 

[joshc]

All the systems you mentioned are weird because of CUDA, and the use of sockets for their CPUs. Start with those things and go from there. Also start with the least amount of things changed/plugged in, and add gradually. Don't stick all your upgrades/changes in there together and expect it all to magically work in one go.

 

[The Finale of Seem]

2 things I would mention: my 9600 always needed a "running start" (power on, chime, power off, then on again quickly) to boot properly. Additionally, I've found that because the mobo (and therefore the CPU card) are mounted in a vertical orientation, the CPU slot and/or card has started sagging after 30 years and needs to be in just the right spot for it to boot.

This was kinda my experience as well. First boot would often chime, but never give me video. It would usually work on a restart. And I was usually running these on a bench in a horizontal position, but I didn't see a difference in vertical orientation either.

 

[The Finale of Seem]

9600 slots seemed pretty deep and tight to me for the thick CPU to sag in any way.

I would be looking for weak PRAM batteries, marginal power supplies, or general capacitor issues that come with age.

Old computers, Cars, stereo systems etc. all seem to age badly if not fired up every so often.

Already replaced the PRAM batteries. Component level repair will have to be the next step. I'm not holding out hope that it'll fix everything, but if it can bring these things back to normal levels of possessed then I'll call that a win. Heck, I just fiddled with a 7300 that was worse than any of them. Known good G3 upgrades would lock up sometime during boot (worked fine in other machines) with or without the motherboard cache stick. Extremely finicky about PCI cards. I've put things back on the shelf until a recap is in the cards.