Apollo 1260 issue.

[Pollock]

Howdy

I'm having an intermittent issue with one of my A1200's with an Apollo 1260 in it. Sometimes it wont boot, sometimes i get a black screen with HD activity, and it looks like WB is loading with flashing activity light but a black screen, but usually when it doesn't boot, it hangs at the reboot (i use loadmodule to insert a patched SCSI driver, so i can use a larger drive. like most of us do) and on rebooting the first time, it stops with a black screen, if i use the keys to reset, then it usually always boots fine, without a reboot to load the patched scsi.device version.

Could this be a timeout issue with the onboard IDE as it boots twice first time round? once it boots, or once i reset it with the 3 keys, it restarts first time all the time, and it's rock solid, it never hangs or crashes in everyday use. I use an ATX PSU that i wired into the Amiga power input, so it's not that.

anyone else had issues like this?

thanks.

[Hewitson]

I've had similar symptoms with my Apollo 1260. I always assumed the power supply couldn't supply enough juice, but it looks like you've ruled that out.

[Pollock]

Originally Posted by Hewitson

I've had similar symptoms with my Apollo 1260. I always assumed the power supply couldn't supply enough juice, but it looks like you've ruled that out.

thanks for your reply.

i had a quick search around online, and apparently we are not the only ones to see this happening. Still love my 1260 though

[Hewitson]

If I recall correctly you can not reboot my machine with the 3 finger salute when this happens. Pretty sure the PSU has to be switched off and on.

[Pollock]

Originally Posted by Hewitson

If I recall correctly you can not reboot my machine with the 3 finger salute when this happens. Pretty sure the PSU has to be switched off and on.

i know what you mean about the 3 fingered salute. it all depends where it falls over, if it hangs between loadmodule loading a patched scsi.device driver and needing a reboot because of this, then for me, the 3 fingers salute works. I read somewhere that these cards have some inherent issues anyway. This is from Jens (individual computers) who owns the rights to the Apollo cards, taken from Chucky's blog.

Chucky's Blog

"if you know how I feel about Apollos, you should also know that I do not encourage people to keep using them. They are instable by design – it’s inherent in the logic equations, and there is no way to improve the designs. They work to a certain degree (note that I specifically avoid the term “reliable”) if you have the right memory modules, but you can never get them to work totally reliable – no matter what speed and no matter what CPU."




So, i guess everyones is a bit tempremental. Mine has a late mask CPU, so its a 50Mhz part, clocked to 66Mhz, i know i could have gotten a 66Mhz part, but i wanted the FPU, and not an LC or EC version etc. it runs really cool, although overclocked, and when it's in WB, it's rock solid.

[Photon]

I'm assuming you've reverted to 50 MHz and get the exact same problem.

HDDs draw a lot of power on power-up specifically. It might be worth testing to supply it from another output of the PSU.

You shouldn't be starting it like this, you may get disk-validation errors just from booting. Check that ENV: is not assigned to a HDD folder during startup, and that other things don't write to the HDD during startup.

[utri007]

When I had this problem it was just SCART cable. Caps?

I have never noticed any difference in reliability and I have both Blizzard and Apollo 68060 accelerators.

[Damion]

edit: Someone mentioned caps, naturally they will be defective if still original.

There might be too much voltage drop at the card, regardless of the PSU you have at the main power input. You can try adding power at the floppy connector, which instantly solved the same problem I had many years ago with an overclocked Apollo 1260 and picoPSU.

While this worked in my case and shouldn't hurt anything, you can read some of the comments in this thread for more info:

http://eab.abime.net/showthread.php?t=63504&page=2

PS: In addition to the boot problems, mine would also occasionally black screen when running demanding programs (demos etc)..

[Hewitson]

Jens' comments about the Apollo are garbage. I can only assume he is bitter because he bought the rights to the cards, but wasn't able to reproduce them.

Are they as good as a Blizzard? No. But they're still damn fine accelerators.

[Daedalus]

Originally Posted by Damion

While this worked in my case and shouldn't hurt anything, you can read some of the comments in this thread for more info:

There are a couple of small things that I think should be added. First, the traces are plenty wide enough for any current the A1200 can reasonably be expected to handle, since the A1200 motherboard has dedicated ground and Vcc layers inside the board. The floppy power connector is connected directly to these so there are no thin traces to burn out.

Second, while it's not ideal to bypass the filtering, it should also be remembered that that filtering was there to deal with a cheap external power supply on a long, low voltage cable, which is quite a different situation to a good PSU in the chassis. If you're using an external PSU and injecting power through the floppy connector, I'd be a little more wary but I still can't imagine it's a problem with any decent quality PSU.

Finally, if there's a heavy current customer on the A1200 like a power-hungry accelerator, it won't be drawing all the current through the floppy port. A significant portion will come through the original power connector too. Not that it matters, given how the floppy power connector is connected.

Originally Posted by Hewitson

Are they as good as a Blizzard? No. But they're still damn fine accelerators.

Even back when they were first released, magazines mentioned instabilities and requiring timing fixes during reviews. The consensus was that they were marginally faster than, but less reliable than, the equivalent Blizzard board.

[Pollock]

Originally Posted by Damion

edit: Someone mentioned caps, naturally they will be defective if still original.

There might be too much voltage drop at the card, regardless of the PSU you have at the main power input. You can try adding power at the floppy connector, which instantly solved the same problem I had many years ago with an overclocked Apollo 1260 and picoPSU.

While this worked in my case and shouldn't hurt anything, you can read some of the comments in this thread for more info:

http://eab.abime.net/showthread.php?t=63504&page=2

PS: In addition to the boot problems, mine would also occasionally black screen when running demanding programs (demos etc)..

Hi Damion

The board was recapped about 7 months ago, I'm pretty sure it's not the PSU, i have measured the juice at the PSU output, and the juice actually going to the CPU core, and they are bang on, it's just sometimes when it boots, 8/10 times it boots fine, and it once it booted it's rock solid, no matter what i throw at it demanding demos, Doom ... it 100% stable.

[Pollock]

Originally Posted by Hewitson

Jens' comments about the Apollo are garbage. I can only assume he is bitter because he bought the rights to the cards, but wasn't able to reproduce them.

Are they as good as a Blizzard? No. But they're still damn fine accelerators.

i would agree with that, while Blizzard cards are a bit more stable in their design, the memory access speeds on the Apollo are quicker (if i recall correctly) or maybe something to do with the buffering i have heard Blizzard cards use, again, more knowledgeable people hopefully will chime in. it's still a great card.

[Photon]

Did you have a chance to try my suggestions? I'm sure we can nail it if we test differences this way.

[paul1981]

Originally Posted by utri007

When I had this problem it was just SCART cable. Caps?

I have never noticed any difference in reliability and I have both Blizzard and Apollo 68060 accelerators.

Same here. My Apollo1260/80MHz has always been rock solid. It's never missed a beat! It should really be run from a 4.5A A500 PSU or beefier though. I ran mine for 10 years from an A600 PSU with a little fan inside to cool the PSU! This was a heavily loaded system with FASTATA, Melody Sound Card, 2.5" HDD... 80MHz 060... PSU did fail eventually and now I use a better A500 type one rated at 4.5A!

@Pollock
I'd wreckon there's some software issue. Filesystem on RDB perhaps? Corrupted or incompatible LoadModule command? Corrupted or wrong/dodgy version of scsi.device that you're trying to load? Or slow spin up of your HDD? Wasn't there that wire that has to be cut on the IDE ribbon to stop the drive powering down on resets? Or was that just 3.5" HDD's as they are 12V driven? Also worthwhile doing a thorough RAM check I'd say. Just throwing some ideas around really. But the speed thing you mention points to software to me, as in...when it boots it boots and is rock solid. You don't get rock solid unless it really is.

[Pollock]

Originally Posted by paul1981

Same here. My Apollo1260/80MHz has always been rock solid. It's never missed a beat! It should really be run from a 4.5A A500 PSU or beefier though. I ran mine for 10 years from an A600 PSU with a little fan inside to cool the PSU! This was a heavily loaded system with FASTATA, Melody Sound Card, 2.5" HDD... 80MHz 060... PSU did fail eventually and now I use a better A500 type one rated at 4.5A!

@Pollock
I'd wreckon there's some software issue. Filesystem on RDB perhaps? Corrupted or incompatible LoadModule command? Corrupted or wrong/dodgy version of scsi.device that you're trying to load? Or slow spin up of your HDD? Wasn't there that wire that has to be cut on the IDE ribbon to stop the drive powering down on resets? Or was that just 3.5" HDD's as they are 12V driven? Also worthwhile doing a thorough RAM check I'd say. Just throwing some ideas around really. But the speed thing you mention points to software to me, as in...when it boots it boots and is rock solid. You don't get rock solid unless it really is.

Thanks for your input mate.

If it was software, then i am tempted to think the behaviour would be able to be replicated, but it can't be. 85/100 times, it boots fine. Once it loads the patched scsi driver with loadmodule and reboots, that's where it normally hangs, if at all. a 3 fingered salute, and it's normally back up. it's a 3.0 ROM, and i seem to remember a delay added to drives on KS3.1, or maybe i am dreaming, at any rate, I'm not using a traditional drive, its a CF card. 32GB with a patched scsi driver, power is taken care of a Dell SFF ATX PSU taken from a micro desktop, with 150W. i didn't want to use anything larger due to minimum load.

Something that has happened, that makes me think it's memory, is sometimes during big file transfers in DOpus, or loading a bigger game, like Hexen, when it returns back to the desktop, it says i have 32MB, when i have 2*32MB SIMMS, a reboot gives me the 64MB, but that happens very rarely, but it has happened.

[paul1981]

Originally Posted by Pollock

Thanks for your input mate.

If it was software, then i am tempted to think the behaviour would be able to be replicated, but it can't be. 85/100 times, it boots fine. Once it loads the patched scsi driver with loadmodule and reboots, that's where it normally hangs, if at all. a 3 fingered salute, and it's normally back up. it's a 3.0 ROM, and i seem to remember a delay added to drives on KS3.1, or maybe i am dreaming, at any rate, I'm not using a traditional drive, its a CF card. 32GB with a patched scsi driver, power is taken care of a Dell SFF ATX PSU taken from a micro desktop, with 150W. i didn't want to use anything larger due to minimum load.

Something that has happened, that makes me think it's memory, is sometimes during big file transfers in DOpus, or loading a bigger game, like Hexen, when it returns back to the desktop, it says i have 32MB, when i have 2*32MB SIMMS, a reboot gives me the 64MB, but that happens very rarely, but it has happened.

What I also forgot to mention about my setup is that it hasn't even been recapped! Neither motherboard, PSU or Apollo card has ever been recapped. Nor my Prelude 1200 (sorry, always mix up melody with prelude) - there are smd caps on there too. So my setup is way overdue for service, but the bloody thing just keeps on working like a champ. Yes, touching wood right now...

First of all, the fact that the problem only happens 15% of the time does not rule out a software issue as mentioned. As daft as it sounds, I wouldn't be surprised if it's due to your 3.0 roms. Apparently, we're not supposed to use 3.0 anymore so I doubt much stuff gets tested with it. I was only forced into 3.1 because my FastATA Mk.III wouldn't boot without it. As for the RAM issue you mention, that does indeed sound promising. I've never known that before, but if the system is stripped of 32MB of RAM all of a sudden then I would definitely expect a crash. Also, doesn't LoadModule load it's modules into the higher memory? If so then this explains the hanging at start as the RAM disappears. It sounds as though a memory test program should be used now.

Physically, is everything looking good on the Apollo? Mine's a bit crusty from a battery leak, but it survived. SIMM sockets good? Solder joints good on the SIMM sockets? Any green IC legs or anything? Chips seated well?

[utri007]

My rev 1d4 motherboard & Apollo didn't work with KS3.1 before I got timing fixes. It didn't boot at all, so I soft kicked KS3,1 and it worked. Now when I have new caps and timing fixes everything works.

[CentralPerk]

Try using RemApollo to load the patched scsi.device. Download it from Aminet and apply this update, then put the patched scsi.device into DEVS:Modules and add the following lines to the very beginning of the startup-sequence:

DRAP
IF WARN
C:REMAPOLLO QUIET MODULES DEVS:Modules/scsi.device
EndIF

[Pollock]

I think i have cracked it gents, it looks like a nasty stick of RAM. My Apollo 1260 has an additional SIMM socket soldered onto the underside, with an additional 32MB or RAM installed, sometimes, Workbench would just "drop" 32Mb and it would vanish, yet Workbench wouldn't crash. Apps that used more memory like Quake and Hexen etc, just wouldn't run. So i powered down, removed the additional stick of RAM, powered up, and it's fine, games that wouldn't run, but used more RAM, like Quake as mentioned, now work fine, and the sometimes not booting issue is gone! it boots first time, every time now, albeit with only 32Mb of RAM and not 64Mb, still 32MB should be enough. I have heard getting SIMM's to run on the Apollo cards is a bit hit and miss, maybe because this is a 50Mhz part @66Mhz, that's the issue, i wanted the MMU and FPU of a full fat version, although it's a late mask CPU, and runs really cool considering.

Now to install RemApollo and get that working.

So, if you have an Apollo 1260 that sometime wont boot, just occasionally with no reason .... check your RAM.

[zipper]

Apollo cards are picky because of the timings used. Blizzards are much easier.