Need help with increasingly disturbing hardware problem

[mr_a500]

Amiga 500 with Viper 520 (33Mhz 68020/68882, 8Mb Fast RAM, 3.0 ROMs, 40Gb 2.5" IDE HD), Mini-Megi 2Mb Chip, ECS Denise, WB 3.1.

Super-long description of problem follows:

My A500 is crashing more and more often and I need help to find where the possible problem could be. I'm sure it's a hardware problem because I just tested two older harddrives containing my old software configurations that didn't have this problem before and now do. So it shouldn't be software related or harddrive related (Maxtransfer, filesystem, etc). Here is the history and possible hints (that may or may not be related, but I'll list anyway just in case):

About a year and a half ago, my keyboard reset (Amiga-Amiga-Ctrl) stopped working. About a year ago, I started getting a random crash in IBrowse when dragging vertical scrollbars (main and edit scrollbars). Then I started getting lots of WHDload random "Line emulator 1111" failures (but usually OK after a reboot). I don't know if the IBrowse scrollbar crash continued because I started scrolling with keyboard instead of mouse.

About 9 months ago I changed the IBrowse default to not display pictures (for speed) and only displayed pictures on certain sites. A few months after that, I switched back to display pictures and started getting IBrowse Network crashes (4 at a time - one for each open connection), IBrowse crashes and Miami crashes.

It is getting worse and now I get a guaranteed crash every single time after I have browsed about 20 pages with pictures on. It seems to happen when all Fast RAM is used up and there is 500-800K of Chip left. Memory doesn't seem to be getting freed. After rebooting after the crash, Scout usually says the last failure was either "Line emulator 1111", "Line emulator 1010", general CPU fault, error in exec or illegal instruction. GURUs are usually 80000004, 80000005 (these two the most common), 80000003, 80000006 and 8000000B. I also get some rare strange numbers.

WHDload fails much more often, usually with "Illegal instruction $10" or "Line emulator 1111". Some games that ran fine before now never run. I couldn't even use the latest WHDload 16.6 because it always fails.

I'm getting strange failures in programs that never had problems before. I've had lots of crashes starting ABCdir with FKey (instant crash when pressing key) that I never had before this month. I get increasing problems where double-click in ABCdir causes ABCdir to freeze - most times I can close ABCdir after that, but some times it is totally non-responsive. I've used ABCdir every day for 4 years and never had this happening before. Also, SongPlayer (set to play AHI 14bit calibrated on 16 colour WB screen) now crashes after playing for about 19 seconds. If I play while switching to an 8 or 2 colour screen, it doesn't crash (major clue here?). This also only started recently.

I tried memory defragmenters, memory tests and any diagnostic programs I could find on Aminet (there weren't many). I tried configuring IBrowse to leave 1Mb Chip and 1Mb fast, but this has no effect (besides, I never had problems before). I made sure all chips were firmly pressed down and all cards removed and reinserted.

So does anyone know what this problem sounds like and how I can diagnose it? If it is a CPU or FPU problem, how can I determine it for sure? Are there any good hardware diagnostic programs?


Edit: I just saw in Scout the previous failure says "Alert task: $DEADBABE"! Could I have a virus?? I have the latest VirusZ and it shows nothing.

[mr_a500]

Not sure if it's related, but there was a virus that inserted DEADBABE into memory:

That scared me, but it doesn't look related. I checked my conclip and it looks OK. The size is different and I don't see any of the strings described when looking at the binary. I don't get any of the stupid popups described. I also did another virus check of Workbench and nothing came up. Thanks for the link though. If it is a virus, then it must be new not to show up with the latest xvs.library. Does anybody know how to check for a virus not yet known? If it's a link virus, I guess I can do filesize checks against backups or compare dates.

The only thing I can think of off the top of my head is the old open her up and reseat all custom chips. Worth a shot.

As I said, I already did that.

The nice thing about AF, it is just about any hardware configuration you need and it is the FASTEST Amiga compatible possible fore modern stuff.

I have AmigaForever, but I don't have a PC and don't ever plan on buying one (I bought AF years ago when I had a work notebook to use). WinUAE is OK, but still far from perfect. I'd still rather use a dying real Amiga than a flawed emulation (especially if it's running on Windows). When I buy a Mac in a year or so, I'll try UAE then.

Sometimes old computers just get tired.
It's clearly a serious problem. Can you change your chip mem expansion or try running without it for a while?

I already tried running without the A501 (and opened the A501 to check for battery leaks), but still got the crashes. Maybe I've been pushing the poor old thing too hard and it's finally wearing out. I just wish I could isolate where the problem is to see if it is something I can replace or fix.

[mr_a500]

I have uploaded a diagnostic disk with repoair utilities on my website at www.ianstedman.co.uk/PU288.DMS

Thanks. I ran all the tests, but I think they are too old to be useful (most 1987 or 1988). One test said I had 1Mb Chip when I actually have 2Mb Chip and it said I had no harddrive (have 40Gb HD). I ran the s17 test and all tests passed when I disabled disk test and _blit. The disk test fails (probably expects 1988 SCSI harddrive and OFS) and the _blit test works for a bit, then crashes (maybe not expecting 2Mb Agnus).

Are you using the original A500 PSU?

Yes. I don't think power is a problem or I'd see screen flickering when drives are in use or get crashes during heavy load like copying between HD and floppy (and I don't).

Have you tried removing and re-inserting the CPU, Kickstart and other socketed chips?

The CPU and Kickstart chips are on the Viper and are not socketed. As I said before, I already removed and re-inserted all cards and pushed down all socketed chips.

[mr_a500]

Look on the underside of the A500 PSU, what are the ratings?
Some A500 PSUs can only supply 5V @ 2.5 Amps, your system may be demanding too much. Not all A500 PSUs are the same, see here.

My power supply is 5V @ 4.5 Amps.

if I were you, I would build the system from ground up to locate the problem. this I mean first bare system and workbench floppy. then fresh workbench install on hd. then 030 and some apps... and so on.

The 020 and HD are on the same card (HD can't be run without having 020 enabled). If I remove the card, I won't be able to run any of the programs where the crash is predictable (no HD, no 3.0, no RAM).

The only thing I can do is remove the Mini-Megi and replace the original Agnus and test for the crash. (bit of a pain that requires removal of chip, replacement of Agnus and removal of Gary chip from chip extender and back in slot, then put it all back again later) I think I'll try something easier for now, like removing the FPU and see what happens.

Does anybody know what exactly line emulator 1111 and 1010 errors are? Can the many illegal instruction errors be caused by a "dying" CPU? Can an overheated CPU develop flaws which can cause these failures? Now that I remember, these problems only started getting bad after I did a 3-day VistaPro animation render. Could I have fried (or mildly baked) my CPU? (but other CPU intensive tasks like image decoding don't seem to cause problems)

[mr_a500]

Good idea. I've got an old ratty obscenely yellow A500 (that smells of cat piss) that I got recently for parts. I'll plug the Viper in that one (and hope it doesn't foul my beloved Viper) and if the problem still occurs, it must then be a problem on the card. If not, then it's a problem with something in the original A500.