VOGONS


Strange floppy boot problem

Topic actions

First post, by EvieSigma

User metadata
Rank Oldbie
Rank
Oldbie

I recently acquired a PC with a very unusual problem: when trying to boot off a floppy disk (for example, a MS-DOS floppy) the system constantly restarts itself after starting to read the disk, this is typically where it restarts:

PXL_20220130_183017212.jpg
Filename
PXL_20220130_183017212.jpg
File size
175.55 KiB
Views
829 views
File license
CC-BY-4.0

Sometimes it actually gets to the point where I can start the DOS install, but then it restarts before it can start writing to the hard disk. What exactly is going on here? I've never had another computer do this before, it's extremely unusual. Could it somehow be the hard drive that's at fault, even if the system crashes before the hard drive even gets written to?

Reply 2 of 33, by EvieSigma

User metadata
Rank Oldbie
Rank
Oldbie
Meatball wrote on 2022-01-30, 18:42:

What happens when you disconnect the hard drive? Does the bootup complete?

Nope! Tried it just now and even without the hard drive connected it still restarts itself upon reading the floppy disk.

Reply 3 of 33, by douglar

User metadata
Rank Oldbie
Rank
Oldbie

I think you might have a bad simm or something is unstable at your current clock speed.

Try removing 4 or 6 simms and see if the issue goes away.

Also see if you can jumper your motherboard to run at a slower speed and make sure your ISA bus is running at 8mhz in the bios.

If that doesn’t fix it, try swapping put cards to see if you have a failing add-in board.

Reply 4 of 33, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

SLC2 chip? What CPU is that, exactly? And did you set the BIOS properly for it, this looks to me like you've have it set to FLUSH and the input is not connected or even supported by the chipset. So each DMA will cause cache and RAM to desync, a bad thing to have in a computer system. Try BARB setting, or if there is no such option, disable CPU internal cache in BIOS - if either helps, you know you can't use FLUSH for cache coherency. Pity but not uncommon on many mobos.

Reply 5 of 33, by EvieSigma

User metadata
Rank Oldbie
Rank
Oldbie
douglar wrote on 2022-01-30, 20:24:
I think you might have a bad simm or something is unstable at your current clock speed. […]
Show full quote

I think you might have a bad simm or something is unstable at your current clock speed.

Try removing 4 or 6 simms and see if the issue goes away.

Also see if you can jumper your motherboard to run at a slower speed and make sure your ISA bus is running at 8mhz in the bios.

If that doesn’t fix it, try swapping put cards to see if you have a failing add-in board.

I only have 2 RAM SIMMs but I'll check the BIOS to see if some default setting (I don't have a CMOS battery right now) is set incorrectly. For some absurd reason this BIOS defaults to not even having floppy seek enabled...

Looking at my BIOS settings, I don't see anything about BARB or FLUSH and since this CPU is soldered to the board and has a heat sink glued on, I don't even know what speed it's running at to adjust the ISA bus speed.

PXL_20220130_233313562.jpg
Filename
PXL_20220130_233313562.jpg
File size
252.75 KiB
Views
747 views
File license
CC-BY-4.0
PXL_20220130_233323953.jpg
Filename
PXL_20220130_233323953.jpg
File size
315.59 KiB
Views
747 views
File license
CC-BY-4.0

Reply 6 of 33, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie
EvieSigma wrote on 2022-01-30, 23:17:

Looking at my BIOS settings, I don't see anything about BARB or FLUSH and since this CPU is soldered to the board and has a heat sink glued on, I don't even know what speed it's running at to adjust the ISA bus speed.

Oh, if that's a mobo with actual IBM chip on it then please do not try to remove that heatsink. You will rip the CPU off the mobo, and damage the pads, before that glue lets go. I bought a mobo with such damage for parts, pity, it looked nice and was a rare case of SX-class chip with cache, but since it's IBM it has different control lines to Cyrix CPUs.

Anyway, do try to test with "Internal cache" set to disabled. Although with soldered CPU the mobo should correctly handle the DMAs, but it can be damaged somehow. Also your DRAM and cache settings might be too optimistic, try 1WS on cache and DRAM R/W. That's before or after you test with cache disabled of course, no point in setting waitstates for disabled cache. If that helps then try to lower those WS to zero, one at a time, and re-test.

Reply 8 of 33, by EvieSigma

User metadata
Rank Oldbie
Rank
Oldbie
douglar wrote on 2022-01-31, 01:44:

Deunan has good suggestions.

AT clock selection would determine the divisor for the ISA bus.

What motherboard / cpu do you have?

That is a good question, I don't actually know! I cannot find a clear manufacturer name or model number on the motherboard anywhere, and with the CPU heatsink being quite securely stuck on, the only way I can figure out the CPU is getting booted into DOS and running a program like HWiNFO.

Reply 12 of 33, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie

So it is the ALARIS LEOPARD mobo. Chipset matches, CPU is IBM SLC2, also matches. That chip would be either 50 or 66MHz (clock-doubled) with 25 or 33MHz bus. I've never seen a 40MHz version, though I suppose it might just work with overclock, but I doubt it. In any case the CLK2/8 is perfect for a 66MHz, and if there is a /10 divider strap that would be ideal for 80MHz. And even if it was 80, with /8 divider it makes ISA clock 10MHz, most cards will tolerate that.

If ISA clock was somehow an issue with FDC I would expect it to throw read errors (bad data or CRC), not just glitch and reset every time. So I say it's cache related or perhaps DRAM being too slow for 0WS, I mentioned that above.

Reply 14 of 33, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie

Battery sounds like a good thing to do. Mkarcher found a thing in some BIOSes where if the battery status is low then they will always report the floppy drive to DOS as a 360k 5.25", no matter what gets set in the CMOS settings: Re: Strange Floppy Disk behaviour 486 . So maybe DOS gets part way through booting (the BIOS can get started using the floppy drive settings you set), then DOS asks the BIOS for floppy drive types attached, gets told the drive it's booting from is a 360k drive, and then can't read from the disk.

Reply 15 of 33, by EvieSigma

User metadata
Rank Oldbie
Rank
Oldbie
snufkin wrote on 2022-01-31, 11:01:

Battery sounds like a good thing to do. Mkarcher found a thing in some BIOSes where if the battery status is low then they will always report the floppy drive to DOS as a 360k 5.25", no matter what gets set in the CMOS settings: Re: Strange Floppy Disk behaviour 486 . So maybe DOS gets part way through booting (the BIOS can get started using the floppy drive settings you set), then DOS asks the BIOS for floppy drive types attached, gets told the drive it's booting from is a 360k drive, and then can't read from the disk.

I had never heard of that issue but it's a possibility. I'll have to get one of those adapters that lets you mount a CR2032 holder to the solder points for a barrel battery, I don't think this board has an external battery header.

Reply 16 of 33, by snufkin

User metadata
Rank Oldbie
Rank
Oldbie
EvieSigma wrote on 2022-01-31, 18:06:

I had never heard of that issue but it's a possibility. I'll have to get one of those adapters that lets you mount a CR2032 holder to the solder points for a barrel battery, I don't think this board has an external battery header.

Oh, it was definitely a weird one. Depending on your board revision, it looks like the Alaris Leopard has an external connector the other side of the AT Keyboard connector from the on board battery. Any board photos?

Reply 17 of 33, by EvieSigma

User metadata
Rank Oldbie
Rank
Oldbie
Deunan wrote on 2022-01-31, 00:26:
EvieSigma wrote on 2022-01-30, 23:17:

Looking at my BIOS settings, I don't see anything about BARB or FLUSH and since this CPU is soldered to the board and has a heat sink glued on, I don't even know what speed it's running at to adjust the ISA bus speed.

Oh, if that's a mobo with actual IBM chip on it then please do not try to remove that heatsink. You will rip the CPU off the mobo, and damage the pads, before that glue lets go. I bought a mobo with such damage for parts, pity, it looked nice and was a rare case of SX-class chip with cache, but since it's IBM it has different control lines to Cyrix CPUs.

Anyway, do try to test with "Internal cache" set to disabled. Although with soldered CPU the mobo should correctly handle the DMAs, but it can be damaged somehow. Also your DRAM and cache settings might be too optimistic, try 1WS on cache and DRAM R/W. That's before or after you test with cache disabled of course, no point in setting waitstates for disabled cache. If that helps then try to lower those WS to zero, one at a time, and re-test.

So I tried setting cache and DRAM to 1WS and that did nothing to solve the problem, and the same goes for disabling internal cache. So maybe it's a battery thing?

Reply 18 of 33, by Deunan

User metadata
Rank Oldbie
Rank
Oldbie
EvieSigma wrote on 2022-01-31, 23:56:

So I tried setting cache and DRAM to 1WS and that did nothing to solve the problem, and the same goes for disabling internal cache. So maybe it's a battery thing?

Welp, so much for easy fixes. You can also try disabling the hidden refresh, and then see if you can add even more WS to the '206 chip access, and the DMAs (especially the 8-bit one). These are the options on the right side of the advanced section. This is to see if there's maybe some sort of conflict between FDC DMA and RAM refresh (which would corrupt the RAM and cause hangs or reboots). If that doesn't help, and neither does the battery, then I'd assume some mobo damage. Inspect it, especially the area around battery since that's where corrosion would start.

Reply 19 of 33, by EvieSigma

User metadata
Rank Oldbie
Rank
Oldbie

Can you buy CR2032 holders that plug into the external battery header on a motherboard or do you have to build one? I know I've seen some people on YouTube using them.