Try to set L2 cache off in BIOS (if it possible)

Sounds like BIOS can't read the drive - but the loaded OS can

Even Win3.1 has it's own drivers for the floppy....

Can you boot directly to DOS? - With Win 3.1 you can prevent it from auto-starting, and IIRC in Win95 there's a "restart in DOS" option.

DOS goes through BIOS for low-level disk access.

Also, it could be that something in your CMOS setup has gotten "not quite right" - Try removing the battery and causing a factor reset is BIOS - then set things up the way it should be again.

Babasha wrote on 2026-08-01, 10:09:

Try to set L2 cache off in BIOS (if it possible)

Thanks for the input - it was (well, it only specified "cache" but surely it means level 2), and I did disable it, but unfortunately to no avail.

DaveDDS wrote on 2026-08-01, 10:43:

Sounds like BIOS can't read the drive - but the loaded OS can […]
Show full quote

Sounds like BIOS can't read the drive - but the loaded OS can

Even Win3.1 has it's own drivers for the floppy....

Can you boot directly to DOS? - With Win 3.1 you can prevent it from auto-starting, and IIRC in Win95 there's a "restart in DOS" option.

DOS goes through BIOS for low-level disk access.

Also, it could be that something in your CMOS setup has gotten "not quite right" - Try removing the battery and causing a factor reset is BIOS - then set things up the way it should be again.

You are absolutely right - I was assuming it was using the same drivers also in DOS mode, but as it turns out, it is also unable to read floppies when restarted into DOS mode from Windows.

I could try removing the battery, and even replacing it while I'm at it (although it now seems to keep the settings), but I'm quite certain that after I initially noticed this issue a few years ago, I did replace the battery then, as the old one was dead. The BIOS allows the settings to be reset into default values, but I don't believe it would do an actual "full" reset, as taking the battery out would. That being said, I did try it (the "softer" reset), again to no avail.

Does the hard drive use overlay software?
If so, the boot order in bios needs to be C: then A: so it loads first. Then there may be a hotkey to boot from floppy.
The reason I mention it is certain Phoenix BIOSes of this era attempt a /256/63 geometry instead of /255/63 and this causes the DOS boot sector to hang trying to enumerate HDDs and parse partition tables.
As someone else said P24D L1 cache issues can exhibit like this, but it a cleared cmos wouldn't make that suddenly appear unlike the above.
BTW I hadn't seen a pre-frogdesign PB tower case before. This must have been one of the last ones.

drMPlanck wrote on 2026-08-01, 11:01:

You are absolutely right - I was assuming it was using the same drivers also in DOS mode, but as it turns out, it is also unable to read floppies when restarted into DOS mode from Windows.

Your BIOS might default to 5 1/4" 1.2MB when the battery dies as well.

Does the BIOS have "disable floppy seek at startup" - if so turn it OFF

This will tell you if the drive is being correctly accessed by BIOS at startup, and on many BIOS will help auto-detect DD .vs. HD drives
(BIOS steps out 45 tracks, then back 40 tracks - if track-0 sensor comes on, drive is DD, if not drive is HD)

Since you can boot in DOS mode, you can run ImageDisk - IMD accesses the floppy controller directly (not through DOS or BIOS)
It will also let you do quite a few low level tests: Seek, format, read, write etc.
This will give you a pretty good tool to verify that the hardware by itself is working, and may provide some more information on any problems it does see.

I did try to take out the battery for a minute, no effect. I didn't remember this anymore, but now I noticed that the last time I replaced the battery I had switched into using an external one (I did also remove the jumper so the motherboard "knows" to use the external and not the internal). However, looking at the user manual, I noticed that even though the battery connector had two pins for GND, I had only connected one to the battery, assuming it made no difference. Didn't have any connector pins ready at hand, but maybe I need to try this as well. As I remember, this fault presented itself already before I replaced the battery, but it was a few years ago so I cannot be quite sure.

The motherboard in question is this: https://theretroweb.com/motherboards/s/packard-bell-pb430

jakethompson1 wrote on 2026-08-03, 19:25:

Does the hard drive use overlay software? If so, the boot order in bios needs to be C: then A: so it loads first. Then there may […]
Show full quote

Does the hard drive use overlay software?
If so, the boot order in bios needs to be C: then A: so it loads first. Then there may be a hotkey to boot from floppy.
The reason I mention it is certain Phoenix BIOSes of this era attempt a /256/63 geometry instead of /255/63 and this causes the DOS boot sector to hang trying to enumerate HDDs and parse partition tables.
As someone else said P24D L1 cache issues can exhibit like this, but it a cleared cmos wouldn't make that suddenly appear unlike the above.
BTW I hadn't seen a pre-frogdesign PB tower case before. This must have been one of the last ones.

I've tried without the HDD connected (and configured in BIOS). But yeah, as mentioned, I did find that the same issue occurs also in DOS-mode.

I believe so as well, this must be one of the last ones before the new styling from PB. This was purchased new in September 1994.

jakethompson1 wrote on 2026-08-03, 19:29:

drMPlanck wrote on 2026-08-01, 11:01:

You are absolutely right - I was assuming it was using the same drivers also in DOS mode, but as it turns out, it is also unable to read floppies when restarted into DOS mode from Windows.

Your BIOS might default to 5 1/4" 1.2MB when the battery dies as well.

At least the settings show correctly. I also believe that the default setting for A: is the 3 1/2" 1.44MB drive, which was pre-installed when new.

DaveDDS wrote on 2026-08-03, 21:23:

Does the BIOS have "disable floppy seek at startup" - if so turn it OFF […]
Show full quote

Does the BIOS have "disable floppy seek at startup" - if so turn it OFF

This will tell you if the drive is being correctly accessed by BIOS at startup, and on many BIOS will help auto-detect DD .vs. HD drives
(BIOS steps out 45 tracks, then back 40 tracks - if track-0 sensor comes on, drive is DD, if not drive is HD)

Since you can boot in DOS mode, you can run ImageDisk - IMD accesses the floppy controller directly (not through DOS or BIOS)
It will also let you do quite a few low level tests: Seek, format, read, write etc.
This will give you a pretty good tool to verify that the hardware by itself is working, and may provide some more information on any problems it does see.

Unfortunately, the BIOS has very limited settings.

Thanks, I will try ImageDisk. Actually I also ran across this while researching this issue and I thought if I should give it a try, so now I certainly will 😀

I have been playing a little bit with ImageDisk. I'm not sure if I understand enough on the subject to know exactly what I'm looking for. Out of curiosity, I ran the TESTFDC, the result was FNF SD 250/300/500, PNP DD 250/300/500 and FNF DD 128b sectors 250/300/500 (F = failed, N = not tested, P = passed). I'm not sure, but given my HW I suppose that should be alright? I inserted a 2DD diskette (previously used in an Amiga , don't know whether that makes a difference here) when the program requested a low-density disk, and a normal 1.44MB HD one, when high-density was requested.

In the "main program" I also ran the align/test -function, in which it at least seemed that in the program I was able to control the disk drive to seek between tracks, read etc. just fine. Then again, this is hardly surprising given that this program does not rely on the same drivers as used by BIOS/in DOS, and that I'm also able to use the disk drive just fine within Windows. But again I'm not really sure what to look for here.

Would you have anything specific to test with ImageDisk in mind, that could give me some clues?

BTW, quite an impressive collection of SW you have written, has to be said. It's fantastic to have folks like yourself around in these kinds of communities.

I created ImageDisk specifically to be able to archive/preserve ANY floppy that the PC hardware could handle - the PC used a Nec765 based FDC which it more capable than Microsoft/IBM lets on (but it does have some limitations that other FDCs don't).

For this reason, ImageDisk does NOT impose "arbitrary" restrictions baseed on what Msoft "thinks" you would want to do on a PC - so it's a very good idea to understand how different floppy drives and formats work at a more in-depth level than "typical" PC users need to. The ImageDisk documentation and interactive help can help a lot with this!

A PC supports 2 different types of 5.25" drives - a DD(360k) drive has 40 tracks and rotates at 300rpm. An HD(1.2m) drive has 80 tracks and rotates at 360rpm - Because of this difference in rotation rate, DD data has to be read/written at 250kps on a DD drive and at 300kbps on an HD drive - typical BIOS/OS "drivers" are configured for a specific drive type and automatically set the write speed - ImageDisk is far lower-level than that and while I records the data rate used in the .IMD file, it allows you to "translate" rates, so the .IMD will be as it it had been accessed on different drive type.

250kpbs on an HD drive makes no sense (has never been used) as does 300k on a DD drive - hence TESTFDC not test these "invalid" combinations. The 'N'ot-tested at 300kbps tells me you are configured for a DD drive.

Also keep in mine that MANY PCs use "all on one" system chips which include a (somewhat) Nec765 compatible FDC NOT a separate discrete Nec765 chip.

SingleDensity has never officially been used on PCs, hence many system chips don't implement/support it - SD also requires different data-separation capability and many PCs don't use higher-end data separator chips (also often included in an "all in one" system chip)

So it's not surprising that you got FNF on the SD tests. Sector sizes can also vary, with 128bytes/sector only being used on very early 8" disk systems (well before PC) - hence many system chips don't implement this - in fact the PC only officially used 512byts/sector - so YMMV.

Test/Align can be used to test seeking as well as reading/writing tracks at the various data rates and densities (basically manully doing what TESTFDC does, only you can focus on a specific operatiin while you are debugging it.

To answer your question, it look like you have figgured out and done the "low level" stuff - all I would suggest now is to F)ormats full disk (note this is a low-level format only and don't make a Msoft filesystem) - just to see if there are any errors.

Then I'd try R)eading and W)riting a full disk.

If these work, it means your hardware is working - the problems you are having are most likely caused by something in system BIOS (or the DOS you are using)

drMPlanck wrote on 2026-08-08, 15:09:

BTW, quite an impressive collection of SW you have written, has to be said. It's fantastic to have folks like yourself around in these kinds of communities.

Thanks, that's how I try to look at almost all the time. (Sometimes it seems like more of a "wasted life" doing all that stuff!)

And sigh - since I my injury - Vision/stability isn't super great now ... so I tend to be less involved in the "Very Old Games ...." part of the community, but fortunately "Very Old Guys" and "Old Systems" is still quite applicable, so hopefully I can still help some.