Hi there! It seems to be a timing issue.
Some component has trouble keeping up with the rest.

You can try adding waitstates or slow down ISA bus a bit.
Maybe that helps. At least for troubleshooting..

Also, how fast are the SIMM modules? How many ns?
For a fast AT, 70ns or less are recommended.
Roughly, I mean. I'm no expert here. FPM RAM?! EDO RAM?!

Tip: You can have a quick system even if the clock rate is a bit lower.
Instead of running things at high speed, try removing the bottlenecks or reduce overhead.

If your 386SX runs at 14 or 12 MHz, but you manage to have a little cache, or manage to run at 0 waistates,
the end result might be smoother than a fast, but stuttering system.

Good luck. 🙂🤞

PS: CheckIt! has a diagnostic function for motherboard and memory, among other things.
Please try running them, if you can. Maybe that narrows down the issue a bit.

PS: IDE controllers in the 486 days were usually merely IDE host adapters.
They had no intelligence. All they did was attaching IDE hard disks to the ISA bus.
Your CF card gets the ISA bus timing directly, thus. If you overclock/underclock ISA, it has a direct effect on the CF card.
Some cards perhaps have trouble with speeds they consider non-standard.

PS: Size of the motherboard or signal reflections due to the way the PCB traces are layed out (say hard 45° or 90° edges) are also a factor.
Because they have an effect to noise level.
Traces too long work like antennas and can get in resonance if the operation frequency matches by accident (not so good).
Anyway, this is a topic someone could dedicate a whole book about.
I just mention it here, to give an idea that there are some more, possibly unknown, factors. So don't question your sanity.

I would agree that timing seems to be an issue, given that it works at half speed but not at the full 16Mhz.

There are no RAM SIPPs installed. The on-board RAM is made up of 2Mb via Intel P21010-08 chips rated at 80ns.

The disk controller is no-name, as far as I can tell. It's a multi-IO card, serial/parallel, floppy/IDE controller. I'm tempted to add an XT-IDE or XT-CF, but I don't know if the controller would conflict and I would still like to use the floppy drives.

Once I have a keyboard (the motherboard is AT, the oldest I have on-hand are PS/2) I can delve further into it.

https://imgur.com/a/WhZOVwHHere's an image of the controller card - It's marked as a "Quantra" brand - which is also actually marked on the 14-inch CRT I collected with this PC. I'm sure it's just some reseller - but I haven't found much data yet.

There are quite a few jumpers on it, so I'm looking for a datasheet now.

elvovirto wrote on 2023-02-12, 02:58:

when running at 16Mhz, characters are missing from the display. So instead of saying American Megatrends Inc, it'll look more like "Am ric n M atr nds I c" or something similar.

I have a 286 mobo that does this, but not with all VGA cards. Some work better then others, and also much depends on which slot VGA card is in. The one next to '245 buffer chip (closest to keyboard connector on this mobo) works best.
It could be one of the driver chips being weak but I think it's due to timing on newer, faster ISA cards and lack of line terminating resistors that 386+ mobo do have on ISA slots. Curiously some manuals do say to install the graphics card in either the first or the last slot.

Deunan wrote on 2023-02-12, 17:03:

elvovirto wrote on 2023-02-12, 02:58:

when running at 16Mhz, characters are missing from the display. So instead of saying American Megatrends Inc, it'll look more like "Am ric n M atr nds I c" or something similar.

I have a 286 mobo that does this, but not with all VGA cards. Some work better then others, and also much depends on which slot VGA card is in. The one next to '245 buffer chip (closest to keyboard connector on this mobo) works best.
It could be one of the driver chips being weak but I think it's due to timing on newer, faster ISA cards and lack of line terminating resistors that 386+ mobo do have on ISA slots. Curiously some manuals do say to install the graphics card in either the first or the last slot.

Interesting - now I have to question if it's an issue with the CF adapter, the IO card or the video card.

Fortunately, swapping slots is easy enough, but after trying various combinations, the issue persists. If only I didn't toss god knows how much stuff over the last 20 years thinking "eh, I'll never need that again."

Not the first case of modern IDE drive causing graphical glitches Re: SATA HARD Disk in 286/386 Mobo: Is it possible?

rasz_pl wrote on 2023-02-12, 17:40:

Not the first case of modern IDE drive causing graphical glitches Re: SATA HARD Disk in 286/386 Mobo: Is it possible? https://ww […]
Show full quote

Not the first case of modern IDE drive causing graphical glitches Re: SATA HARD Disk in 286/386 Mobo: Is it possible?

Look at that 🤣, very similar effect, but it looks like the issue there was some signalling that SATA doesn't provide on the ISA bus. Given that CF is really just IDE, I wouldn't think the issue would be the same - although I don't know enough overall to rule it out.

ATA2 signaling was causing detection failure, graphical glitches were independent issue rayer was unable to resolve

Quick initial follow-up -

I received my SD to IDE (one of the cheapo ebay jobs) and it boots without graphical glitches at both 8 and 16Mhz.

I'm STILL waiting on my ps2 to AT keyboard adapters to arrive and then I can perform some actual real-world testing and I'll follow-up.

Further testing is underway - Here's where it gets weird...er.

The BIOS on the system board is too old to support autodetect for hard disk parameters. I used Tom Warren's WhatIDE to get cylinder/track/head info off of the SD card so I could input it manually.

DOS 6.21 install disks see it, partition it, format it and then the install fails at 99% with an error accessing the fixed disk. If you boot to a formatted floppy and then try to run anything off C:, it'll just sit until eventually it produces an abort, retry, fail.

You can FDISK the drive, format and SYS the drive, but it'll sit for a while on boot before eventually stating non-system disk or disk error.

I haven't been able to find ANY information on the disk controller card and it's jumper settings. I may need a new thread to try and track that down if anyone's ever seen one before.

Unfortunately I don't have spares of basically anything. I've seen a couple other reports of similar failures while installing DOS using these IDE/SD cards. If anyone wants to chime in, I'd appreciate it.

elvovirto wrote on 2023-02-18, 23:26:

DOS 6.21 install disks see it, partition it, format it and then the install fails at 99% with an error accessing the fixed disk. If you boot to a formatted floppy and then try to run anything off C:, it'll just sit until eventually it produces an abort, retry, fail.

Does your BIOS have "ant-virus" options that redirect and lock access to first sector of HDDs? If not, try doing fdisk /MBR booting from floppy, I've had DOS install itself and then fail to boot because, for some reason, the boot sector had the partition table but not a valid OS loader code in it.

It has no such thing - it's extremely bare and not user-friendly. Basic options are date/time, floppy a/b type, hard disk type (currently user 47, no autodetect). Advanced options allow for modifying wait states.

I tried an fdisk /mbr from a boot floppy after creating and formatting a DOS partition on the card and it accesses the floppy for a moment and returns to prompt without any notifications (I do not recall if this is normal). It made no difference. The SD card cannot be booted. Files CAN be written TO the SD card, but not run from or copied from the SD card. They will all eventually A,R,F.

What size is the SD card you're using? That BIOS most definitely won't support drives larger than 504MB, which I suspect your SD is. You need to either use a Disk Drive Overlay (such as OnTrack or EZDrive) or restrict the drive size to 504MB (1024 cylinders, 16 heads and 63 sectors per track). There are other solutions like adding an option ROM BIOS which supports LBA, but I just mentioned the most common approaches to solve this issue.

I started with a 2g card, assumed the same and switched to a pair of 32Mb cards I had on hand. There seems to be no difference.

elvovirto wrote on 2023-02-18, 23:26:

Further testing is underway - Here's where it gets weird...er. […]
Show full quote

Further testing is underway - Here's where it gets weird...er.

The BIOS on the system board is too old to support autodetect for hard disk parameters. I used Tom Warren's WhatIDE to get cylinder/track/head info off of the SD card so I could input it manually.

DOS 6.21 install disks see it, partition it, format it and then the install fails at 99% with an error accessing the fixed disk. If you boot to a formatted floppy and then try to run anything off C:, it'll just sit until eventually it produces an abort, retry, fail.

You can FDISK the drive, format and SYS the drive, but it'll sit for a while on boot before eventually stating non-system disk or disk error.

I haven't been able to find ANY information on the disk controller card and it's jumper settings. I may need a new thread to try and track that down if anyone's ever seen one before.

Unfortunately I don't have spares of basically anything. I've seen a couple other reports of similar failures while installing DOS using these IDE/SD cards. If anyone wants to chime in, I'd appreciate it.

It could be this, especially with a 1989 BIOS. Re: JCS286/SCAT (82C235) Motherboard

Your original screenshot was using OnTrack; but now you are talking about the 504MB limit. Are you still using OnTrack? It might be enough to get around this problem.

Possibly related to the hang on boot after installing DOS

elvovirto wrote on 2023-02-19, 00:32:

I tried an fdisk /mbr from a boot floppy after creating and formatting a DOS partition on the card and it accesses the floppy for a moment and returns to prompt without any notifications (I do not recall if this is normal).

This is normal behavior btw, you don't get any message with this command

konc wrote on 2023-02-19, 13:32:

elvovirto wrote on 2023-02-19, 00:32:

I tried an fdisk /mbr from a boot floppy after creating and formatting a DOS partition on the card and it accesses the floppy for a moment and returns to prompt without any notifications (I do not recall if this is normal).

This is normal behavior btw, you don't get any message with this command

Awesome, thanks for this, now I can at least know that for certain.

jakethompson1 wrote on 2023-02-19, 06:32:

It could be this, especially with a 1989 BIOS. Re: JCS286/SCAT (82C235) Motherboard

Your original screenshot was using OnTrack; but now you are talking about the 504MB limit. Are you still using OnTrack? It might be enough to get around this problem.

I haven't tried using Ontrack or any other overlay.

I did go so far as to prepare the cards using Rufus and RMPrepUSB, but that got me nowhere. I'm currently playing with 32Mb cards. The next smallest I have is 2Gb and after that it's basically 32Gb cards.

I honestly don't know much about BIOS replacement. I'm attaching a picture of the existing BIOS ROMs.