There is VERY CLEARLY incorrect geometry being reported.

What kind of CF card are you using? It may not be properly supporting old CHS modes.

wierd_w wrote on 2024-06-06, 07:32:

There is VERY CLEARLY incorrect geometry being reported.

What kind of CF card are you using? It may not be properly supporting old CHS modes.

It’s a Kingston 8gb ( https://www.amazon.co.uk/Kingston-CF-8 ... B001NXDB4A)

I did clear (set it to NONE) the CHS in bios and boot with a disk and run ideinfo again with the similar error.

I am going to go out on a limb, and ask you to aggressively test your memory, then also suggest moving the IDE controller to a diff slot.

If integrated, try disabling and using a multio card.

The values reported look like either the high or low byte is stuck high, and ghe ident string is gibberish.

So, either

some part of the 1st mb of ram is bad... and data being queried is being stored wrong...

The data bus leads in that isa slot have crud on them, and data coming back from the query is corrupt.

The controller is in some other way hooziered

The IDE device is hooziered.

So, to me, the diagnostic order of elimination, from easiest to hardest fix, is:

Aggressive ram test.

Test with different IDE device

Test of controller in different slot.

Test of different controller.

In that order.

Is the 500MB HDD known to work? I've seen some bad HDDs do that.

What is your actual CF card? One time you say it's 128MB the other time it's 8GB. 8GB is way to big.

kixs wrote on 2024-06-06, 23:15:

Is the 500MB HDD known to work? I've seen some bad HDDs do that.

What is your actual CF card? One time you say it's 128MB the other time it's 8GB. 8GB is way to big.

Yes, 500mb hdd are working on my 486DX2

I have both CF card 8gb and 128mb .

wierd_w wrote on 2024-06-06, 18:35:

I am going to go out on a limb, and ask you to aggressively test your memory, then also suggest moving the IDE controller to a d […]
Show full quote

I am going to go out on a limb, and ask you to aggressively test your memory, then also suggest moving the IDE controller to a diff slot.

If integrated, try disabling and using a multio card.

The values reported look like either the high or low byte is stuck high, and ghe ident string is gibberish.

So, either

some part of the 1st mb of ram is bad... and data being queried is being stored wrong...

The data bus leads in that isa slot have crud on them, and data coming back from the query is corrupt.

The controller is in some other way hooziered

The IDE device is hooziered.

So, to me, the diagnostic order of elimination, from easiest to hardest fix, is:

Aggressive ram test.

Test with different IDE device

Test of controller in different slot.

Test of different controller.

In that order.

Did a memtest86+ 4.0 last night and it passed all tests. I also tried different IDE HDD, 500 MB (working HDD that I use in my 486dx2) and ideinfo.exe also output the same error.

Next is to test different controllers which I currently do not have and will be ordering.

Since I'm ordering, I'm thinking to order lo-tech XT-CF (https://texelec.com/product/lo-tech-xt-cf-lite-rev-2/) or may be going Picomem (https://texelec.com/product/picomem/?highlight=picomem) route.

amontre wrote on 2024-06-06, 02:41:

Is this the sign of failing board ?

The geometry should be 992/8/32, and that's 0x03E0/0x0008/0x0020 in hexadecimal. What you have there on the screen is 23520/22536/22560, which is 0x5BE0/0x5808/0x5820. Notice anything interesting?
If we subtract 0x5800 from each value we get correct results. In fact it's even easier than this, we need to mask bits rather than subtract, a simple logical AND will suffice. In other words some of the bits in the upper byte are stuck at 1, but not all. It's not a simple issue of the upper part of the 16-bit ISA slot not working at all. It does work but it's glitched somehow.

Try different slot - in fact try all of them, because these are daisy-chained but you don't know in what order (it can be more of a T connection than a simple chain as well). Make sure the IDE card is clean, try a different one perhaps.
Also check your BIOS settings, perhaps you've set some very aggresive ISA clock divider?

Deunan wrote on 2024-06-07, 13:31:

The geometry should be 992/8/32, and that's 0x03E0/0x0008/0x0020 in hexadecimal. What you have there on the screen is 23520/225 […]
Show full quote

amontre wrote on 2024-06-06, 02:41:

Is this the sign of failing board ?

The geometry should be 992/8/32, and that's 0x03E0/0x0008/0x0020 in hexadecimal. What you have there on the screen is 23520/22536/22560, which is 0x5BE0/0x5808/0x5820. Notice anything interesting?
If we subtract 0x5800 from each value we get correct results. In fact it's even easier than this, we need to mask bits rather than subtract, a simple logical AND will suffice. In other words some of the bits in the upper byte are stuck at 1, but not all. It's not a simple issue of the upper part of the 16-bit ISA slot not working at all. It does work but it's glitched somehow.

Try different slot - in fact try all of them, because these are daisy-chained but you don't know in what order (it can be more of a T connection than a simple chain as well). Make sure the IDE card is clean, try a different one perhaps.
Also check your BIOS settings, perhaps you've set some very aggresive ISA clock divider?

Yes you are correct 992/8/32 is the CHS I got when connect it on other computer. The mother board has only 1 IDE hdd .

I did try different IDE with the same results on ideinfo.exe

amontre wrote on 2024-06-07, 15:26:

The mother board has only 1 IDE hdd .

Ah, so it's a mobo with built-in IDE controller. In that case it's quite possible there is a problem with signal connection to that IDE connector. The glitching bits are 11, 12 and 14, That would be pins 10, 12 and 16 on the connector.
First check if the pins of the connector are not corroded or very dirty. Probably not so the next step would be to check if the mobo doesn't have a scratch somewhere that cut these 3 signals. The chip that provides HDC and FDC functions is that Acer M5105 that's also on the photo. Check if the pins I mentioned are connected to some of the pins of this chip. If there is no connection you need to look for mobo damage. If there is a connection... well, the chip might be damaged.

Deunan wrote on 2024-06-07, 21:29:

amontre wrote on 2024-06-07, 15:26:

The mother board has only 1 IDE hdd .

Ah, so it's a mobo with built-in IDE controller. In that case it's quite possible there is a problem with signal connection to that IDE connector. The glitching bits are 11, 12 and 14, That would be pins 10, 12 and 16 on the connector.
First check if the pins of the connector are not corroded or very dirty. Probably not so the next step would be to check if the mobo doesn't have a scratch somewhere that cut these 3 signals. The chip that provides HDC and FDC functions is that Acer M5105 that's also on the photo. Check if the pins I mentioned are connected to some of the pins of this chip. If there is no connection you need to look for mobo damage. If there is a connection... well, the chip might be damaged.

If i add multi io card or something like Lo-tech XT-CF-lite rev.2 , i can bypass using the onboard IDE right ?

Probably not. Unless there is a jumper on the mobo to disable the built-in IDE, which I doubt. Also that card you mentioned (Lo-tech XT-CF-lite rev.2) is 8-bit only and most likely uses XT, not AT, port for the IDE channel. So it will be slow at best, might not work properly at all in AT system.

What you need is multi-I/O card that can be jumpered to secondary IDE channel, but using normal AT port addresses. And then you'll need XTIDE ROM extension with that as well because the mobo BIOS surely won't be looking for secondary IDE.
With this combo you can leave current on-board IDE as-is, just not connected to anything, and XTIDE can be configured to look for HDDs on the second channel. That would in effect be a bypass and it would not require the existing IDE to be disabled. The problem with this setup is you'd need two extra ISA cards (the I/O one and a network card to be a carrier for XTIDE ROM) and you are already limited with the number of free slots. And there might be some other issues, like which chip is now going to be the floppy controller - most likely it'll need to be the mobo one (since it can't be disabled I bet) and you'd have to configure the extra I/O card accordingly.

All in all it would still be best to try and fix the on-board IDE channel, if possible.

I have solved this by booting XUB on XT-CF