@Bill1962, Intel's x86 architecture is little-endian.

https://en.wikipedia.org/wiki/Endianness

_________________
A backup a day keeps DR away.

Just a quick one, I let the disk stay in the computer, just removed the 40 pin cable, and hooked up the adapter there, and let the computer power supply power the disk, and then put the USB end of the cable in the other computer. No difference. And, I took the disk with me to the local computer shop, and they had several older adapters, and some of them let the disk run, and some not when you plugged the adapter in the disk. But, in the moment they put the USB connector in the computer the disk died. With all of their adapters.

The bios have predefined disks up to more than 1000 cylinders, and more than 100 Mb, so I can not imagine that is an issue. If I want a smaller disk to try, I have to buy one.

Ok, computer on the workbench. 0 volt signal on C17. Tested during writing FF for 1500 cylinders so I had plenty of time to test all the data lines, both from 0 to 7, and 8 to 15. There were also a little different voltage coming out. Some data lines gets down to just a bit over 3 volt.

I have ordered another motherboard. Or, is there anything to do about it ? I assume the problem lays in some circuits since it is not missing on only one socket. And it is 0 volt, it is not like it is 0.7 volt, or 1.3 volts or whatever. What I can do is when it has stopped writing the FF, to take the cards out and look for a shortening to ground for that C17 pin, it is awfully close to D18 which is ground.


Bill

All the ISA slots are bussed together, and all the pins are wired to the SARC RC2016 chip.

http://ru.pc-history.com/wp-content/uploads/m396f.jpg
https://www.philscomputerlab.com/uploads/3/7/2/3/37231621/2885086_orig.jpg
http://stason.org/TULARC/pc/motherboards/F/FORCOM-TECHNOLOGY-CORPORATION-486-M396F-1.png

With power disconnected, what resistance do you measure between C17 and Ground (as you have already suggested)?

BTW, some VGA cards can autodetect an 8-bit or 16-bit slot, while others need to be jumpered accordingly. Perhaps your VGA card is running in 8-bit mode?

_________________
A backup a day keeps DR away.

You might like to test the resistance of the VGA card and your multi-I/O cards to exclude the possibility that one of them is responsible for the apparent short between C17 and ground.

_________________
A backup a day keeps DR away.

Resistance between C17 and Gnd. is about 2 MOhm. And there is no shortening between C17 and Gnd. on any of the IO cards, two hdd controllers or the VGA card.

I wonder if that Sarc RC2016A5 chip can be found somewhere, but I guess it would be impossible for a 56 year old to solder that thing in place.

And now I need to learn about that "little endian" stuff that I did not understand. Well, I got that it is not straight forward, but I am not at the point that I could say it was Bit 14 that was the issue.

I have no idea about the VGA card, it is one of those things I have no contact with. I look at the monitor to see if it is anything there, that is about what kind of contact I have with it.


Bill

Frank, if the drive is not informing correctly one of the bits, could that affect the parameters that the BIOS use to auto configure the disk ?

@Bill1962, "little-endian" means that 0x1234 will be stored as 0x3412, ie low byte ("little end") first.

@rogfanther, I did check the Identify Device output. However, bit #14 does not impact on the CHS parameters,


http://www.users.on.net/~fzabkar/IDE-identify/FIND-ATA.DOC


So CHS -> 0x03E9 / 0x000F / 0x0011 -> 1001/15/17

However, the problem affects the partition table.

AIUI, here is the corrupt table (not visible in OP's screenshot):


This is how a good partition table should look:

_________________
A backup a day keeps DR away.

Hm, I need to in a very good mood for it and sit down with a piece of paper and see if I can figure out this "endian" stuff. We never went this deep in the computer education I have had. I need to understand it fully, so it sticks in my "skull"

There is one chip that sits in a socket, Sarc 6042, but I assume since you guys have not mentioned it even once it has nothing to do with this problem. However, it can not hurt to stick a thin screwdriver between the chip and the socket to move the pins in case of corrosion. And then push it back again.

I have been all over the board with a magnifier and a flashlight. And it looks very clean, nothing that should indicate anything wrong. However, I do not have any pinout for the 2016 circuit, so I can not check to see if it comes any signal out there. Or, maybe it even come from the processor which by the way gets very hot.


Bill

I suspect that "SARC" is a PCChips rebranding of the ALI chipset at stason.org.

Similarly, the 6042 chip would be the keyboard controller with its own "keyboard BIOS". The original IBM PC-AT had an Intel 8042 microcontroller in this position.

The other chip would be a real-time clock (RTC) and CMOS RAM.

The 16-bit data bus should originate at the SARC 2016 chip. Bus lines would be grouped together. They should be fairly easy to trace back from the ISA slots.

_________________
A backup a day keeps DR away.

In my attempt to reseat the 6042 IC, I ended up getting it totally loose, and here is a picture of the socket under.

I am not sure about this, but I have an idea that it is pin 39 we are talking about. The socket is a 40 pin socket, and on the right side, not the top one in the picture, number two down from the top. The spring tab on the outside is there, but on the inside it is pushed down all the way to the bottom of the socket.


Bill

Pin 39 is "TEST 1":

http://pdf.datasheetcatalog.com/datasheet_pdf/intel/8041AH_to_8742AH.pdf

Test 0 and TEST 1 are wired to Keyboard Data and Keyboard Clock (in the original IBM PC-AT).

http://www.users.on.net/~fzabkar/PC-AT/at-17of22.JPG

If the keyboard works, then I don't think it matters.

_________________
A backup a day keeps DR away.

Here is how the good and bad partition tables look in a disc editor (DMDE):

corrupt


good

_________________
A backup a day keeps DR away.

Why would the partition table start with head number 1, and not 0 ? I have seen it, but that is like missing a lot of capacity. It is about 8.67 Mb in one head on this drive.

I was also looking at the Glitchworks XT/IDE controller, if that is a possibility since I already have the board for another project. It would use only the lower 8 bits.


Bill

It starts on C/H/S 0/1/1. This means that only 16 sectors are unused.


An XT controller would be a lot more trouble. Try it. :-)

_________________
A backup a day keeps DR away.

Hello again, sorry this took so long time. I have bought a new (old) motherboard, a new (old) Video card, and brand new multi IO card. Today I put it together. Beep Beep Beep from the speaker, and nothing on the monitor. The monitor say no signal. So, I took my old Video card, and used that. And oh dear, you are not going to believe this, but the computer does not write 1 on bit 14, just like the old one. Everything identical.

I don't know if I shall laugh, or cry, or if there are some other emotional reactions I have never heard about I can use.

It must be the old video card that cause bit 14 to go low. But, there are no shortening between bit 14 and ground on neither the motherboard, nor the multi io card, nor the video card. So, it must be something that happens when it is powered up, and therefor impossible to find with a ohm meter.

The sad thing is that I have no idea how the new video card I bought is set up. It has some jumpers, but I am not able to find any info about this card online.

Good old Trident ....

Anything written in the back of the card ? There used to be manuals for this stuff, but I believe I let go of those a long time ago.

Without the SuperIO card, only with the new vga card, does any of your boards boot ?
( preferrably both )

I have a "VGA 8900C" user manual, but its jumper layout differs greatly (Asia Data, DSC-010C).

The jumpers at J4 could be selecting one of the "16-bit" IRQ pins (D3, D4, D5). A continuity check with a multimeter should confirm this. If correct, then ISTM that no IRQ is currently selected. I would also check where IRQ2 (pin B4, redirected to IRQ9) goes. My manual provides a jumper for disabling/enabling IRQ9 (= IRQ2).

It doesn't help, but here is the FCC filing for your card:

https://fccid.io/KC6-8900VGA

_________________
A backup a day keeps DR away.

BTW, I don´t remember needing to change jumper settings in our old vga cards. It seems to remember that there was some option to change the memory address used, but nothing in the type of needing to change jumpers to avoid conflicts. ( that was the domain of network cards.. )

@Bill1962, does the computer starts/access bios with only the vga card connected ?

With the new Video card, I hear Beep beep beep, the monitor is totally black, no video signal, so I can not tell how far it goes. So, the first I did was to remove the io card, to just see if something came up. I wondered if there was a conflict between the video and io card. But, still black monitor.
Then I used the old video card, alone, so only the motherboard and the video card, and it came up and I could access bios, and change the date and clock, and that was pretty much it.

I have found some information about a similar video card, and there is something about interleave, and a frequency, and some of them seems to have some setting for 8 bit or 16 bit motherboards, but since I can not find any information about this particular card, I am not sure of anything. It is written with a marker pen on the back of the card.... Good Well, it looks good, but looks and works is two different things.

On the back side it says, GC318 -PT4855 S/N:A1809-032, and on the warranty label it says: SN: 82313432 A

Jumper 4 is open on all 3 switches. What does ISTM mean ?

I found the FCC place earlier, but could not find any jumper settings there.