most likely unobtanium. What exactly do you need? it looks like one of those common at the time integrated [GAME, 2 SERIAL, PARALLEL PRINTER PORT].

I have a mobo with integrated multi i/o.
Com ports are not working.
Trying to figure out why.

Perhaps a -12V rail is absent?

pshipkov wrote on 2023-01-01, 16:28:

I have a mobo with integrated multi i/o.
Com ports are not working.
Trying to figure out why.

Missing +/-12V would result in dysfunctional external communication, but correct operation in loopback mode. I once experienced a super I/O chip on a mainboard where the floppy part failed to work and the serial ports didn't run at all, not even in loopback mode. In that case, the source were corroded traces (NiCd battery, as usual) to the crystal that generates the baud rate and the FDC clock.

or if on headers you got wrong pinout? worst case scenario you can buy $10 ISA IO card and transplant it

Found a replacement and did the swap. Same result.
All is good with the PCB/assembly. No visible damage.
Can be a cold joint somewhere internally of course.
Without a schematic i am in the dark.
May need to find a working card with such chip, capture voltages/signals, then compare with the troubled mobo.

Any other ideas ?

For what it's worth I have one Super I/O card with 863F+865F combo. Both serial ports are damaged. The 865F has built-in RS232 level shifters and I suspect these are not very robust, either improper use (like shorts, which good RS232 chips should tolerate) or just age kills them. This is not a typical level shifter chip done with custom CMOS process or bipolar tech, it also has some 5V ISA decoding logic in it.

On my card first serial is dead (receive line is deaf but transmit still works), second kinda works but is unable to power serial mouse. What I've noticed is turning on RTS and DTS on the first port will somewhat affect the voltages output on the second port as well, and this is just enough for my mouse to start working. So I didn't trash it only because of this ugly workaround - these chips are unobtainium, and with yours also dead I suspect they are all faulty/fragile anyway so finding a donor card that works is going to be difficult. Also the card is just bad in general. The 863F is super picky about what CF card it'll work with, out of a dozen I have just one works properly.

Thanks for the notes.

Actually i swapped the chip with sourced "new" one that should be functional, but still no mouse.
So, so the problem can be anywhere really.
Still feel that if i have a good starting point for comparison i may be able to figure something out, if lucky.
Looking now at ebay to see what's available for cheap.

Few more details:
Running ctmouse /m results in no mouse.
Running it with ctmouse /m /s11 (or /s13) - system hangs.

Perhaps running some tests would help narrow down the cause of the problem? I like to jumper RxD with TxD and see if I can get echo - a simple DOS terminal is built into Norton Commander 4+
It's also good to test the voltages on the output lines (TxD, RTS, DTR) to make sure they meet RS232 requirements (quite frankly often they don't even on seemingly working ports, I've seen +/- 5V being used). It's also possible to jumper RTS/CTS and DTR/DSR pairs (or even try combination like RTS to DSR, etc) to see if terminal software recognizes the signals. This way it's usually possible to figure out which pin is dead. If you have another PC with known good serial port and a null-modem cable then it's even easier to test all this.
Since the 865F has the voltage shifters in it, the chip connects directly to the pins of the serial port socket. You can trace these to make sure everything is OK, and if there are any (possibly faulty) filtering parts like chokes and capacitors connected. In general the mouse will require pretty much all signals to work properly - RxD for data, TxD for control, other lines for providing enough power to the electronics (which is very low power but does need a few mAs, well the ball ones anyway).

Thanks for the notes.
This prompted me to actually compare pin voltages. Not sure why i didn't do it early.

Pin voltages on a working command port:

1   00 11 00 00 00
2-> 00 11 00 11 00

On the not working one:

1   00 11 00 00 00
2-> 00 00 11 11 00

This reminds me that there are (at least two) RS232 pinout standards.
Looks like i am hitting this case here.

(-> denotes first pin on the on-board COM1 header)

pshipkov wrote on 2023-01-02, 18:10:

This reminds me that there are (at least two) RS232 pinout standards.

Exactly. The traditional pin numbering of the 2.54mm headers is

12   4   6   8   .
21   3   5   7   9

The standard press-fit IDC connectors that connect these pins to a ribbon cable will get the pins in order, so on the cable, the layout is like this:

11 2 3 4 5 6 7 8 9 .

The official numbering on pins on a D-subminiature connector, size E, is

11   2   3   4   5
2  6   7   8   9

note that these ASCII arts are in the same scale, i.e. the horizontal pin distance between 1 and 2 on the D-sub connector is 2.54mm, just as the horizontal distance of the pins on the IDC header. Traditional cables connected pins with the same number. This requires hand soldering of the cable wires to the D-subminiature connector. There are press-fit D-subminiature connectors, but they only work if the cable has the signals in the matching geometry, which (quite obviously) is

11 6 2 7 3 8 4 9 5 .

To put the signals in this order, the pin header on the I/O card needs to be layed out with pins in the non-traditional order

16   7   8   9   .
21   2   3   4   5

A coarse rule of thumb is that pre-VL multi-I/O boards typically use the traditional layout of the pin header, whereas onboard I/O headers on mainboards use the layout that makes the break-out cables easier (and thus cheaper) to manufacture. As pin 5 is ground on 9-pin RS-232 ports, it's easy to buzz out what type of pin header you have, by checking whether the ground pin is at the edge or in the center of the 10-pin connector.

Yep.
Just rewired the pins and is all good.

Thanks guys and thanks for the additional notes mkarcher.