Ther are two types of COM port wirings, one where each write goes sequentially to each pin and ones where wires are staggered. Neither are compatible, older things use sequential while newer use staggered. Most of my AT cases have sequential wiring but many of my P1 boards expect staggered and all P2 and newer only use staggered from my experience.

12345
6789 (sequential)

vs

13579
2468 (staggered)

Thank you for that information! I wasn't aware of that.
This means that i should always try 2 different options to set up a loopback. (pin 2 and 3 connected)
Thank you!

Unfortunately it didn't give me any results this time. Also tried separate set of 2 com ports i have laying around just to make sure, it's not cabling related problem.

What could i try next?

There are chips that convert 5V levels of the UARTs to +/-12V levels of RS232, usually done with 75232 chips (20 pin chips). Make sure they get +5V, +12V and -12V or things will not function, if all voltages are ok then you'll try to measure the inputs and outputs and see if there's any activity when COM ports are written etc. If inputs show stuff and outputs don't you know the level converter chip is damaged, but if inputs are gone too then the problem is in the UART itself. This is assuming there are no configuration issues.

Suured tänud sulle, Tiido!

I'll try it out soon. I have proper hooks available to attach to chips.

Should ordinary multimeter be sufficient or should i measure the inputs/outputs in some other way?

hahaha, np 😜

Multimeter is probably not gonna be fast enough to register a bip of data passing on UART output but a LED will briefly light up (use something like 1k series resistor for COM port side, due to +12V rather than +5V). Reducing baud rate to lowest possible will make the bip longer and more easily noticable.

I had a problem with this too on 486 motherboards. I finally got it to work after rearranging the wires in the serial ribbon plastic connector. I don’t remember the exact order of the wiring but I was able to find a diagram on the internet to rearrange the wires. Google “serial connector diagram or wiring”.

I think I'm onto something.

I was able to find TX line on 5V side with LED. By juggling with COM port pins, i also confirmed 12V RX side. (Seeing jumbled characters appearing on screen)
What seems to be case (and looking by bare eyes, seem to be confirmed by LED), 12V TX is stuck on HIGH state.

So onto a chip change, i guess? It's interesting, that both ports are down, each has it's own chip for driver...

Page i also used for reference: http://www.frontx.com/pro/cpx102_2.html

My motherboard pins seem to comply with the last picture the webpage has. (TXD on Line 3)
As my ports are all with clip-on ribbons (not pin-by-pin, so i could rearrange them), then as i understand, i would need staggered pattern on the port side, not in a sequence as on that pages last com port soldering is shown. Unfortunately i have only sequential ones, therefore i need to start soldering as well, i guess 😜

That seems to indicate that the chips are indeed faulty. I can imagine it is possible to damage the chips by connecting devices to them with COM header with wrong wiring. One of my VLB IO cards' COM ports stopped working when I used wrong type cable on it, no mouse ever worked afterwards on it and I'm pretty sure the driver chip is dead on it.
Nowdays I make sure what the pin ordering is on the ports. One way to tell is to see if a fat GND trace connects middle of the header (sequential) or edge of the header (staggered).

Tiido wrote:

One way to tell is to see if a fat GND trace connects middle of the header (sequential) or edge of the header (staggered).

I'm going to remember that probably as long as i live now...

Thankfully, when i wasn't (unfortunately) too much interested in giving new lives to computers, i did desolder lots of chips from them and stored those away -> seems, i have 2 replacements already with me.
So let's see, how much time i have during christmas time to maybe repair it. I really love that board, as it pretty much has everything i want a Pentium 1 machine to have...

So, first of all good news: Both COM ports are now working! And even better - i had to change only one 75232 chip... but not for reasons i'm proud of...

Lesson nr 1: NEVER USE PICO-PSU TO TEST STUFF. (Yes, it's easy and nice, BUT IT DOESN'T SUPPLY -12V!)
When 75232 doesn't get it's negative 12V, it's not going to work.

Lesson nr 2: Do stuff one-by-one with easyest things first.
I jumped on to the motherboard being certain to replace those two chips. If i had resoldered the ports and just gave it a quick try, i wouldn't have touched the motherboard...

Only mistery is, why i wasn't able to do a loopback test, when i tried different RX-TX pins in the very beginning...

Yet another lessons to take with me and maybe for someone else reading this later to not make same mistakse i've made.

I now have fully working motherboard with one 75232 having very bad-looking solder joints. 😜 (Those SMD chips are a bit over my skills with soldering iron...)

A tiny UP in the light of the whole thread.

I now ordered this picu-psu from aliexpress: https://www.aliexpress.com/item/DC-DC-24pin-A … 2833975761.html
This one properly supplies -12V as well. Tested all the ATX voltages by the specs, and this one is fine. (Y)