It sounds like the level converter chip has partially given up, on this particular board it is the smaller Winbond QFP near the COM ports. W83768F on some photos I could find.

Did you ever test your Gerald's Retro PC Zimodem on any other computer ? If so what what other computer ? If not I would not.
It could have a flaw that caused a overload in chip Tiido mentioned. The datasheet for W83768F has under absolute max ratings for certain things and besides a flaw if you used a USB wall-wart type PSU into the Zimodem it could have supplied more volts than typical causing a voltage overload on COM1. Just a thought....

Tiido wrote on 2022-12-01, 02:33:

It sounds like the level converter chip has partially given up.

Thanks, I've found them at utsource, hope it will help.

Horun wrote on 2022-12-01, 03:12:

Did you ever test your Gerald's Retro PC Zimodem on any other computer ? If so what what other computer ? If not I would not.

Oh, no, I did not. But I guess next time I have to establish a policy to test on some more expendable hardware.
Like I said before, I've used so many RS232 devices in the past, and in so random and messy environments, but never actually fried anything.
Definitely I was not expecting permanent damage.

Thanks!

--
mg

Whole idea behind RS232 was being bulletproof. Its hard to fry level translators.
Do you have a multimeter? If you are game for a hack it might be possible to patch your Serial port by remapping less used lines. In theory full RS232 uses 5 inputs and 3 outputs. Most stuff with only use 1 input and one output.

As for the ESP8266 device, can you post picture of the thing? Have you rechecked pinout? I dont think even wrong wiring or pushing 5V into one of the pins would destroy RS232 transceiver, most likely port on your board was simply tired after 27 years 😀.

In my experience there a very few coincidences with anything. If mouse worked fine on COM1 before testing the "untested" Zimodem on COM1 (which did not work) and now mouse does not work on COM1.
Makes me suspicious of the Zimodem. Example of same but different: My AGP slot quit working after testing an untested AGP video card but AGP slot worked fine with original AGP card.
Or my floppy controller worked fine until I tested an untested floppy drive but it worked fine with the original floppy drive 😁

At the moment I'm waiting for the replacement chip, I suppose that would be optimal solution, to have those two ports fully operational. And in the meantime I'm using COM2.
The Zimodem was generally a sealed theoretically and ready to go box with DB9, so no, were not checking it before plugging in, unfortunately. I will make some photos tomorrow and post.
I do have multimeter, yes.

--
mg

>Patched RX with TX with a piece of wire and checked with terminal emulator, also to no avail.

Put a multimeter at pin 3 (TX) and write something in terminal. My bet its only RX that blew.
Do the same with Zimodem and look at the voltage on its TX pin, should be within +- 30 Volts.
RS232 was designed to be idiot proof, You can short it to ground, to power supply, it should survive all of that abuse. There is no way for Zimodem to generate a signal able to kill a transceiver from its max232 (or whatever translator its using).

What might have happened is you powered Zimodem from another computer, you have bad ground(or no ground), and while plugging it in shield touched serial pins. Bad grounding issues is another kettle of fish altogether.

If the 83768 was set for Low-V then the maximum for driver part is not +-30v but just above VCC/VDD max's (+/-13v), just an observation...

At first I thought that cant be right and assumed you must have seen VSS/VDD max, but there indeed is no mention of any other max characteristics 😮. Good old 1489 this is supposed to be replacing is +-30V. The f Winbond... did they really manufactured serial transceiver not conforming to EIA/TIA-232 standards?

The data sheet does state for the line driver/receivers: RS 232C (which is the low-V version, quote from diff source): RS232C is just the third version of the EIA's Recommended Standard 232. The current version is revision F. The major difference between the original standard and the C version was that the voltage levels for the signals were reduced from +/-25V to +/-12v and +/-5V. Not sure if that is correct but somewhat matches the sheet...