Absolutely, the test requires a loopback dongle which connects a few more pins to some other pins, including 2 to 3, RxD & TxD connected.
Without loopback dongle, the test is mostly meaningless.

I guess the pinout for a checkit serial loopback dongle can still be found on the internet somewhere.

edit: added themobrules reference for dongle pinout Re: serial mouse not detected on a 386 motherboard

I ordered the female end and will assemble such dongle.
My point was that without it i was not sure if this test is even meaningful.
Thanks for confirming that.

I do have oscilloscope. Can run a fdd-copy test and observe B22.
Will post the result later today.

pshipkov wrote on 2023-06-19, 21:30:

I do have oscilloscope. Can run a fdd-copy test and observe B22.
Will post the result later today.

Oh, cool. If you hook it up to the computer serial port pin 2 and type in the terminal program you should see the TxD signal change as well. Signal should change be between +/-12V

weedeewee wrote on 2023-06-19, 21:43:

pshipkov wrote on 2023-06-19, 21:30:

I do have oscilloscope. Can run a fdd-copy test and observe B22.
Will post the result later today.

Oh, cool. If you hook it up to the computer serial port pin 2 and type in the terminal program you should see the TxD signal change as well. Signal should change be between +/-12V

Can this thing run Windows 95? 8 or 16 MB of RAM installed?
I've got some old VB programs that toggle RXD, TXD etc via checkboxes.

Signal on ISA pin 22 changes with FDD-copy.
However Serial pin 2 is constant when outputting from terminal.

pshipkov wrote on 2023-06-19, 23:37:

Signal on ISA pin 22 changes with FDD-copy.
However Serial pin 2 is constant when outputting from terminal.

oh how I hate the internet these days... I might as well ask chatgpt in stead of googling it.
Sorry, try pin 3 .

May I suggest something? Check if the UART on that multi-i/o card is an 16450.
If so, I recommend to desolder it and throw it away. 16450's should have been exterminated since the 90s already, when quick modems appeared. Like an i8088, it's not even worth to be saved. ;)

As a replacement, I suggest to install a socket and get an 16550AFN or equivalent chip.
Modem programs, mice drivers and Windows 3.1x will be grateful for the replacement.
Anyway, this is just a suggestion, by the guy with the bad ideas.

According to the CheckIt results, which I've never seen as worse IRL,-with or without loop-back dongle-, the UART might be faulty.

Edit: According to my knowledge, CheckIt can do test the internal FiFo buffer without a dongle just fine.
As TheMobRules wrote, though, the question for the dongle must be correctly answered ("N").

pshipkov wrote on 2023-06-17, 22:55:

Tried this on a working setup - yes, there is some spamming in the console. But the same does not happen with the motherboard in […]
Show full quote

Tried this on a working setup - yes, there is some spamming in the console.
But the same does not happen with the motherboard in question here.

Noticed something about the CTMOUSE.EXE driver.
In the working system CTMOUSE /M reports "Installed at COM1 (03F8h/IRQ4) in Microsoft mouse mode". (or something like that)
But in the problematic mobo it reports "Installed at COM1 (03F8h/IRQ4) in Mouse systems mode".

---

+-12,22V - which is ok.
Currently i am using the ATX PSU with adapter for AT.
It produces slightly lower +-5 voltage. It was a problem once.
That's why i checked with trusted AT PSU that produces 5.0V solid.
No difference in this case.

---

So this looks like some physical damage somewhere.
At the same time video and sound cards work fine.

Is there a convention or rules which data lines on the ISA bus pass the mouse signals ?

So that post, since older mice you had to hold a button to set Mouse Systems mode, makes me think that one of the control lines DSR or DTR or one of those is stuck.

Also I had a similar problem to this a long time ago on a PCChips/VIA socket 7 motherboard, and went nuts trying to sort it out. What seemed to be the cause in the end was some BIOS corruption, the BIOS had got a bit error and wasn't handling serial ports correctly.

Jo22 wrote on 2023-06-20, 07:37:

May I suggest something? Check if the UART on that multi-i/o card is an 16450. If so, I recommend to desolder it and throw it aw […]
Show full quote

May I suggest something? Check if the UART on that multi-i/o card is an 16450.
If so, I recommend to desolder it and throw it away. 16450's should have been exterminated since the 90s already, when quick modems appeared. Like an i8088, it's not even worth to be saved. 😉

As a replacement, I suggest to install a socket and get an 16550AFN or equivalent chip.
Modem programs, mice drivers and Windows 3.1x will be grateful for the replacement.
Anyway, this is just a suggestion, by the guy with the bad ideas.

According to the CheckIt results, which I've never seen as worse IRL,-with or without loop-back dongle-, the UART might be faulty.

Edit: According to my knowledge, CheckIt can do test the internal FiFo buffer without a dongle just fine.
As TheMobRules wrote, though, the question for the dongle must be correctly answered ("N").

Ok, I'll bite.
It's a good idea to replace a 16450 with a 16550 considering the later has a fifo.
Though it's completely beside the problem since the card is working in another computer according to pshipkov.
just an FYI.

edit: googling the card mentioned earlier in the thread shows only one big chip. the available photos do not have enough resolution for me to clearly identify it.

Pin 3 is silent too.

Picture of the controllers i am testing with.
Tested with other models as well - UMC and GoldStar based ones - same result with the mouse on this mobo.
I use these because they handle very high ISA frequencies fine, especially the green one.

No such chips on them.

Winbond W83757.. Interesting. That's a Super I/O chip, actually. Strange find for a multi-i/o card, I assumed to find it on a motherboard only. A Pentium or late 486 model, maybe.
When I used ISA multi-i/o cards, it was all about 386/486 PCs.

https://pdf1.alldatasheet.com/datasheet-pdf/v … OND/W83757.html

Winbond is reported to be high failure rate for this.

Cheers,

pshipkov wrote on 2023-06-20, 15:37:

Pin 3 is silent too.

No output whatsoever ? !
and hard drive and floppy drive are working ?
and the card works in another computer.
Have you tried another video card?

@pentiumspeed
So, can you expand on your note please ?
Do you mean issues with serial ports, or something broader ?

As a side note - these controllers, along with Promise EIDE Pro (ISA), are probably the best i touched so far.
They are normal controllers really - handle both serial wiring standards, for example.
My point is mostly in terms of overclockability.
The green controller handles 27.5MHz ISA frequencies just fine. For reference.

@weedeewee
Yes to all your questions.
To confirm to your first question - the output is constant.

pshipkov wrote on 2023-06-20, 16:13:

@weedeewee
Yes to all your questions.

You should add a summary of all the things tried and measured to every reply you give here from now on. 😀

Stumped I am. its kinda like my mainboard with odd dma problem. dos didn't work, yet somehow windows does work.

edit: do you have any other multi io card that uses more chips instead of just this one? like a card with actual 16450/16550 chips on ? just for testing.

I know. We went in circles few times, but that's fine. I appreciate all the feedback. This is a puzzling issue for sure.

Speaking of puzzling issues - i know very well what you mean about DMA not functional in DOS, but working fine in Windows.
LuckyStar LS-486E revisions C2 and D are exactly like that.

I have less integrated cards. Let me see if at least one of them sports these chips.

pshipkov wrote on 2023-06-20, 16:34:

I know. We went in circles few times, but that's fine. I appreciate all the feedback. This is a puzzling issue for sure.

Speaking of puzzling issues - i know very well what you mean about DMA not functional in DOS, but working fine in Windows.
LuckyStar LS-486E revisions C2 and D are exactly like that.

There will be more circles because hardly anyone will read this whole thread.
So to recap :

The cards works on another mainboard
All voltages are present and accounted for +12,+5,-5,-12
Both floppy & harddrive work, also tested with different video card.
Card fails serial I/O, no matter what settings are used for address & IRQ, yet work ok on another mainboard.

Really? wow thanks, first I hear about it. mine's an ADI F4DXP-UC5D-I , different chipset from the lucky star