butjer1010 wrote on 2025-07-10, 09:57:

I have arduino memory tester, will try to test the modules. So TMM2018 can also be tested in T48? Will try that also...
tibpal20l8-25cnt and tibpal16l8-25cnt - could i test them somehow?
Thanks

A "PAL" is a programmable chip. So to test whether it behaves "correctly", you need to know how it is programmed, so except for basic tests like "are all inputs high impedance", there is nothing you can easily do with it.

Deunan wrote on 2025-07-10, 08:49:

There's also that TMM2018 which I suppose is the character RAM? That one is static memory and the pinout should be compatible with 6116, so the programmer might be able to test it.

Character information is stored in the main video memory on the EGA/VGA architecture, so "character RAM" is not the function of that chip. Maybe it is used for shadow registers for legacy 6845 emulation.

ASEM was an Italian manufacturer of PC:s in the late 1980:s

There are some youtube videos and tests of their PC:s in the magazines of the day, though I have not found any pictures of your card (other than your own).
It was possibly mounted in a 286 PC from their DESK 4000 or DESK 5000 series available late 1987, though I have not seen any detailed tests of those other than the DESK 5030, tested in MC Microcomputer in march 1989, but that one has a different looking VGA type graphics card. (Digging through archive.org)

Edit: they also had a similar mini tower series named 'Thor' with 286/386 inside. The top line 'Thor 9000' was tested in the magazine, but it was equipped with a Hercules card at the time.

Rwolf wrote on 2025-07-10, 17:18:

ASEM was an Italian manufacturer of PC:s in the late 1980:s

There are some youtube videos and tests of their PC:s in the magazines of the day, though I have not found any pictures of your card (other than your own).
It was possibly mounted in a 286 PC from their DESK 4000 or DESK 5000 series available late 1987, though I have not seen any detailed tests of those other than the DESK 5030, tested in MC Microcomputer in march 1989, but that one has a different looking VGA type graphics card. (Digging through archive.org)

Good to know, maybe i can find something more for this card...
Thanks

mkarcher wrote on 2025-07-10, 16:12:

A "PAL" is a programmable chip. So to test whether it behaves "correctly", you need to know how it is programmed, so except for […]
Show full quote

butjer1010 wrote on 2025-07-10, 09:57:

I have arduino memory tester, will try to test the modules. So TMM2018 can also be tested in T48? Will try that also...
tibpal20l8-25cnt and tibpal16l8-25cnt - could i test them somehow?
Thanks

A "PAL" is a programmable chip. So to test whether it behaves "correctly", you need to know how it is programmed, so except for basic tests like "are all inputs high impedance", there is nothing you can easily do with it.

Deunan wrote on 2025-07-10, 08:49:

There's also that TMM2018 which I suppose is the character RAM? That one is static memory and the pinout should be compatible with 6116, so the programmer might be able to test it.

Character information is stored in the main video memory on the EGA/VGA architecture, so "character RAM" is not the function of that chip. Maybe it is used for shadow registers for legacy 6845 emulation.

So only RAM can be tested more.... Thanks

Edit : I cannot test RAM chips : Didn't realize they were 4464 (18pins), my tester is for 4116-4164(256). Maybe i have few spares from my Amigas to try...

I have only 514256 from my Amigas, need to go to boxes to see if some VGA card has 4464 on it.
Before that, maybe to try those 2 jumpers, if the beeps stops....

butjer1010 wrote on 2025-07-11, 12:55:

I have only 514256 from my Amigas, need to go to boxes to see if some VGA card has 4464 on it.
Before that, maybe to try those 2 jumpers, if the beeps stops....

I dug into the BIOS of your EGA card. It has a beep function in it, that emits a single "standard" beep. On a 486 system, that beep will be a quite short beep, because it is timed just by executing CPU instruction (XOR CX,CX // LOOP 🤑. This "beep" function is used in two places: It is used if the video BIOS is asked to print a BEL character (character number 7, Control-G) in "TTY emulation" mode, and it is also executed once(!) if the memory test fails. That memory test covers the 4464 chips.

In case you get more than one beep, those beeps are not emitted by the EGA BIOS. Most mainboard BIOSes have a beep code for "no graphics card detected". On AMI, it is "8 beeps". Depending on the AMI BIOS version, the "significant beeps" are introduced by some higher pitched "error indication" beeps, which do not count. My first 286 system emitted two high-pitched beeps followed by 8 medium-pitched beeps in case the graphics card was missing or faulty. On Award, the code for "no working graphics card" would be "1 long, 2 short". If you get either of these codes, it seems your card is completely dead - well, unless one of the jumper fixes it. The original IBM EGA card had a jumper to set the I/O address to 2B0..2DF instead of 3B0..3DF - and this will make the card "disappear" for any software that does not explicitly try to access a secondary EGA card. I never used a function like that on any video card, and I expect most clones to not have it at all, but an EGA card set into the 2xx I/O port region would explain your symptoms without any hardware damage.

mkarcher wrote on 2025-07-11, 18:07:

butjer1010 wrote on 2025-07-11, 12:55:

I have only 514256 from my Amigas, need to go to boxes to see if some VGA card has 4464 on it.
Before that, maybe to try those 2 jumpers, if the beeps stops....

I dug into the BIOS of your EGA card. It has a beep function in it, that emits a single "standard" beep. On a 486 system, that beep will be a quite short beep, because it is timed just by executing CPU instruction (XOR CX,CX // LOOP 🤑. This "beep" function is used in two places: It is used if the video BIOS is asked to print a BEL character (character number 7, Control-G) in "TTY emulation" mode, and it is also executed once(!) if the memory test fails. That memory test covers the 4464 chips.

In case you get more than one beep, those beeps are not emitted by the EGA BIOS. Most mainboard BIOSes have a beep code for "no graphics card detected". On AMI, it is "8 beeps". Depending on the AMI BIOS version, the "significant beeps" are introduced by some higher pitched "error indication" beeps, which do not count. My first 286 system emitted two high-pitched beeps followed by 8 medium-pitched beeps in case the graphics card was missing or faulty. On Award, the code for "no working graphics card" would be "1 long, 2 short". If you get either of these codes, it seems your card is completely dead - well, unless one of the jumper fixes it. The original IBM EGA card had a jumper to set the I/O address to 2B0..2DF instead of 3B0..3DF - and this will make the card "disappear" for any software that does not explicitly try to access a secondary EGA card. I never used a function like that on any video card, and I expect most clones to not have it at all, but an EGA card set into the 2xx I/O port region would explain your symptoms without any hardware damage.

Beeps are 2 short, very quick, than 8 short, slower than first two beeps, and than after the keyboard blinks again, 2 short beeps again. Tried with both jumpers, all 4 positions, nothing changed 🙁
Found 8x chips 41464, and replace the memory on the card. Nothing had changed either.
So, probably the card is dead!

butjer1010 wrote on 2025-07-11, 18:40:

Beeps are 2 short, very quick, than 8 short, slower than first two beeps, and than after the keyboard blinks again, 2 short beeps again. Tried with both jumpers, all 4 positions, nothing changed 🙁
Found 8x chips 41464, and replace the memory on the card. Nothing had changed either.
So, probably the card is dead!

Sounds like an AMI BIOS telling you "no CGA/MDA/EGA/VGA-like graphics card installed (2+8 beeps)", and the later two beeps like mean "POST errors, press F1 to continue". So after pressing F1, the system will likely boot. Obviously without an image though, because the graphics card doesn't work.

mkarcher wrote on 2025-07-11, 19:19:

butjer1010 wrote on 2025-07-11, 18:40:

Beeps are 2 short, very quick, than 8 short, slower than first two beeps, and than after the keyboard blinks again, 2 short beeps again. Tried with both jumpers, all 4 positions, nothing changed 🙁
Found 8x chips 41464, and replace the memory on the card. Nothing had changed either.
So, probably the card is dead!

Sounds like an AMI BIOS telling you "no CGA/MDA/EGA/VGA-like graphics card installed (2+8 beeps)", and the later two beeps like mean "POST errors, press F1 to continue". So after pressing F1, the system will likely boot. Obviously without an image though, because the graphics card doesn't work.

Yes, unfortunately 🙁

mkarcher wrote on 2025-07-10, 16:12:

Character information is stored in the main video memory on the EGA/VGA architecture, so "character RAM" is not the function of that chip. Maybe it is used for shadow registers for legacy 6845 emulation.

Could be, but I was thinking (since this card should emulate CGA) that 2k SRAM would be perfect for user-definable CGA character set. But I have no idea what it does.

As for the card still not working, well maybe somebody did swap these PALs around. But I'm not sure it's safe to experiment with that, it's not going to be a short to power rails but could perhaps damage some outputs. There are two 16L8 that might have something to do with the clock switching, and two 20L8 that I belive are address decoders. Assuming the main chip is not simply dead, this is where easy solutions stop and hard work with scope and perhaps logic analyzer begins.

Deunan wrote on 2025-07-12, 11:53:

Could be, but I was thinking (since this card should emulate CGA) that 2k SRAM would be perfect for user-definable CGA character set. But I have no idea what it does.

Generally, my oppinion is that this would be unlikely. You already have the logic to operate CGA-compatible text modes (i.e. the standard modes zero to three) using RAM fonts, why would you implement a second text mode logic for CGA register-level emulation.

Specifically for this card, I looked into the BIOS, and I am quite confident this 2K RAM is working space for the NMI-based 6845 emulation, which can optionally be mapped at A000:0 using proprietary Paradise EGA registers.

Deunan wrote on 2025-07-12, 11:53:

As for the card still not working, well maybe somebody did swap these PALs around. But I'm not sure it's safe to experiment with that, it's not going to be a short to power rails but could perhaps damage some outputs.

Most logic level outputs do tolerate a short time (several seconds) short to either GND or +5V without taking permanent damage (maybe not true for bus driver outputs). A short between two active outputs is even less severe than a short to a power rail. So I expect the swapping of the two 16L8 will not cause permanent damage if just done for a short test whether the card is recognized. As an extra precaution measure, one might want to verify that no I/O pin is shorted to a power rail (this is OK if the PAL has this pin permanently configured as input) before randomly putting a different PAL into that socket.

Deunan wrote on 2025-07-12, 11:53:

There are two 16L8 that might have something to do with the clock switching, and two 20L8 that I belive are address decoders. Assuming the main chip is not simply dead, this is where easy solutions stop and hard work with scope and perhaps logic analyzer begins.

If you look up other PEGA2A-based cards on the Internet, you will find that many of those cards use a 2-chip configuration, with a "PARADISE PBI" secondary VLSI. It seems the PALs and possibly the SRAM as well replace functionality integrated in the PBI chip. I noticed next to the PAL sockets there are indications for the required progam, named "L208801A" to "L208803A", possibly "L208804A" hiding below the last PAL. Some PAL chips have a small ROM area used for a program ID, so if the TIBPALs do, they are not read-protected and supported by the T48, there might a chance that the chips contain that ID in the program ID location, and the correct placement can be verified.

The first test I would do with that card is add a MDA/Hercules card to the same system and check whether the EGA card responds to any I/O, memory or BIOS cycles. This requires you to have a Hercules (clone) card and a solution to display MDA/Hercules video (either an old CRT or some digitizer). If you know what you are doing, you can also CTTY COM1 and use DOS debug to analyze the situation, but I'm afraid this goes way beyond what butjer1010 wants to invest in trying to troubleshoot the card.

[/quote]
The first test I would do with that card is add a MDA/Hercules card to the same system and check whether the EGA card responds to any I/O, memory or BIOS cycles. This requires you to have a Hercules (clone) card and a solution to display MDA/Hercules video (either an old CRT or some digitizer). If you know what you are doing, you can also CTTY COM1 and use DOS debug to analyze the situation, but I'm afraid this goes way beyond what butjer1010 wants to invest in trying to troubleshoot the card.
[/quote]

Problem is not what I want to invest in trying to troubleshoot the card, problem is can I regarding my knowledge. I can do the solder job great, but diagnosing is beyond me 🙁
I have one MDA CRT and 286PC with MDA graphics card, I can bring it down from attic, if there is some easy solution even I can do 😀

butjer1010 wrote on 2025-07-12, 13:51:

I can do the solder job great, but diagnosing is beyond me 🙁
I have one MDA CRT and 286PC with MDA graphics card, I can bring it down from attic, if there is some easy solution even I can do 😀

OK, you can do this: You boot using the MDA card while the non-functional EGA card is installed as well. To do that: If your "Standard CMOS setup" includes a "video type" option, choose "MONO". If your board has a MONO/COLOR jumper, set it to "MONO", then install both cards. I expect DOS to boot perfectly with display on the MDA CRT. At the DOS prompt, do this:

1C:\>DEBUG
2-dC000:0
3C000:0000  FF FF FF FF  FF FF FF FF  FF FF FF FF  FF FF FF FF  .... .... .... ....
4....
5C000:0070  FF FF FF FF  FF FF FF FF  FF FF FF FF  FF FF FF FF  .... .... .... ....
6-i3c2
7FF
8-i2c2
9FF
10-q
11C:\>

This simulated DEBUG run contains both your input and the output of DEBUG I am afraid of - a completely non-responsive EGA card. If it looks significantly different, it is a good sign. In any case, please post the actual result of that DEBUG run (a screen photo or a manually typing it down, the exact values in the dC000:0 block are not that relevant you need to type it down completely, but the first line might give a good impression in whats wrong in case it is not plain FF) a picture from the back side of your card, so I can try to follow some traces to find out whether the reason for the failure might be a broken trace, or at what location you might need to take further measurementes.

Woow, thanks a lot.
Will do that first thing in the morning, and make a picture of screen i get.

I couldn't make it today, as soon as i can make it, i will let You know (too much work to do).