analog_programmer wrote on 2023-07-02, 10:22:

Nexxen wrote on 2023-07-02, 10:13:

Is the 150K still 150 when connected to the motherboard?
Only way to know is to desolder both. I guess that the 8.8 will read 10K, but the 150 is 150K, reading is correct.
https://kiloohm.info/smd3-resistor/154

The measured values of 12.2k instead of 150k for R77 and 8.8k instead of 10k for R78 (this one is not so far from the marked value) are from the resistors as they are soldered to the PCB. I have to unsolder them to check their actual values for sure.

There is still one problem I couldn't catch with the multimeter: If the solder joints of the resistors to the PCB are not good. At the end maybe the hot air gun with plenty of flux will "play" over these resistors.

My mistake, 12.2k isn't a correct reading. As it could be in a circuit you have to desolder it to check 100%. This is the "boring" phase.
You can always touch up the resistors you don't feel right about, won't do any harm.

There is an update with the situation around this cursed TNT2 Pro. Still I do not know if it is for good or worse 😁

Now officially this card has all symptoms of dead BIOS: the screen is completely dark, but system boots normally and no complains about videocard errors during POST (I always use POST test ISA/PCI card). How I achieved this?...

Well, I just removed the EEPROM (at least I thought its an EEPORM until today) chip during my checks for cracked or shorted SMD resistors and caps. None damaged SMD components found so far... I do not know exactly why I removed the BIOS chip, maybe some habit just to mechanically clean its pins as it is removed-reinserted into socket. And after this simple manipulation there was no more videosignal from the card.

So... Today I decided to try again reflashing of BIOS, but "in light" using my "self-revived" PCI S3 as prime videocard in the system. And I noticed something strange. The NVFLASH.EXE finds the TNT2 card, but can not flash nothing - does not recognizes the BIOS chip with "-c" argument. Ok, tried the FLASHROM (DOS version) - same story: with "-p gfxnvidia" arguments it recognizes the card as supported "programmer", but also complains that there is no EEPROM chip supported to read from nor to write to. So I decided to remove the stupid sticker "RIVA TNT2 PRO Copyright(c)99" which was covering the EEPROM to see what unsupported chip it is. SURPRISE! It is Amic A276308-70 OTP EPROM! No way! One time programmable at 12V and readable at 5V EPROM:

So it turns out that several days ago I've reflashed nothing "in the dark" with those autoexec.bat-ish automated flash attempts. It's OTP EPROM chip for even cheaper videocard.

I'm just wondering, can anybody here confirm possible OTP EPROM BIOS "bit rot"?

If this is legit suggestion I'll continue to search for supported EEPROMs 32 or 28 DIP package with acceptable price. There are not so many options thou. NVFLASH 3.18 supports these 64kb 5V EEPROMs:
SST 29EE512 , 64Kx8, 5.0V, 128B page
SST 39SF512, 64Kx8, 5.0V, 1B page
Atmel 29C512, 64Kx8, 5.0V, 128B page
Atmel 49F512, 64Kx8, 5.0,3.0,2.7V, 1B page
WBond W29EE512, 64Kx8, 5.0V, 128B page

Bit rot could happen, sure. My understanding is that OTP EPROMs are no different than standard EPROMs except for the lack of an erase window. I don't think they'd be immune from rot.

Do you have an EPROM programmer? If so:
First make a backup of the chip.
Not sure what the typical decay behavior is - it's possible that the only errors are in the direction of erasure (logical 1s that need to be reburned to 0s). If so then you could get away with simply reprogramming it and not doing an erase cycle (which is impossible with an OTP). It will complain that the chip isn't erased, but if all the programmed 0 bits are destined to be 0s anyway then it doesn't matter.

You'd need an accurate image of the original programming though. If the image is substantially different (from a different card) then of course there's going to be some mismatching bits that need to be erased to 1s.
Whatever you do, make a backup first in case you need to restore it (and you'd need a new EPROM at that point).

As for Flash ROMs - I've run into a situation on an Atari where Flash ROMs weren't compatible where an EPROM was expected, because of some subtle difference in how they behave. But I think that's rare and the card was probably designed to work with Flash, so it would probably work.
Check pinouts to make sure those chips you're considering would match the pinout of the original EPROM. If there's a mismatch then you'll need to look for any (soldered) configuration jumpers that might need to be moved to support the Flash chips.

shamino wrote on 2023-07-12, 23:08:

Do you have an EPROM programmer?

No, I don't have an actual EPROM programmer. Already tried, but I can't read this OTP ROM chip with software like NVFLASH, UNIFLASH or FLASHROM (DOS version) - they just do not recognize Amic A276308-70 chip.

Recently I've purchased some 128kb 5V EEPROM chip to replace another defective BIOS chip on some old Socket 7 mobo. When I receive it, I'll swap DIP28 socket on this videocard with DIP32 and try to flash on it some TNT2 Pro BIOS dump downloaded from internet. I hope this will work, altough the original TNT2 OTP ROM is 64kb.

I just read that OTP ROM chips from that time (as of this TNT2 Pro) have about 20 years data life. This videocard was manufactured in October 2001, so it's "time to die" because of degraded OTP BIOS chip has come.

shamino wrote on 2023-07-12, 23:08:

Check pinouts to make sure those chips you're considering would match the pinout of the original EPROM. If there's a mismatch then you'll need to look for any (soldered) configuration jumpers that might need to be moved to support the Flash chips.

Checked. The pinout of this Amic OTP ROM is same as of standard 5V 64kb or 128kb DIP28 EEPROM. Also there are 4 extra holes for DIP32 chips on the card and they are connected properly, but manufacturer soldered socket for DIP28 chips. So there's not a problem to be used DIP32 EEPROM chip. I'll swap the original DIP28 socket for DIP32.

Good to read that you are making some headway with the card, certainly a fun read this thread.

Fingers crossed a fresh bios chip gets the old girl running again!

analog_programmer wrote on 2023-07-13, 14:21:

shamino wrote on 2023-07-12, 23:08:

Do you have an EPROM programmer?

No, I don't have an actual EPROM programmer. Already tried, but I can't read this OTP ROM chip with software like NVFLASH, UNIFLASH or FLASHROM (DOS version) - they just do not recognize Amic A276308-70 chip.

You should be able to read these chips if you disable chip recognition (i.e. force the chip type). Choosing any kind of 64KB chip (27C512 or similar) should work. IIRC both UNIFLASH and flashrom allow forcing the chip type.

mkarcher wrote on 2023-07-13, 21:17:

You should be able to read these chips if you disable chip recognition (i.e. force the chip type). Choosing any kind of 64KB chip (27C512 or similar) should work. IIRC both UNIFLASH and flashrom allow forcing the chip type.

There is logical sense in this approach, I have to try it. But if the data in the original OTP ROM chip is already corrupted, then what's the point of "refreshing" it with the very same corrupted data?

Update:

I managed to make some dumps from the original OTP EPROM chip useing 3Com PCI NIC adapter as "programmer" and FLASHROM under DOS forcing it to read Atmel AT29C512 chip (as the original EPROM Amic A276308 can't be recognized as supported chip by FLASHROM.EXE) and... surprise! All five dump files read for about five minutes and one after another (without rebooting the machine) differ in many bytes, but in most parts are consistent. I also compared dumps to BIOS from shamino's similar card (which he shared in another topic) and turns out that my dumps are similar, but with many "broken" bytes. I'm attaching an archive with the five broken BIOS dump files, if anyone would be interested to look and compare them with a hex editor to see how the bytes in the chip "mutate" with each subsequent read.

P.S. I have one redundant DIP32 EEPROM chip, but it's 256KB, and now I'm wondering if I should order a new (rather refurbished) appropriately sized suitable EEPROM (64KB) from China...

Final update:

Two weeks ago I finally got brand new compatible EEPROM chip (Atmel AT29C512), I changed the 28-pin socket to 32-pin (everything is properly connected for 32-pin socket on the PCB) and flashed the BIOS-dump file given by shamino. The card didn't work, but at least now the motherboard beeps on POST for "no videocard" error, and with the original degraded OTP BIOS chip no beeps or errors are given on POSTing.

I don't know if this is due to incompatible BIOS flashed or the card is really dead now due to some hardware malfunction. I've no other suitable BIOS-dumps to test them.