VOGONS


Reply 40 of 54, by treeman

User metadata
Rank Oldbie
Rank
Oldbie
darry wrote on 2021-07-24, 11:08:

This might be a dumb idea, but if all that is required for the switching to work is the proper logical level voltage on the switch pins depending if 3D is being used or not, would designing a simple "bodge circuit" that applies the proper logical level voltage based on the state of another signal pin that is always either high or low depending on whether 3D is in use or not be a possible alternative to chip replacement ?

I have thought about this too, it should work if get the voltages right but I am afraid if there is a partial short in the chip then most likely it will be slowly getting damaged with each use. I also note when I took out the 3d monster after measuring voltages the big chip was quiet warm, close to being hot. At the time I didn't think much of it but now I think after only scrolling the game menu a few times and having the pc on for a few minutes + it is winter here about 12 degrees Celsius I think it was hotter then stock.

Reply 41 of 54, by darry

User metadata
Rank l33t++
Rank
l33t++
treeman wrote on 2021-07-24, 11:22:
darry wrote on 2021-07-24, 11:08:

This might be a dumb idea, but if all that is required for the switching to work is the proper logical level voltage on the switch pins depending if 3D is being used or not, would designing a simple "bodge circuit" that applies the proper logical level voltage based on the state of another signal pin that is always either high or low depending on whether 3D is in use or not be a possible alternative to chip replacement ?

I have thought about this too, it should work if get the voltages right but I am afraid if there is a partial short in the chip then most likely it will be slowly getting damaged with each use. I also note when I took out the 3d monster after measuring voltages the big chip was quiet warm, close to being hot. At the time I didn't think much of it but now I think after only scrolling the game menu a few times and having the pc on for a few minutes + it is winter here about 12 degrees Celsius I think it was hotter then stock.

That makes sense, the best long term solution is replacing the damaged chip .

Reply 42 of 54, by treeman

User metadata
Rank Oldbie
Rank
Oldbie

I think I will put it aside for now, keep a lookout for any chips or damaged cards. I really don't want to ruin this card

Reply 43 of 54, by bloodem

User metadata
Rank Oldbie
Rank
Oldbie

Not sure if you've seen this video by Necroware, but it might help if you eventually decide on replacing the chip 😀

1 x PLCC-68 / 2 x PGA132 / 5 x Skt 3 / 9 x Skt 7 / 12 x SS7 / 1 x Skt 8 / 14 x Slot 1 / 5 x Slot A
5 x Skt 370 / 8 x Skt A / 2 x Skt 478 / 2 x Skt 754 / 3 x Skt 939 / 7 x LGA775 / 1 x LGA1155
Current PC: Ryzen 7 5800X3D
Backup PC: Core i7 7700k

Reply 44 of 54, by treeman

User metadata
Rank Oldbie
Rank
Oldbie
bloodem wrote on 2021-07-24, 12:12:

Not sure if you've seen this video by Necroware, but it might help if you eventually decide on replacing the chip 😀

I have seen that video before, he does a nice job. I have replaced a chipset before on a motherboard and I think next time I do such a repair the quickest way would be to clip off all the legs rather then use a heatgun to get it off. Since the chip is broken who cares how it comes off, the safest way for the rest of the card I guess. But it won't be easy or quick getting a replacement.

I am thinking aloud here but I could disconnect the pin from the pcb (maybe it won't be shorting internally if its left unconnected) then feed power externally to the switch chip from another source and attempt to do what Darryl said attach a voltage switch manually

Reply 45 of 54, by bloodem

User metadata
Rank Oldbie
Rank
Oldbie
treeman wrote on 2021-07-24, 12:21:

I have seen that video before, he does a nice job. I have replaced a chipset before on a motherboard and I think next time I do such a repair the quickest way would be to clip off all the legs rather then use a heatgun to get it off. Since the chip is broken who cares how it comes off, the safest way for the rest of the card I guess. But it won't be easy or quick getting a replacement.

IMO, a heat gun is the only way to do it properly and as safely as possible (unless you want to risk damaging some solder pads). I for one never had issues when using the heat gun.
Well, I should mention that the heat gun itself is also important. Industrial heat guns are not recommended for this type of work.

1 x PLCC-68 / 2 x PGA132 / 5 x Skt 3 / 9 x Skt 7 / 12 x SS7 / 1 x Skt 8 / 14 x Slot 1 / 5 x Slot A
5 x Skt 370 / 8 x Skt A / 2 x Skt 478 / 2 x Skt 754 / 3 x Skt 939 / 7 x LGA775 / 1 x LGA1155
Current PC: Ryzen 7 5800X3D
Backup PC: Core i7 7700k

Reply 46 of 54, by maxtherabbit

User metadata
Rank l33t
Rank
l33t
treeman wrote on 2021-07-24, 08:51:
This is most likely bad news, I traced pin 1 going through 2 caps to the bottom big chip, pixelfx processor according to Googl […]
Show full quote

This is most likely bad news, I traced pin 1 going through 2 caps to the bottom big chip, pixelfx processor according to Google. It would make sense that the processor sends the voltage when it is time to switch. So it is partially fried 🙁

IMG-20210724-184226.jpg
This is the pin that is directly connected to pin1 on the switch chip which is always on 5v. I couldn't find a pinout, does 1 exist? prob not. If it was one of the caps I guess it would either be shorted or open.

The good news? I see somebody on ebay is selling these chips, I have done a few qpfc chips before so I can do the job however not really looking forward to it or keen so undecided atm what to do. If my above diagnosis and thinking is correct

Edit: I checked on my working 3d monster and it seems that the pin on the pixelfx has about 1.3ohm resistance to the surrounding capacitors. On the faulty 3d monster that resistance is 0.030 enough to show continuity. However on both cards resistance from the capacitors to ground is ~. 400. So this most likely means there is some short inside the chip 🙁 not a external line where the caps are

I do think your diagnosis is correct. I also think replacing this video chip to solve the autoswitching problem is pure insanity. They always get hot, I don't think that's indicative of any internal short if it's still working in every other way.

Reply 47 of 54, by treeman

User metadata
Rank Oldbie
Rank
Oldbie

Yeah this is definately something I do not want to jump the gun on so I won't rush into anything.

Another test I just did is disconnect the leg from the main chip responsible for the voltage. Testing resistance on the disconnected pin now shows a open circuit, so perhaps I was wrong. Still shows .030 on the now not connected pad.. hmm such a guessing game without schematics

Reply 48 of 54, by treeman

User metadata
Rank Oldbie
Rank
Oldbie

I stripped all the caps and still same resistance, only thing left is to take off the switch chip. Took it off too and still 0.030 resistance, how can this be on a unconnected line?
Argh this time I traced the line under a microscope and yes I missed something.
One of the connecting vias also branch out to a hex inverter. This is the very last thing present on the line. I disconnected to pin that was branched off to my problematic line and now it is a open circuit. I am convinced this hex inverter is the 2nd piece to the puzzle

I guess I just dribbled a bit of garbage posting quicker then thinking here.

IMG-20210725-125613.jpg

I guess this is a good opportunity to swap in some new electrolytic caps for the old ones since they off and source a new hex inverter

Reply 50 of 54, by treeman

User metadata
Rank Oldbie
Rank
Oldbie
weedeewee wrote on 2021-07-25, 08:02:

Great troubleshooting!

Thank you, I am still dissapointed with myself for even thinking about removing the chip. I tend to get too excited and jump to conclusions before I fully think it over. Anyway I also tested the voltages on the working card yesterday. This may help somebody
in 2d
High voltage was 4.4V low voltage 66mv
in 3d
High voltage was 5v low voltage 166mv

Kind of makes it hard to hook up a manual switch varying voltages like that.

Anyway until I source a hex inverter will be a while, hopefully back with good news in weeks or perhaps months

Reply 51 of 54, by weedeewee

User metadata
Rank l33t
Rank
l33t

don't worry too much about it... live & learn or try, try, try try try, try try, try again.
there's a lot of us out there that do that, get too excited and shoot their load before having fully thought it over... sometimes a little experience goes a long way, then you can get excited, go all the way to fix it and still manage to blow it. 🤣

Which high voltage and low voltage did you list there exactly? working board, yes, but which ic, which pin ?

Right to repair is fundamental. You own it, you're allowed to fix it.
How To Ask Questions The Smart Way
Do not ask Why !
https://www.vogonswiki.com/index.php/Serial_port

Reply 52 of 54, by treeman

User metadata
Rank Oldbie
Rank
Oldbie

some dissapointing news, got the hex inverter. Installed it, resistance on the pin/line as correct now as on the working card. (so the hex inverter did have a problem)
System still starts up and doesn't pass through 2d.
Tested and still shows 3d ok.

Seems something is still not triggering the switch. Taking a break for now. I guess taking a measurement of voltages should be the next thing to see if there is any change at all from a working hex inverter.

Reply 53 of 54, by treeman

User metadata
Rank Oldbie
Rank
Oldbie

I tested voltage with the new hex chip and it is at 1.8~1.9V on 2d and 3d not making sense to me.

Going bit crazy here ended up swapping hex inverters from working 3d monster to broken one, results exactly the same. Have the new hex inverter on the working 3d monster now and it still works.

So the hex inverter is working. I am afraid this is gone far enough, changed the switch chip and hex inverter now. Only thing left is the big chip and like everybody said not worth changing. Think going to leave this where it is, got a second 3dfx card that works but doesn't auto switch to 3d, could of been worse

Reply 54 of 54, by treeman

User metadata
Rank Oldbie
Rank
Oldbie

So as final confusion I thought id share my findings which could be helpful how exactly the switch process works:

So the main chip will send 2 voltages to the hex inverter on pin 13 A3 also connected to the switch chip pin 1 BEa. This in turn will send a inverted voltage from pin 12 O3 direct to the switch chip on pin 13BEb
Screenshot-20210815-115728-2.png
Screenshot-20210815-115939.png

So sending a low voltage to switch chip pin1 BEa will also invert a high voltage to pin 13BEb
this is the signal for 2d mode
36.8mv - > A3 (hex inverter) + direct BEa (switch chip)
This gives a output from O3 on the hex inverter of 4.47v sent to the switch chip on BEb

For 3d mode the main chip will send 5.06V to the hex inverter on A3 + direct to switch chip on BEa. Once again we have a inverted voltage of 150.8mv coming out of O3 to BEb
* These readings taken on the working card

Now my chip being stuck on 1.8~1.9V is a mystery why it still works in 3d mode.
I had a good idea, I got a dc regulated power supply so I decided to manually inject voltage to A3 on the hex inverter. I simply soldered a wire to A3 ran it to the dc power supply and ran a ground wire to the pc case.

Starting the pc all plugged to the 2d card I got the bios beeps of no video card. I flipped on my dc power supply sending .3v and after 1 or 2 seconds I got 2d picture. The theory holds up.

I started swarrior 3dfx in dos and got a black screen as the game loaded. Still in 2d mode. I started turning the dial on the voltage up around only 1.5V the switch to 3d took place.

Conclusion:
- The ranges seem to be pretty elastic as don't need the 5V to switch to 3d looks like 1 - 1.5V is the spot, must be why my card is always stuck on 3d giving me a constant 1.8v
- I noticed that sending the voltage to switch to 2d I could turn off my power supply and it would stay in 2d and show the correct voltages from the switch chip, perhaps in my case the main chip needs a jump start?
- looks like I was right that the main chip has a fault at sending the right voltage but other components on the same line were damaged too, hex inverter + switch chip
- My old hex inverter was showing very low resistance to ground on A3, which coincidentally is next to vcc, I didn't check or can't remember but 0.17 resistance to ground sounds like vcc so it was most likely internally shorted to vcc, which would explain why on my original readings I was getting 5v, same power the vcc of the hex inverter to power it.
- It would be not too hard to rig a switch connected to pin A3 on the hex inverter to send either a low or high voltage but for me this would ruin the appeal of this card like putting cheap spoilers on cars. The card works, comes with a catch but will be workable through a kvm switch if needed.