Another epic fix! Such great detective work! You're spot on. When there is any damage and the thing isn't working right, extra scrutiny must be given in the general vicinity of the damage.
An interesting project A while back in the retro web the talked about a project to put a pentium o pentiumoverdrive in a 386. I could be interesting to bring that here
A perfect demonstration of fault finding and where to look, fractures in resistors ...I learned something new here, Thank You. Great result ...awesome video.
Just recently I finished the restoration of very damaged 386 motherboard. Battery ravaged and some crushed components. The Last fix was a resistor arraya with an almost invissible crack. I tried to fix the array common line and Just use one resistor for the cracked one, to no avail, so I used an small array. Later relpaced the whole arrayanes. But for one crushed, I madera my one with discrete resistos as I dont have a replacement
Inspired by your videos, I bought a 3Dfx Voodoo 1 on eBay a few months ago. Like so many of your video examples, some of the 3Dfx chips had bent pins. I have a very modest wifi microscope setup, but I was able to straighten & reflow them. I have been able to get it to work a couple of times, but unfortunately the card still doesn't work reliably. Perhaps it's time to look for further damage on other components to see if perhaps it has a cracked resistor on it somewhere too. Thanks for your videos. Without them, I wouldn't even be attempting these repairs!
I am happy to hear that! Congratulations on your attempt! Fixing pins on 3Dfx chips isn't easy - even on Voodoo 1 cards! Even though the card doesn't work reliably, you are a big step closer to a fully working card! Let me give you a few pointers: Is there any pattern you notice when the card is unreliable? For instance, when the card is cold or hot? I would pay close attention to the solder joints between 3Dfx chip legs and the pads on the PCB. Even if they look to be fine, there could be a poor electrical connection. If you carefully poke those legs and they move (even a tiny bit), I would reflow the entire chip. Next, as you said, check for cracked components. Any scratch on the PCB might give you hints where to look. I hope this helps. Let me know if you can find something or if you need further help!
@@bitsundbolts Well. I need to source a floppy drive or a floppy drive emulator :( , the Win98 CDROM isn't bootable😂 (funny enough, I do have a Win98 SE original boot floppy). Furthermore, I need to dig the archive and find an IDE HDD or be patient as an IDE to mSATA converter on it's way. Setup: Abit BH6 Intel Pentium III @ 750 mhz (originally a P3-1000, but a 133 fsb model) 512 PC-133 SDRAM Diamond Multimedia (Nvidia) V770 Diamond Multimedia Voodoo 2 12 MB in SLI Diamond Multimedia Monster Sound MX300 3Com 3C905x Apart from the SLI Voodoo 2, which was in use in a previous P2-450 system, the setup was put together between October~December 2000. The FDD and IDE drive that I had put into the storage got wet due to a leak in our garage roof last winter. This went unnoticed until June of this year when I was doing a check. Bummer. I think I'll pay our local recycle shop a visit, perhaps the have a fossilized FDD on stock.🤣 Update: Well it appeared the problem is in the setup. After setting the IDE DVD drive as master on the second channel it will actually boot the Win98 CD. Incidentally, I found a 200 GB external drive (Western Digital), so the question is: to shuck or not to shuck ;). The SLI Voodoo 2 is 'recognized' at bootstrap as they appear as multimedia devices.
@@bitsundbolts Currently putting the setup together. The system used to be my main, later on went to my mom when I upgraded to a P4. Setup: Abit BH6 Intel P III-1000 @ 750 (133 fsb model, socket 370 in an Abit Slotket !!!) 512 MB SDRAM PC-133 Diamond Viper V770 (replaced an Intel i740) Diamond Multimedia Voodoo 2 12 mb SLI Diamond Multimedia Monster Sound MX300 3Com 3C905x NIC FDD: no NEC DVD-reader/writer HDD: Either a WDAC 200 GB IDE drive (will require shucking as it is an external drive) or a 128 GB mSATA drive (IDE to mSATA adapter on the way). Initial problem with booting from CD-rom were mitigated, setting it as master on the second channel and giving the Win98 CD a polish did it. For now the SLI Voodoo 2's are detected at bootstrap, so looking good. Once the storage has been solved, real world testing can start.
Brilliant, just came home early from a shit day at work in near freezing temps. Sat down with a cup of tea, watched you tinker and the moment you landed on the damaged array and we could see it cracked real time made me laugh. You took me from down in the dumps to happy as can be in no time, I mean what was the chance of you landing on it like that, even with the visual trace damage. Epic repair all round mate, thank you for the upload.
I am sorry to hear that your day wasn't great! But I am happy that you enjoyed the video and that it made you laugh :) Thanks for watching and hope your day picks up from here!
Hello! Thank's for good content. My brother had the same problem with his card. After much searching, I found that about 15 resistor arrays had lower resistance than expected. After replacing all the faulty resistor arrays, the card worked!
Congratulations on that repair! Interesting that the resistance reduced on so many resistor arrays. That must have been a faulty batch or something bad happened to the card (maybe a lot of heat). Great that you were able to fix it!
@@bitsundbolts I don't know what could have led to such consequences. One day, the card suddenly stopped being detected in the system. And MOJO reported that TMU had about 150MB of memory and FBI had about 1.5GB. I regret that I didn't take a screenshot then.
Thanks Alex great video another card saved from the voodoo master! Very informal on fault finding and repairing crazy to think something so small can make the card faulty
Why would it be crazy? If you're not getting information, power, or any other kind of signal, of course that small thing is to blame. A cheap seal brought down Challenger, after all.
Yes, hehe. Also, maybe a bit unconventional to debug this card, but I have to fight with the knowledge I have. And more often than not, those cards fail due to physical damage. Well, I guess I was lucky here!
@@bitsundbolts I've repaired a few cheap 16 bit I/O cards that were thrown into a bin too. Predating SMI stuff, so it's easy to see where a trace is broken or a resistor cracked. Once it was the leg of an IC that was broken under the PGA. That was fun to fix. Good thing I have a magnifying station too:D
Brilliant work, I will use this approach when logically diagnosing problems. If you see signs of damage - look in line of damage for more. Seems obvious, but it's so easy to overlook the hidden failures in the sea of components. Thank you immensely.
Your troubleshooting skills persuaded me to sub to your channel....I owned this particular flavor 3dfx card back in the day. It's so easy to knock off those small surface mount components from a PCB board with a bit of uncareful handling. Bravo!
Had a windows XP laptop once that kept shutting off randomly. Took it apart down to the board and looked for hours trying to find an issue with it. Finally i accidentally found a tiny little crack in the solder joint of the main input power resistor and resoldered the connections on both side of the resistor. Put the laptop back together and not once did it shut off after that. That laptop had never been taken apart so I'm convinced to this day it was just a factor defect connection. When i told the guy i got from what i ended up doing to fix it he freaked out but was also happy to see it fixed.
Awesome fix! I just fixed an STB Voodoo 2 that threw the same error in Windows. The 2nd resistor from the left on RN18 on my card was damaged and I was able the get the card working again with a bodge wire from the pad to the top of the resistor network. It also needed a pin replaced on one of the TMUs, but that was a lot easier than finding and fixing the resistor network.
When you were Ohming it, I thought I saw it move, but I wasn't sure if it was the microscope, UA-cam being crazy or whatever. But when it snapped during soldering, it erased ALL doubts that's for sure.
Case solved. Great job, detective! I have 2 dead Voodoo2 cards killed probably by the non-standard motherboard slot voltage. They both started, showed 1 second of rendering and died. Then I bought another motherboard, third Voodoo2 and got a perfect retro-PC.
I'm sorry to hear about those two Voodoo cards. I am curious what could be the problem. Have you tried to debug them? You could use a tool called Mojo to see if the cards are detected at all. Maybe they're salvageable - or be used as donor cards.
It is definitely something I came across more than once. So, yes, it makes sense to spend the five minutes to check a damaged area for something that isn't obvious to the eye!
It is so wild to me coming across videos of people resurrecting the hardware I grew up with. My first major gpu purchase was this exact card. Actually it is sitting in its box after perhaps 6 months of use on my shelf now. I played A LOT of warcraft on this thing.
Yes indeed. That solder joint looked a bit off. That is why i scratched it off a bit - and if you look closely, you see that edge of the resistor array move. I just didn't see that while working on the card - only while cutting the video material it became obvious.
@ props to your sir. Watching your videos inspired me to do my own repairs. I have yet to fix a board, but then again most of the boards I have are socket 775 and above. Your methods of visual inspection and fault finding have been helpful to isolate where the problems are on my boards though. Now I just need to learn how to get down to root causes, ie, the faulty components. I haven’t been lucky enough to have broken traces or busted components, but I will keep digging until I figure things out and hopefully fix some boards.
Interesting video. Similar thing: I had a neighbour who bought a car as a non-runner. For 3 months he tried everything to get it to work without success. Traced it down to a faulty ECU. A replacement was only available from a scrap yard for £300, and had no guarantee of it working. I asked him if I could have a look at the inside of the ECU to see if I could spot anything. I took it home openned it up and spotted a burn patch on the board under a burnt out resistor. Had a rummage through my components and pulled out a set of parts that would match. Replaced the parts and gave it back to him, all in under 20 minutes. Told him if it didn't work then he's lost nothing. He put the ECU back in the car and it started first time. A 2 pence component stopped the whole car from working.
Very nice! Yes, those stories are very common! Just a simple component prevents an entire system from working! Great job of attempting a repair and succeeding! I bet your neighbor was very happy - and probably also a bit surprised?
My first 3d accelerator was a Voodoo3 3000 AGP, i LOVED it.. Especially when i got Unreal, and it made such a difference to the game! Still sad about 3dfx going to the grave, but glad nVidia kept their legacy alive :)
You pointed out how some of the components seemed to be at odd angles, like they might have been hand-placed. The reflowing of solder often causes strong surface tension movements. If the pads of the components were not properly designed for size and type of solder to be used, when the solder wets, it can cause the components to move around. The worst case of this is something called Tombstoning, where a cap/res will actually stand up on one end from the surface tension. Otherwise, an excellent video.
I had one of those components once, it was rotated 90 degrees. It still made the right connection, but it was higher than the rest of the SMDs because of the rotation. Somebody mentioned the term Tombstoning at that time as well.
When you found the scratched traces, couldn't you use a multimeter to check if the traces were damaged? Also, at 20:55 a chip is shown "V53C16258HK25 100MHz" with an upside-down "V"-shaped scratch and slight damage to the lower-right edge of the chip body. Would that be worth checking out a bit more? (It's also shown at 21:31 .)
Yes, I could have used the multimeter. I usually remove the solder mask at deep scratches to see if there is an issue, then reinforce the trace with solder (that usually also will remove any moisture that may have accumulated), and finally, I'll reapply fresh solder mask to seal the traces again. The V-shape scratch on the memory is definitely a damage worth investigating - and I probably would have if there wasn't the scratch in the PCB. But I have to say that a scratch on top of a memory chip is less likely to lead to a problem since it's further away from the PCB and the SMD components. Of course, this doesn't mean that you should never look around such damages. It still can be a valuable hint that points you to the real issue.
Its luck anyway the next step was use volt check in the memory and a osciloscope on the chips signaks missing a pain in the ass tests soo it,s luck find in visual inpection the fault, no all times you spot bad resitors on ceramic caps it's more easy if shorted or no decopling is working.
0:48 There's a spelling error there in the Witchery instructions; where it says 'type "witchery" end press Enter' it should say 'type "witchery" _und_ press Enter.'
I'm not sure I agree that this was luck, I think it is very much a result of your methodical and thorough approach. At best a case of "das Glück des Tüchtigen" (luck of the diligent). 😉 Either way, great work as always! 👏 Really looking forward to that 386 board that is so similar to my first own PC! ♥ Make sure to kick some Mecha-Hitler butt on it once you inevitably fix it. It is also a perfectly suitable machine for Defender of Boston by the way...wink wink nudge nudge. 😄
Thanks! I guess I did draw from experience which made me find that broken resistor array quickly. The 386 video is done! It'll be the one for next Thursday - but it ended up being over an hour long 🤣 - an indicator of how much work it was!
No, orientation of resistors does not matter. Even the ceramic capacitor I add on the back of the board is not polarized (does not need a specific orientation). There are, however, other components where orientation is mandatory, but they usually have some sort of a marking to identify positive or negative terminal, or pin number 1.
Good luck! I hope you'll find the issue. I had a Voodoo 1 with a broken resistor on the back which also behaved similarly - yellow exclamation mark in the Windows device manager.
I was lucky to find that one. If I had looked closer through the microscope, I might have seen the edge of the resistor array move slightly when I touched it with the multimeter probes. I noticed that only while editing the video - at that time, I knew what the issue was.
This was awesome! Back in the day my first self built PC was a P2 350mhz 32MB RAM with a Diamond 3DFX Voodoo 2 8MB or maybe 12MB card. Came with Mechawarrior 2 31st Century game designed to be played on that card. I never was able to finish playing that game all the way through and now don't have the 3DFX card. I'd buy one on ebay but don't trust the sellers enough to pay the high prices they're asking. I miss that PC something fierce.
@@bitsundbolts Keep up the great work! Component level repairs are not a possibility for me and my shaking hands. But it's awesome seeing folks keeping older hardware alive so they can still be used.
One of my first bigger projects was fixin a real old oscilloscope (a mix of PCBs and ceramic strips). The manual advised to "just look around and inspect things" before even measuring. I wasted quite a long time because I did not do that...
Most of my successful repairs are due to a thorough visual inspection. I wish I had more experience debugging electric circuits using a multimeter and oscilloscope though. Interesting that this old manual suggests the visual inspection as well - someone knew how most electronics fail 😉
80n is fine for an aging 100n ceramic capacitor, it likely is now measuring 100n plus after it has been reflowed, and the dielelctric has been relaxed from aging.
I had this 3D accelerator, I cant remember what graphics card I had with it though, I do remember how excited I was when I bought this card though, I also bought a Star Wars game with it and had previously upgraded my memory to a 64Mb card to my Pentium II based WIn 98 PC and a week later added another 64Mb card. this PC eventually ran my MAME cabinet.
Amazing that so many components are that crooked! Must have been the times that every card produced was sold instantly. Also is that C116 cap cracked, or is it just a hair or something? Around 22:30.
Can't wait for a public release of witchery, and also V1 support. I have a broken V2 just like the one in the video plus one Orchid Righteous V1 in need of troubleshooting.
I wish I could test Witchery on one of my precious Voodoo 2 boards. I replaced the FBI, I replaced the 4 x FBI memory ICs, checked all resistors, checked all traces... and yet I still get random artefacts that are indicative of an FBI memory fault. 😞
What about the other 4 memory ICs connected to the FBI chip? There should be 8 chips, right? Or is this one of those Voodoo 2 with 1MB memory chips? Unfortunately, I am not sure when Maksim is going to release Witchery - it is still under development.
@@bitsundbolts, long story, but I managed to identify the 4 x memory ICs that were creating the problem (2 on the back and 2 on the front) and I only replaced those four. The card worked perfectly with the other remaining four ICs and the "SSTV2_FBI_MEMSIZE=2" variable (even after I physically removed the ones that I narrowed down as being the source of the problem). So I'm confident that the other four are absolutely OK... Even stranger is the fact that only specific games are affected by the artifacts (GLQuake, Quake 2 and some Direct3D titles). Unreal, on the other hand, works flawlessly, so my guess is that only a few specific memory bits are affected. I've repaired countless other Voodoo 2 cards in the meantime, but this one... it's been bugging me for more than a year now, I've never seen such a strange behaviour. At one point, I was sure that the problem was the FBI itself, I replaced it with a new one (I have a few spare NOS FBI chips), and I couldn't believe it when, after replacing the chip, I saw that the same problem persisted. At this point it will be absolutely incredible if I ever manage to find the problem. 😅
@@bitsundbolts, long story, but I managed to identify the 4 x memory ICs that were creating the problem (2 on the back and 2 on the front) and I only replaced those four. The card worked perfectly with the other remaining four ICs and the FBI memsize variable (even after I physically removed the ones that I narrowed down as being the source of the problem). So I'm confident that the other four are absolutely OK... Even stranger is the fact that only specific games are affected by the artifacts (GLQuake, Quake 2 and some Direct3D titles). Unreal, on the other hand, works flawlessly, so my guess is that only a few specific memory bits are affected. I've repaired countless other Voodoo 2 cards in the meantime, but this one... it's been bugging me for more than a year now, I've never seen such a strange behaviour. At one point, I was sure that the problem was the FBI itself, I replaced it with a new one (I have a few spare NOS FBI chips), and I couldn't believe it when, after replacing the chip, I saw that the same problem persisted. At this point it will be absolutely incredible if I ever manage to find the problem.
@@bitsundbolts, it's a long story, but I managed to identify the 4 x memory ICs that were creating the problem - 2 on the back and 2 on the front - and I only replaced those four. The card worked perfectly with the other remaining four ICs and the FBI memsize variable (even after I physically removed the ones that I narrowed down as being the source of the problem). So I'm confident that the other four are absolutely OK... Oh, and even stranger is the fact that only specific games are affected by the artifacts (GLQuake, Quake 2 and some Direct3D titles). Unreal, on the other hand, works flawlessly, so my guess is that only a few specific memory addresses are affected. I've repaired countless other Voodoo 2 cards in the meantime, but this one... it's been bugging me for more than a year now, I've never seen such a strange behaviour. At one point, I was sure that the problem was the FBI itself, I replaced it with a new one (I have a few spare NOS FBI chips), and I couldn't believe it when, after replacing the chip, I saw that the same problem persisted. At this point it will be absolutely incredible if I ever manage to find the problem.
@@bitsundbolts, it's a long story, but I managed to identify the 4 x memory ICs that were creating the problem - 2 on the back and 2 on the front - and I only replaced those four. The card worked perfectly with the other remaining four ICs and the FBI memsize variable (even after I physically removed the ones that I narrowed down as being the source of the problem). So I'm confident that the other four are absolutely OK...
Nice work.. I just picked up a two V2 12mb cards. I saw them working on a video. But scared to test them after I saw the box it was shipped in.. they look okay. Got them for my collection , not to use..
dont know if it mattered but spotted that dead resistor array, may of been manually replaced, compared to the repair n the others next to it, it was upside down.
I think those resistor arrays come from a roll for pick-and-place machines. There is no guarantee that they all face the same direction when packaged in those rolls (or wheels). I believe this was factory made. The crack was caused by whatever scratched over the PCB and maybe also hit the 3Dfx chip.
@@znk0rfake fingers weird, misinformation weird, or being the tool of capital to harvest free labor from unprofited workers thereby reproducing inherent flaws within capital markets weird? AI does all three, so that’s gonna be hard to beat!
What good luck to stumble across the problem so quickly! Was there a bit of fuzzy corrosion around some things or does the microscope just really highlight normal 25 year old dust?
There was a layer of dust over the card. I cleaned it after I was done with the video - now it looks clean again. And yes, I was very lucky to find the issue that quickly - if I ever would have found it. Nevertheless, I learned something. A broken resistor that can't can't be identified as such, can be the reason for a faulty card. So, if there isn't anything obvious, we have to check the resistors!
Hi, i have a Quadro FX4500 with no display and a damaged component labelled L514 (mentioned on Vogons) which i guess is an inductor? But how do i tell value (no markings)? I cant see any other obvious flaws
Unfortunately, I don't know how to get the value of this inductor. If I had this issue, then I would try to find a similar looking component on another device (an inductor of course) and see if the card changes its behavior. An inductor will measure like a bridge - continuity. It resists current flow changes if I understand this device correctly.
The resistor you de-soldered was upside down. At least, according to the letters on it. Was this the case on your "good" board as well? Or did someone already try to repair this card earlier?
@@roelbrook7559 30 years ago computer vision lacked processing power to bother with resistor orientation. It was probably upside down in the tape and passed the automated tests.
It is very common to have resistor arrays placed in different orientations. A resistor can be placed either way. Those components most likely come from a wheel with thousands of resistor arrays. A pick-and-place machine is then populating PCBs. It all depends on the orientation the resistor/array/network had in that wheel. I believe this resistor was added by the factory and just got hit in that corner and cracked.
Another epic fix! Such great detective work! You're spot on. When there is any damage and the thing isn't working right, extra scrutiny must be given in the general vicinity of the damage.
Thank you Adrian! Yes! I have seen it multiple times! A damaged PCB can give a lot of hints where to look for more damage.
Your guy's repair videos is like after work emotional therapy for me. Haha
An interesting project
A while back in the retro web the talked about a project to put a pentium o pentiumoverdrive in a 386.
I could be interesting to bring that here
A perfect demonstration of fault finding and where to look, fractures in resistors ...I learned something new here, Thank You. Great result ...awesome video.
Thank you!
Just recently I finished the restoration of very damaged 386 motherboard. Battery ravaged and some crushed components.
The Last fix was a resistor arraya with an almost invissible crack.
I tried to fix the array common line and Just use one resistor for the cracked one, to no avail, so I used an small array. Later relpaced the whole arrayanes. But for one crushed, I madera my one with discrete resistos as I dont have a replacement
My first 3D accelerator was a 3Dfx Voodoo 2, so I have a weakness for these cards, and seeing one restored to its glory warms my heart.
I loved my Voodoo 2! It felt like it was billions of years ahead of anything else at the time.
Same here.
I still have mine around somewhere in a box
Really cool video! Oh the memories with Voodoo 2. And that Unreal engine castle flyby... Oh boy. 💖
Inspired by your videos, I bought a 3Dfx Voodoo 1 on eBay a few months ago. Like so many of your video examples, some of the 3Dfx chips had bent pins. I have a very modest wifi microscope setup, but I was able to straighten & reflow them. I have been able to get it to work a couple of times, but unfortunately the card still doesn't work reliably. Perhaps it's time to look for further damage on other components to see if perhaps it has a cracked resistor on it somewhere too. Thanks for your videos. Without them, I wouldn't even be attempting these repairs!
I am happy to hear that! Congratulations on your attempt! Fixing pins on 3Dfx chips isn't easy - even on Voodoo 1 cards! Even though the card doesn't work reliably, you are a big step closer to a fully working card! Let me give you a few pointers: Is there any pattern you notice when the card is unreliable? For instance, when the card is cold or hot? I would pay close attention to the solder joints between 3Dfx chip legs and the pads on the PCB. Even if they look to be fine, there could be a poor electrical connection. If you carefully poke those legs and they move (even a tiny bit), I would reflow the entire chip. Next, as you said, check for cracked components. Any scratch on the PCB might give you hints where to look. I hope this helps. Let me know if you can find something or if you need further help!
voodoo 2 sli was a monster setup, especially for quake 1/2 tournament rigs
Great fix. You got me inspired, now I'll have to pull my SLI Voodoo 2 setup from the mothballs and see if Win98/ME is willing to play ball ;) .
Good luck! I hope your setup still works! 😃
@@bitsundbolts Well. I need to source a floppy drive or a floppy drive emulator :( , the Win98 CDROM isn't bootable😂 (funny enough, I do have a Win98 SE original boot floppy). Furthermore, I need to dig the archive and find an IDE HDD or be patient as an IDE to mSATA converter on it's way.
Setup:
Abit BH6
Intel Pentium III @ 750 mhz (originally a P3-1000, but a 133 fsb model)
512 PC-133 SDRAM
Diamond Multimedia (Nvidia) V770
Diamond Multimedia Voodoo 2 12 MB in SLI
Diamond Multimedia Monster Sound MX300
3Com 3C905x
Apart from the SLI Voodoo 2, which was in use in a previous P2-450 system, the setup was put together between October~December 2000.
The FDD and IDE drive that I had put into the storage got wet due to a leak in our garage roof last winter. This went unnoticed until June of this year when I was doing a check. Bummer.
I think I'll pay our local recycle shop a visit, perhaps the have a fossilized FDD on stock.🤣
Update:
Well it appeared the problem is in the setup. After setting the IDE DVD drive as master on the second channel it will actually boot the Win98 CD.
Incidentally, I found a 200 GB external drive (Western Digital), so the question is: to shuck or not to shuck ;).
The SLI Voodoo 2 is 'recognized' at bootstrap as they appear as multimedia devices.
@@bitsundbolts
Currently putting the setup together. The system used to be my main, later on went to my mom when I upgraded to a P4.
Setup:
Abit BH6
Intel P III-1000 @ 750 (133 fsb model, socket 370 in an Abit Slotket !!!)
512 MB SDRAM PC-133
Diamond Viper V770 (replaced an Intel i740)
Diamond Multimedia Voodoo 2 12 mb SLI
Diamond Multimedia Monster Sound MX300
3Com 3C905x NIC
FDD: no
NEC DVD-reader/writer
HDD: Either a WDAC 200 GB IDE drive (will require shucking as it is an external drive) or a 128 GB mSATA drive (IDE to mSATA adapter on the way).
Initial problem with booting from CD-rom were mitigated, setting it as master on the second channel and giving the Win98 CD a polish did it.
For now the SLI Voodoo 2's are detected at bootstrap, so looking good. Once the storage has been solved, real world testing can start.
Very satisfying fix!
Brilliant, just came home early from a shit day at work in near freezing temps. Sat down with a cup of tea, watched you tinker and the moment you landed on the damaged array and we could see it cracked real time made me laugh.
You took me from down in the dumps to happy as can be in no time, I mean what was the chance of you landing on it like that, even with the visual trace damage. Epic repair all round mate, thank you for the upload.
I am sorry to hear that your day wasn't great! But I am happy that you enjoyed the video and that it made you laugh :) Thanks for watching and hope your day picks up from here!
Hello! Thank's for good content.
My brother had the same problem with his card. After much searching, I found that about 15 resistor arrays had lower resistance than expected. After replacing all the faulty resistor arrays, the card worked!
Congratulations on that repair! Interesting that the resistance reduced on so many resistor arrays. That must have been a faulty batch or something bad happened to the card (maybe a lot of heat). Great that you were able to fix it!
@@bitsundbolts I don't know what could have led to such consequences. One day, the card suddenly stopped being detected in the system. And MOJO reported that TMU had about 150MB of memory and FBI had about 1.5GB. I regret that I didn't take a screenshot then.
Great diagnosis and repair! Love seeing hardware being brought back to life.
I love the thumbnail. I really enjoy these types of repair video. It just goes to show that attention to detail is key.
Thank you!
Finally! One more 3Dfx Wonder...now Witchery! Thank you so much!
Great Tech !!! Amazing Skill&Work done !!!
Thanks!
Thanks Alex great video another card saved from the voodoo master! Very informal on fault finding and repairing crazy to think something so small can make the card faulty
Why would it be crazy? If you're not getting information, power, or any other kind of signal, of course that small thing is to blame. A cheap seal brought down Challenger, after all.
Yes, hehe. Also, maybe a bit unconventional to debug this card, but I have to fight with the knowledge I have. And more often than not, those cards fail due to physical damage. Well, I guess I was lucky here!
@@bitsundbolts I've repaired a few cheap 16 bit I/O cards that were thrown into a bin too. Predating SMI stuff, so it's easy to see where a trace is broken or a resistor cracked. Once it was the leg of an IC that was broken under the PGA. That was fun to fix. Good thing I have a magnifying station too:D
awesome dude... this was my frist gfx card i bought !!!. was cool to see it again, and awesome you found the problem :) GJ
Thanks! 👍
Love your videos. Keep bringing 3DFX's back to life!
Oh, I have two more Voodoo 2 cards that are probably on a different level to repair! Unpacking them today to assess the situation :)
@bitsundbolts Awesome! I can't wait!!!
No idea how I stumbled upon this and your channel but damn that was interesting, thank you!
Glad to hear that! Thanks 😊
Thanks for sharing your knowledge
Congratulations !! God job under the microscope.
Awating for the recover corrosión Project.
Brilliant work, I will use this approach when logically diagnosing problems. If you see signs of damage - look in line of damage for more. Seems obvious, but it's so easy to overlook the hidden failures in the sea of components. Thank you immensely.
You're welcome! I had this happen to me several times until I found that pattern. A deep scratch in the PCB means potentially damaged SMD components.
the joy we had back then having one of these GPU's was much more than what we have now if we have a RTX4090
Your troubleshooting skills persuaded me to sub to your channel....I owned this particular flavor 3dfx card back in the day. It's so easy to knock off those small surface mount components from a PCB board with a bit of uncareful handling. Bravo!
Agreed! It's sad that such faults may be the reason for such cards to be tossed out, even though there's nothing wrong but a small SMD resistor.
nice work man,voodoo 2 was my first 3d card back in the day.,good memories
Mesmerising watching this 😊
Thank you Phil! 🙂
I remember booting up Doom 2 when I got this bad boy. Took me all summer to save up for it. Everything was so smooth and round not a pixel to be seen.
Had a windows XP laptop once that kept shutting off randomly. Took it apart down to the board and looked for hours trying to find an issue with it. Finally i accidentally found a tiny little crack in the solder joint of the main input power resistor and resoldered the connections on both side of the resistor. Put the laptop back together and not once did it shut off after that. That laptop had never been taken apart so I'm convinced to this day it was just a factor defect connection. When i told the guy i got from what i ended up doing to fix it he freaked out but was also happy to see it fixed.
Those are the great successes that stay with us! Great fix on that laptop!
Very nice job... You are so meticulous and precise. Thank you for sharing your knowledge.
3dfx repair videos are the best of your channel
I have more coming 😃 don't worry...
Awesome fix! I just fixed an STB Voodoo 2 that threw the same error in Windows. The 2nd resistor from the left on RN18 on my card was damaged and I was able the get the card working again with a bodge wire from the pad to the top of the resistor network. It also needed a pin replaced on one of the TMUs, but that was a lot easier than finding and fixing the resistor network.
Congratulations on the fix! Always good to hear when one of those cards was successfully repaired!
When you were Ohming it, I thought I saw it move, but I wasn't sure if it was the microscope, UA-cam being crazy or whatever.
But when it snapped during soldering, it erased ALL doubts that's for sure.
Yes - you are right. It does move very slightly! I only noticed when editing the video and after I found out that this corner was broken.
Case solved. Great job, detective!
I have 2 dead Voodoo2 cards killed probably by the non-standard motherboard slot voltage. They both started, showed 1 second of rendering and died. Then I bought another motherboard, third Voodoo2 and got a perfect retro-PC.
I'm sorry to hear about those two Voodoo cards. I am curious what could be the problem. Have you tried to debug them? You could use a tool called Mojo to see if the cards are detected at all. Maybe they're salvageable - or be used as donor cards.
Simple but also not simple! Well done!
Thank you Tony!
The FCC ID points to the card being manufactured by Joytech Computer Co, Ltd.
Ah, interesting! Never heard of that company though.
Good name for a "personal entertainment" company
Great find! This is a good reminder that you should not be lazy and check every part thoroughly which you did. Congrats!
It is definitely something I came across more than once. So, yes, it makes sense to spend the five minutes to check a damaged area for something that isn't obvious to the eye!
It is so wild to me coming across videos of people resurrecting the hardware I grew up with. My first major gpu purchase was this exact card. Actually it is sitting in its box after perhaps 6 months of use on my shelf now. I played A LOT of warcraft on this thing.
It is nice to refresh those memories! Most of my viewers grew up during that time.
u say luck i say that was pure intuition. Great eye my friend. but I suspect you noticed on some level that it looked off.
Yes indeed. That solder joint looked a bit off. That is why i scratched it off a bit - and if you look closely, you see that edge of the resistor array move. I just didn't see that while working on the card - only while cutting the video material it became obvious.
Find the scratch, find the problem! You make it seem so easy that I'm starting to believe I can also fix everything 😉
Haha, of course you can! The more you fix those things, the easier it gets! But this one was lucky - I fully admit that 😉
i had one of those years ago! good fix!
Excellent work! 👍
Thank you!
@ props to your sir. Watching your videos inspired me to do my own repairs. I have yet to fix a board, but then again most of the boards I have are socket 775 and above. Your methods of visual inspection and fault finding have been helpful to isolate where the problems are on my boards though. Now I just need to learn how to get down to root causes, ie, the faulty components. I haven’t been lucky enough to have broken traces or busted components, but I will keep digging until I figure things out and hopefully fix some boards.
Interesting video.
Similar thing: I had a neighbour who bought a car as a non-runner. For 3 months he tried everything to get it to work without success. Traced it down to a faulty ECU. A replacement was only available from a scrap yard for £300, and had no guarantee of it working. I asked him if I could have a look at the inside of the ECU to see if I could spot anything. I took it home openned it up and spotted a burn patch on the board under a burnt out resistor. Had a rummage through my components and pulled out a set of parts that would match. Replaced the parts and gave it back to him, all in under 20 minutes. Told him if it didn't work then he's lost nothing. He put the ECU back in the car and it started first time. A 2 pence component stopped the whole car from working.
Very nice! Yes, those stories are very common! Just a simple component prevents an entire system from working! Great job of attempting a repair and succeeding! I bet your neighbor was very happy - and probably also a bit surprised?
Good job on this one.
On the other hand, your microscope is incredible 🤯
My first 3d accelerator was a Voodoo3 3000 AGP, i LOVED it.. Especially when i got Unreal, and it made such a difference to the game! Still sad about 3dfx going to the grave, but glad nVidia kept their legacy alive :)
Nice job. You have good instincts!
Another great fix Alex. You are becoming a Voodoo card whisperer 😀
Hehe, thanks...
If you draw a line between the scratch on the board and the dent on the chip does it not cross the resistor right where it broke?
Well, it may not be 100% crossing, but it is close enough to be suspicious!
Your microscope footage is always excellent quality
You pointed out how some of the components seemed to be at odd angles, like they might have been hand-placed. The reflowing of solder often causes strong surface tension movements. If the pads of the components were not properly designed for size and type of solder to be used, when the solder wets, it can cause the components to move around. The worst case of this is something called Tombstoning, where a cap/res will actually stand up on one end from the surface tension.
Otherwise, an excellent video.
I had one of those components once, it was rotated 90 degrees. It still made the right connection, but it was higher than the rest of the SMDs because of the rotation. Somebody mentioned the term Tombstoning at that time as well.
lol, @ 7:57 "tmu bruce is busy", he aint got no time for these antics i guess. good work on the resuscitation!
perfect job! :) congratulations :)
Thanks!
When you found the scratched traces, couldn't you use a multimeter to check if the traces were damaged?
Also, at 20:55 a chip is shown "V53C16258HK25 100MHz" with an upside-down "V"-shaped scratch and slight damage to the lower-right edge of the chip body. Would that be worth checking out a bit more? (It's also shown at 21:31 .)
Yes, I could have used the multimeter. I usually remove the solder mask at deep scratches to see if there is an issue, then reinforce the trace with solder (that usually also will remove any moisture that may have accumulated), and finally, I'll reapply fresh solder mask to seal the traces again.
The V-shape scratch on the memory is definitely a damage worth investigating - and I probably would have if there wasn't the scratch in the PCB.
But I have to say that a scratch on top of a memory chip is less likely to lead to a problem since it's further away from the PCB and the SMD components. Of course, this doesn't mean that you should never look around such damages. It still can be a valuable hint that points you to the real issue.
Definitely skill, not luck. Good work!
Its luck anyway the next step was use volt check in the memory and a osciloscope on the chips signaks missing a pain in the ass tests soo it,s luck find in visual inpection the fault, no all times you spot bad resitors on ceramic caps it's more easy if shorted or no decopling is working.
New fan here, subscribed! 😁
Thanks!! 🙏
You deserve much more subscribers
Haha, thank you! Well, I'm working on it. It's a slow grind!
0:48 There's a spelling error there in the Witchery instructions; where it says 'type "witchery" end press Enter' it should say 'type "witchery" _und_ press Enter.'
Well done super neat soldering
Thanks 👍
I'm not sure I agree that this was luck, I think it is very much a result of your methodical and thorough approach. At best a case of "das Glück des Tüchtigen" (luck of the diligent). 😉 Either way, great work as always! 👏
Really looking forward to that 386 board that is so similar to my first own PC! ♥ Make sure to kick some Mecha-Hitler butt on it once you inevitably fix it. It is also a perfectly suitable machine for Defender of Boston by the way...wink wink nudge nudge. 😄
Thanks! I guess I did draw from experience which made me find that broken resistor array quickly.
The 386 video is done! It'll be the one for next Thursday - but it ended up being over an hour long 🤣 - an indicator of how much work it was!
Good find 😎👍
Does orientation matter with resistors? it was upside down when you removed it but orientated up afterwards...
no, it doesn't.
@ thank you.
No, orientation of resistors does not matter. Even the ceramic capacitor I add on the back of the board is not polarized (does not need a specific orientation). There are, however, other components where orientation is mandatory, but they usually have some sort of a marking to identify positive or negative terminal, or pin number 1.
Great work!
Thanks!
Wonderful, I have a VOODOO 1 with the same problem - I thought the drivers were not suitable. Now I know where to look for the problem....
Good luck! I hope you'll find the issue. I had a Voodoo 1 with a broken resistor on the back which also behaved similarly - yellow exclamation mark in the Windows device manager.
nice find! I'm not sure that would have ever occurred to me.
I was lucky to find that one. If I had looked closer through the microscope, I might have seen the edge of the resistor array move slightly when I touched it with the multimeter probes. I noticed that only while editing the video - at that time, I knew what the issue was.
The video card is a Joytech Apollo 3D FAST 2 PCI as per the FCC
Awesome video!
Thanks!
Your thumbnail made me click, your content made me stay
Ah nice! I usually get trashed for using AI generated thumbnails :)
This was awesome! Back in the day my first self built PC was a P2 350mhz 32MB RAM with a Diamond 3DFX Voodoo 2 8MB or maybe 12MB card. Came with Mechawarrior 2 31st Century game designed to be played on that card. I never was able to finish playing that game all the way through and now don't have the 3DFX card. I'd buy one on ebay but don't trust the sellers enough to pay the high prices they're asking. I miss that PC something fierce.
Yeah, prices are very high. I did buy cards from ebay, but none of those "untested" ones. Even those go for a lot of money these days.
@@bitsundbolts Keep up the great work! Component level repairs are not a possibility for me and my shaking hands. But it's awesome seeing folks keeping older hardware alive so they can still be used.
Great job!
Thanks!
One of my first bigger projects was fixin a real old oscilloscope (a mix of PCBs and ceramic strips). The manual advised to "just look around and inspect things" before even measuring. I wasted quite a long time because I did not do that...
Most of my successful repairs are due to a thorough visual inspection.
I wish I had more experience debugging electric circuits using a multimeter and oscilloscope though.
Interesting that this old manual suggests the visual inspection as well - someone knew how most electronics fail 😉
awesome 💪💪💪
An excellent fix...😊
Thank you!
When I had my PC upgraded in 1998 it had a 3dfx voodoo card put in, and the repair shop left this icon on the desktop - Turok dinosaur hunter!
Skill + Luck = Retro Tech Success 🙂
80n is fine for an aging 100n ceramic capacitor, it likely is now measuring 100n plus after it has been reflowed, and the dielelctric has been relaxed from aging.
Good job!👍
I had this 3D accelerator, I cant remember what graphics card I had with it though, I do remember how excited I was when I bought this card though, I also bought a Star Wars game with it and had previously upgraded my memory to a 64Mb card to my Pentium II based WIn 98 PC and a week later added another 64Mb card. this PC eventually ran my MAME cabinet.
Nice looking voodoo2 card, i had 1 simular, tink mine lacked some ram chips tho
Amazing that so many components are that crooked! Must have been the times that every card produced was sold instantly.
Also is that C116 cap cracked, or is it just a hair or something? Around 22:30.
It looks like it, but I think it is only left-overs from cleaning or wiping off the card.
This guy is awesome
Can't wait for a public release of witchery, and also V1 support. I have a broken V2 just like the one in the video plus one Orchid Righteous V1 in need of troubleshooting.
I hope it will be released soon! Voodoo 1 Support is only beginning to be developed.
I wish I could test Witchery on one of my precious Voodoo 2 boards. I replaced the FBI, I replaced the 4 x FBI memory ICs, checked all resistors, checked all traces... and yet I still get random artefacts that are indicative of an FBI memory fault. 😞
What about the other 4 memory ICs connected to the FBI chip? There should be 8 chips, right? Or is this one of those Voodoo 2 with 1MB memory chips? Unfortunately, I am not sure when Maksim is going to release Witchery - it is still under development.
@@bitsundbolts, long story, but I managed to identify the 4 x memory ICs that were creating the problem (2 on the back and 2 on the front) and I only replaced those four. The card worked perfectly with the other remaining four ICs and the "SSTV2_FBI_MEMSIZE=2" variable (even after I physically removed the ones that I narrowed down as being the source of the problem). So I'm confident that the other four are absolutely OK...
Even stranger is the fact that only specific games are affected by the artifacts (GLQuake, Quake 2 and some Direct3D titles). Unreal, on the other hand, works flawlessly, so my guess is that only a few specific memory bits are affected. I've repaired countless other Voodoo 2 cards in the meantime, but this one... it's been bugging me for more than a year now, I've never seen such a strange behaviour. At one point, I was sure that the problem was the FBI itself, I replaced it with a new one (I have a few spare NOS FBI chips), and I couldn't believe it when, after replacing the chip, I saw that the same problem persisted. At this point it will be absolutely incredible if I ever manage to find the problem. 😅
@@bitsundbolts, long story, but I managed to identify the 4 x memory ICs that were creating the problem (2 on the back and 2 on the front) and I only replaced those four. The card worked perfectly with the other remaining four ICs and the FBI memsize variable (even after I physically removed the ones that I narrowed down as being the source of the problem). So I'm confident that the other four are absolutely OK...
Even stranger is the fact that only specific games are affected by the artifacts (GLQuake, Quake 2 and some Direct3D titles). Unreal, on the other hand, works flawlessly, so my guess is that only a few specific memory bits are affected. I've repaired countless other Voodoo 2 cards in the meantime, but this one... it's been bugging me for more than a year now, I've never seen such a strange behaviour. At one point, I was sure that the problem was the FBI itself, I replaced it with a new one (I have a few spare NOS FBI chips), and I couldn't believe it when, after replacing the chip, I saw that the same problem persisted. At this point it will be absolutely incredible if I ever manage to find the problem.
@@bitsundbolts, it's a long story, but I managed to identify the 4 x memory ICs that were creating the problem - 2 on the back and 2 on the front - and I only replaced those four. The card worked perfectly with the other remaining four ICs and the FBI memsize variable (even after I physically removed the ones that I narrowed down as being the source of the problem). So I'm confident that the other four are absolutely OK...
Oh, and even stranger is the fact that only specific games are affected by the artifacts (GLQuake, Quake 2 and some Direct3D titles). Unreal, on the other hand, works flawlessly, so my guess is that only a few specific memory addresses are affected. I've repaired countless other Voodoo 2 cards in the meantime, but this one... it's been bugging me for more than a year now, I've never seen such a strange behaviour. At one point, I was sure that the problem was the FBI itself, I replaced it with a new one (I have a few spare NOS FBI chips), and I couldn't believe it when, after replacing the chip, I saw that the same problem persisted. At this point it will be absolutely incredible if I ever manage to find the problem.
@@bitsundbolts, it's a long story, but I managed to identify the 4 x memory ICs that were creating the problem - 2 on the back and 2 on the front - and I only replaced those four. The card worked perfectly with the other remaining four ICs and the FBI memsize variable (even after I physically removed the ones that I narrowed down as being the source of the problem). So I'm confident that the other four are absolutely OK...
loved Voodoo cards had several
Nice work
Nice work.. I just picked up a two V2 12mb cards. I saw them working on a video. But scared to test them after I saw the box it was shipped in.. they look okay. Got them for my collection , not to use..
I do not know if this is coincidence, but resistor arrays also often fail on the DVD drive boards of the first Xbox
But those are usually in a case - there might be a different reason for why those resistor arrays fail in the first Xbox - maybe heat?
dont know if it mattered but spotted that dead resistor array, may of been manually replaced, compared to the repair n the others next to it, it was upside down.
I think those resistor arrays come from a roll for pick-and-place machines. There is no guarantee that they all face the same direction when packaged in those rolls (or wheels). I believe this was factory made. The crack was caused by whatever scratched over the PCB and maybe also hit the 3Dfx chip.
That card looks like it was produced on a Friday @ 4:55 PM before a 4 day holiday weekend, and the PnP machine was down so it was handsoldered.
Haha, true!
Why didnt you check new resistors array before soldering ?
I guess I just assumed that it was working.
Think i still got my voodoo 2 and 3 somewhere
really dont like these ai pics in the thumbnail
I'm sorry...
@bitsundbolts love your content otherwise just wanted to let you know that I normally don't touch any videos with ai "art" on the thumbnail.
@LetsPlayKeldeo you folks a effing weird.
@@znk0rfake fingers weird, misinformation weird, or being the tool of capital to harvest free labor from unprofited workers thereby reproducing inherent flaws within capital markets weird? AI does all three, so that’s gonna be hard to beat!
The content is what matters
Once again, no disappointment!!
TY European bro !
you sir are a true WARLOCK
Hehe, perfect match for Voodoo and Witchery!
Awesome!
Thanks!
Not sure if it's just the capture, but at the end in Unreal, the color of the sky randomly flickers. I wonder if it's an issue.
I see. The one to the right of the large tower. I have a feeling that this is an encoding issue. I'll check the original footage tomorrow.
I was always mildly annoyed by benchmarks of that time, because they looked way more fun than most games I owned ;)
Haha, yes! It was frustrating to see 3D benchmarks that seemed like an awesome game but couldn't be played.
What good luck to stumble across the problem so quickly!
Was there a bit of fuzzy corrosion around some things or does the microscope just really highlight normal 25 year old dust?
There was a layer of dust over the card. I cleaned it after I was done with the video - now it looks clean again. And yes, I was very lucky to find the issue that quickly - if I ever would have found it. Nevertheless, I learned something. A broken resistor that can't can't be identified as such, can be the reason for a faulty card. So, if there isn't anything obvious, we have to check the resistors!
Hi, i have a Quadro FX4500 with no display and a damaged component labelled L514 (mentioned on Vogons) which i guess is an inductor? But how do i tell value (no markings)? I cant see any other obvious flaws
Unfortunately, I don't know how to get the value of this inductor. If I had this issue, then I would try to find a similar looking component on another device (an inductor of course) and see if the card changes its behavior. An inductor will measure like a bridge - continuity. It resists current flow changes if I understand this device correctly.
24:00 Maybe you should heat the solder one more time...🙃
Haha, sometimes I'm just not happy with the orientation of those tiny SMD components and get sidetracked. I know this is painful to watch 😅
@@bitsundbolts Especially when you're the very same behavioral 'victim' yourself. "Just one more dab"
The resistor you de-soldered was upside down. At least, according to the letters on it. Was this the case on your "good" board as well? Or did someone already try to repair this card earlier?
They're not directional.
Up and down are the same
@@myne00 I know. But I was wondering because I wouldn't think the robot making the card would have put it upside down.
@@roelbrook7559 30 years ago computer vision lacked processing power to bother with resistor orientation.
It was probably upside down in the tape and passed the automated tests.
I mean, did you not react immediately at the numbers being upside down compared to all the others?
Probably already altered once before.
It is very common to have resistor arrays placed in different orientations. A resistor can be placed either way. Those components most likely come from a wheel with thousands of resistor arrays. A pick-and-place machine is then populating PCBs. It all depends on the orientation the resistor/array/network had in that wheel. I believe this resistor was added by the factory and just got hit in that corner and cracked.
Хорошая работа !!!!
Thanks!