@@vswitchzero greetings vswitch. Watching this video, in which you use an S3 Virge DX, I ask you a question for help. I have an S3 Virge DX card (AOpen PT 75), which I had never used before (so dont know its previous state). after a good cleaning and soldering a capacitor that had come loose. I tried it and it did not give an image, the PC emitted several beeps. After an additional review and more cleaning, it managed to show the vbios report, but then it goes dark. Reviewing it more in depth I saw that it had a position where another capacitor seemed to go and in the images that I looked for I saw that yes, it was missing another capacitor (those are the two on both sides of the PCI connector i this card). In the images I cannot see the data of the capacitor, but visually it is the same as the one that I soldered before. So I matched this capacitor and put it on, but the card remains the same, it shows the information from the vbios, and that's it, it doesn't give an image. during boot up i keep touching the chip and i feel it getting hotter than i think it should, so i turn off the pc. Can you give me any advice on this? also I have two AOpen VI15G motherboards , the revision with support for 3.3v CPUs (it has the power regulator). They are part of some PCs (the other is a Pentium) that I had at the end of the 90s or the beginning of this century and all the parts and pieces were stored in working conditions almost 20 years ago. A few months ago nostalgia assaulted me and I am reassembling them. All the parts needed a lot of cleaning, from the PSU to the cards, due to the dust accumulated even inside the boxes, in addition to some slight oxidation. Everything was pristine after cleaning. one of the two psu need some repair but the other ones is working flawlessly. I have some 486 but the best one is an Am5x86. I didn't remember if the jumpers were set for use that one (i think they were, but not sure at all, 20 years bro), so I looked for the documentation and I found two documents, one original from the first version of the motherboard, which only supports 5v, and another with jumper settings updated for the revision with 3V support). I configured the motherboard as an Intel 486DX4, also as an AMD DX4 (the board is from 2 years before the 5x86), based on some recommendations I read around. both are 3.3/3.45 volt processors, but when measuring the voltages on the socket pins and on the regulator output, it's always 5V no matter the configuration. It happens equally on both motherboards, I don't think the regulator is defective in both cases. I would appreciate if you help me with the correct configuration to be able to use the Am5x86 (or 486DX4 from Intel or AMD, as a valid config as I see). The other 486 compatible I have is an AMD 486DX2-66 but of course the primary interest is the Am5x86. I still need to check them with the DX2s to make sure. the PentiumMMX is also not working, no post, but i will see it latter Any help will be very appreciated.
Pre-heating absolutely makes a huge difference. You'll get more bulk heat soaked into the board meaning you don't need to dwell as long with the hot air station.
This really is invaluable information. I have a Banshee that also pops up with the same error message when flashing, but it doesn't show up under device manager sooo. Who knows.
I really enjoy your videos, the way you describe your techniques and logical troubleshooting is really interesting to follow. You explain things clearly and your filming quality is absolutely top notch. Thanks for these interesting adventures you bring us on!
Congratulations on 11K and a nice repair! Those tiny traces are really easy to miss. Especially on those cards where it's quite busy with SMDs everywhere.
Thanks very much! Very true - I spent a lot more time checking the back because of all the scratches all over but those are all easier to spot since there are no SMDs anywhere :)
Flux is needed to clean surfaces from oxides during soldering. But when you desolder with a hot air station, it only interferes - you do not see the moment when the solder becomes shiny, completely melted. The flux evaporates rapidly, as the temperature of the hot air station is too high for it, and produces too much smoke, which makes the desoldering process more difficult. So there is a chance to tear off pads. But for soldering components back or cleaning pads - there is not such thing as too much flux.
Nice Job. It appearance to me that you have the right approach fixing it. I had that same model once. I still have fond memories of this card. I really miss 3dfx :*(
Hurrah! Nice work. I really appreciate the style of your videos they're really chill, great narration and nice editing, good work! Oh and correct 3 attempt insertion of USB plug 😂
Excellent work. 25:35 it would be really interesting to see if the Voodoo 3 is able to run at over 200 mhz with 5ns or even faster memory. Apparently the fastest Voodoo 3 ever released was made by Falcon Northwest runnung at 200 mhz, but I never saw one on the internet and I wonder if such a card even exists.
Very interesting about the Falcon Northwest Voodoo 3. I've never heard of that one before. I'm definitely going to try to get this card past 200MHz, stay tuned :)
You are absolutely right about using flux, never ever do fine soldering (micro/smd) without proper amount of flux, or as Alex @ northridgefix would say, flux helps with the flow of solder.
You're becoming quite an expert on Voodoos. I have a V2 that has seen better days, no clue what happened with it, but it won't even be detected by OS or bios.
Best of luck with your V2! If it isn't detected - take a closer look at the frame buffer chip (the middle one near the slot). If there are any shorted pins, that could do it. Also make sure the regulator is providing the correct voltage output. Could also be failed resistor arrays, which I've seen before too.
I actually saw a voodoo 3 on another channel that had vbios issues. So he flashed it. It worked. He examined the old rom. Only a few changed bytes. 9 in total. So he reflashed the original. Still worked perfectly. lol It's so bizzare.
It is very strange indeed. I had a similar experience with a socket A motherboard a while back. The BIOS was bricked/corrupted and the system wouldn't post. I pulled the image off via my external programmer, and its checksum matched the one I downloaded from the ASUS site exactly. Flashed it back and it worked just fine. I suspect the external programmers do a better job of reading "borderline" cells that don't have enough charge left in them than the board/card does. I remember hearing that it can take too long to read those cells too, which can be an issue. So the image pulled by the programmer is good, but the card/board doesn't pull the same data when it tries to read it. Flashing it again restores all those flaky cells. If I find another EEPROM in this kind of state, I'd like to do more tests on this.
@@vswitchzero yeah that was my guess as well. That it was marginal and just not fast enough to POST. But flashing I guess doesn't care about speed? /shrug
I haven't done significant soldering in a while, but I did a lot of small electronics repair for my father's business back in the mid 00's, similar SMC sizes to these. Anyway, we always used a flux impregnated solder and it worked fantastically. It was also lead based I'm pretty sure, and at least back then, the non-lead based solders were crap and required much higher temps.
Thanks for your comment and for sharing! I have never tried lead-free solder before (since most of what I repair is from before that era) but I've heard its much harder to work with.
@@vswitchzero They had a certain brand they loved with the flux in it and stock piled it. The business went under (tech they fixed basically went obsolete), but he still has some somewhere in his garage. But because it was flux impregnated, we never needed extra flux and would just melt a little extra solder if you needed to and then wick away the excess.
Congrats for the milestone, very well deserved! Nice video and I have a feeling the seller where you found the card is the same I source some of my stuff from 🙂 Preheating is a must. Someone might say ‘I’ve never used it and it’s always worked for me’ but just blasting silly temps to an IC is not good for either the component nor the pcb. For the broken trace, I’d say the pin would be floating and not low all the time - so it would probably be seen low and high at random by the chip. Anyways, I enjoyed the video and well done!
Thanks so much Tony, I really appreciate it! Congratulations on all of the success on your channel as well. I really enjoy watching your repair videos and have learnt a lot from you :) .. Thanks for the tip on preheating. I will look into picking one up. If you have any recommendations, it would be appreciated! .. also great point on the pin "floating".. I automatically assumed it would just behave as low when disconnected but there is clearly more to it! Thanks again.
Great work 👌 👏.Preheating is necessary when there is a bga ICs on the board .Preheating before removing the IC and after soldering the IC. It must be go from room temperature to soldering and de-soldering temperature in gradual manner. Because the thermal shock may not affect the removed BGA IC, but affects another BGA IC adjacent to it, and breakage occurs at the BGA solder points.
Thanks very much! And thanks for the info. I'm hoping that one of these days I'll be able to attempt some BGA work. That's one area I've never ventured into yet.
Thanks very much! It's always such an exciting feeling to get something working (especially after numerous failed attemptes).. that "aha" moment is priceless :D
If you haven't under preheater - better to mix the solder of chip with the rose alloy by a soldering iron, and then unsolder it hot air gun. Rose alloy has a much lower temperature of melting and it's hardens with muuuuch more time that ordinary solder, that is why unsolder chips with it much easier
I remember walking into Futureshop and buying my Voodoo 3 2000 in early 2000. The only piece of hardware that I'm not sure what I did with when I had to upgrade. I should track down one someday.
I have so many fond memories of going to Futureshop back in the 90s. I still remember going there to buy my first modem, and again later on to buy my first 3dfx Voodoo card too. Was always an exciting place to visit :-)
Hi, yes a multi plane PCB certainly needs a pre heat, it will make desoldering much easier, less tracks and pads pulled off, Small boards you can pre heat the underneath with your hot air gun before doing the yop. Bob. UK
Thanks for this video! I'm trying to troubleshoot a Voodoo 3 2000 PCI that isn't displaying, but the system will POST with it in, and found your vid. I'm very impressed with your troubleshooting skills, I wish I had that familiarity with soldering and circuitry.
15:45 When plugging in a USB cable, the open side (where you can see through the two square holes) ON THE CABLE always faces away from the PCB. That almost always means that the open side will face up on a laptop, or to the left on a desktop PC when reaching around the back. There are some exceptions to this, like most notably two IKEA items I have, a charger and a battery bank where the USB ports are upside down from the norm, with the "open" side facing down from the top.
The pin was likely defaulted to high, the busted trace stopped it being pulled low (to ground). It's the only situation that fits the data sheet. Congrats on 11k 👍
Thanks for your comment! Yes, I'm pretty sure you are right about that. I assumed that a disconnected pin would be 0V, but I checked the pin again with the system powered up and during normal operation it measures 3.3V. So I think you are right that the trace pulls it low instead of the other way around.
@@vswitchzero you can see that CS, OE and WE pins all have "bars" over their text names. This means that they are active low signals, i.e. must be pulled low to be considered active (asserted). When they are high the signals are inactive (negated). For many types of ROMs and RAMs you need to assert both CS and OE to read from the device, or CS and WE to write to it. In some cases, OE can be permanently tied to ground, and reads are effected simply by asserting CS, or writes can be made by assessing both CS and WE, where WE effectively overrides OE. This can save a little bit of circuitry during design. I don't know if I've ever come across a datasheet which indicated whether an unconnected pin on one of these devices would be pulled high or low by default (I've certainly never noticed it if it was mentioned), but since they are very likely based on CMOS manufacturing processes, an unconnected pin will float to some random state and could thus result in unpredictable behaviour. What value it floats to could be influenced by noise from neighbouring circuitry, or by simply moving your hand or some test instrument near it. And this can even change from moment to moment, which can lead to very confusing behaviour during troubleshooting. Given they are active low signals it would make more sense for them to be pulled high if unconnected, to make sure the device is disabled and wouldn't interfere with any others that are fully functional. But I suspect they aren't pulled in either direction and are expected to be connected to something that sets their state. I would also suggest that "output disable" mode can't really be broken. That mode simply means that, since OE is negated, the ROM won't ever drive the data bus if CS is asserted.
Top work as always, I can only aspire to be as good as you at this kind of stuff! Ive modded a few retro consoles but my skills aren't anywhere near good enough to attempt such small work as you do. Well done :) You have a new Patron in me, thats for sure.
Wow, thanks so much for becoming a Patron! I really appreciate your support! I had almost zero soldering skill only a couple of years ago. It just takes practice, patience and some decent quality tools. If I can do this, anyone can, that's for sure :) .. thanks again!
That was the best one yet, learned a lot today. I love the way you troubleshoot these cards and explain every step so well, you're very intelligent. Congrats on the milestone man looking forward to more!
I scrapped so many of these sorts of things in the early 00s. Basically I was buying each generation as they came out from the first Voodoos for developing on. It turned into a giant pile of stuff... that eventually went for recycling. I carefully archived all the 80s 8-bit kit I had, but never thought anyone would collect PC video cards. I'm sorry!!
I never would have imagined back then either. I always sold off my aging hardware to fund the next upgrade, or just gave stuff away when it was too old to be useful. Hard to think that people may be trying to collect GTX 1080s and cards like that in the future :D
Amazing video. Again I learned so much from this. Super valuable information that I will definitely come back to if any of my Voodoo 3 cards stop working.
Good efforts man, thanks for sharing Man in older times we had lots of display/video card choice now there be just AMD versus Nvidia with Intel as 3rd choice Intel jumped all the way from i740 to Arc gpus 380, 750 & 770 but never the 580 it was cancelled :/
Don't feel bad about the false start on troubleshooting. When you have to magnify the board to that level, it's so easy to miss some damage like the kind you eventually found.
Not being connected is not the same as having a grounded pin or a "low" applied. The pin might actually be floating as it's called. Basically it's level is undefined as the datasheet specifies what happens if the pin is properly connected not when it's not. It might have a defined level if there are internal pullups or pulldowns. ( And they are activated). But I would not expect them to be there on a flash chip.
Big congratz for the milestone and another repaired card of course! Found your channel a while back and have been enjoying it since and learning myself while I do!
Yeah, that makes a lot of sense. I was thinking of it in too much of an "on/off" situation and there is clearly more to it than that. Appreciate the comment!
Great repair. I have a Voodoo 3 2000 PCI that I don't use because it locks up Windows 98, usually when exiting Windows DOS prompt or a DOS game. No clue what is wrong with it..
Something like that would be really useful! I did look around for something of that sort came up empty. Probably wouldn't be too complicated to design but not my area of strength unfortunately.
I have the same model card that hasn't worked since I bought it (system wouldn't post). Thanks to this video I tried the switcheroo trick with a PCI card, but my system still won't post, so I get no video output at all. Any ideas on what could be wrong? Edit: tested on a different system...it's completely dead
Thanks for your comment! If the card isn't detected at all, it could be power related (check the regulator for the correct voltage output and test the ferrite beads/filters to the left of the 3dfx chip to make sure there is near zero impedance across them). Unfortunately, some of these types of issues are due to BGA contact issues, which can be much more difficult to fix. I have a V3 here that I believe that is the issue on. One of these days I'll need to learn how to do BGA rework.
Thanks for the tip. Haven't tried no-clean stuff yet, but the flux I use is really easy to get off without much effort. I used some cheap flux paste in the past that required a lot more work to remove.
hi vswitch. Watching this video, in which you use an S3 Virge DX, I ask you a question for help. I have an S3 Virge DX card (AOpen PT 75), which I had never used before (so dont know its previous state). after a good cleaning and soldering a capacitor that had come loose. I tried it and it did not give an image, the PC emitted several beeps. After an additional review and more cleaning, I managed to show the vbios report, but then it goes dark. Reviewing it more in depth I saw that it had a position where another capacitor seemed to go and in the images that I looked for I saw that yes, it was missing another capacitor (those are the two on both sides of the PCI connector). In the images I cannot see the data of the capacitor, but visually it is the same as the one that I soldered before. So I matched this capacitor and put it on, but the card remains the same, it shows the information from the vbios, and that's it, it doesn't give an image. during boot up i keep touching the chip and i feel it getting hotter than i think it should, so i turn off the pc. Can you give me any advice on this? also I have two AOpen VI15G motherboards , the revision with support for 3.3v CPUs (it has the power regulator). They are part of some PCs (the other is a Pentium) that I had at the end of the 90s or the beginning of this century and all the parts and pieces were stored in working conditions almost 20 years ago. A few months ago nostalgia assaulted me and I am reassembling them. All the parts needed a lot of cleaning, from the PSU to the cards, due to the dust accumulated even inside the boxes, in addition to some slight oxidation. Everything was pristine after cleaning. one of the two psu need some repair but the other ones is working flawlessly. I have some 486 but the best one is an Am5x86. I didn't remember if the jumpers were set for use that one, so I looked for the documentation and I found two documents, one original from the first version of the motherboard, which only supports 5v, and another with jumper settings updated for the revision with 3V support). I configured the motherboard as an Intel 486DX4, also as an AMD DX4 (the board is from 2 years before the 5x86), based on some recommendations I read around. both are 3.3/3.45 volt processors, but when measuring the voltages on the socket pins and on the regulator output, it's always 5V no matter the configuration. It happens equally on both motherboards, I don't think the regulator is defective in both cases. I would appreciate if you help me with the correct configuration to be able to use the Am5x86 (or 486DX4 from Intel or AMD, as a valid config as I see). The other 486 compatible I have is an AMD 486DX2-66 but of course the primary interest is the Am5x86. I still need to check them with the DX2s to make sure. the PentiumMMX is also not working, no post, but i will see it latter Any help will be very appreciated.
That's strange that the system displays the vbios text strings but doesn't display anything from the BIOS post process. I can't say I've ever seen that before. Do you happen to have another video card you could test out in the system to see if the behavior is the same? It is normal for the Virge chip to get warm after some time, but if it gets really hot (like so hot you can't leave your finger on it for longer than a second or two) that could be an indication of some other problem with it. If the same thing happens with another card, I'd look at the motherboard and BIOS. I've seen strange things happen with power supply issues too, so could be power related also. Sorry I can't really be of more help on that one. In regards to the AM5x86, boards that don't officially support it can usually still work. It's usually just the write-back L1 cache that can't be enabled. The jumper settings for a DX2 or DX4 should be fine, but to get it to run at 133MHz, you'll usually need to set the multiplier selection jumper that's normally used to set a DX4 CPU to run at a 2X multiplier instead of 3x. Once you set that, the AM5x86 should run at the 4x multiplier. I did find this PDF document on the retro web that may help. There is a column for 5x86: theretroweb.com/motherboard/manual/vi15g-settings-61a731bc3ef30571520750.pdf That's strange that the regulator is outputting 5V. So both the input and output pins are reading 5V? Usually boards like this will just bypass the regulator when configured for 5V operation so that the regulator isn't used at all. But if you are measuring 5V on the VCC pin of the socket, then that's definitely a problem. I don't see a manual for that board with 3V support but perhaps it was just configured for a 5V CPU prior and needs some re-jumpering?
@@vswitchzero thanks for your answer. Since the Socket 7 motherboard doesn't work either, I am testing the card in an infinitely more modern PC, a 4th generation Intel, its board still has PCI. on that one I also tried a S3 Trio 64+ (2MB) and Win10 boot and showed image.... in a few shades of gray, LOL About the temperature, well, I don't think it reaches that much, I just don't think it would get hot without doing any work. I'll see how hot it really gets, I never give it enough time just in case I do have that documentation, it is the one I have used, but I read comments that this 5x86 config refers only to Cyrix's, and for this reason it is recommended to use the DX4 configuration. on the vogon web someone has it with an intel dx4-100 (another with a dx4 overdrive or something like that, but this is not valid), but in the images you can only see a part of the configuration jumpers. I tried to register on the web to ask but it didn't work for me. all the images I found are with DX2, and one (from the theretroweb website) with the supposed Intel DX4 (W/B) configuration, but his one is strange because it is teh 5v only version of the board (unless they use an overdrive, I suppose) PS: I just gave the regulator a close look. on the ADJ leg of the voltage regulator (the "transistor" like on the left, if the board is positioned so that they are below, the one with a heatsink, it is an LT1085CT,) there are 3.3 volts!!!. even the vein that reaches it is thicker, but short, it comes from the main side of the board. however on the output leg (vout, the middle), the input leg and the VCC pins on the socket there is 5v. does that mean its dead? I measure no shorts, not in the power lines, nor in the regulators. the other component is a NDP406BL FET, it has 12v on the G leg and 5 on the other two. I must check if the same thing happens on the other board
@@vswitchzero Well, I managed to get 3.3-3.5 volts, with two configurations, the AMD DX2 and the AMD DX4. the "3.45v" settings for the intel DX4 deliver 5V!!. however they still won't boot. I'm afraid if the 5x86 was burned when I put it with one of the intel jumper settings. I tried with an amd dx2 (3.3) and it doesn't boot either, no beeps or anything. I will check the boards more thoroughly
With name-brand capacitors from reliable sources (like this one) you can pretty much trust the datasheet. 0.85 Ohm at 100kHz is Panasonics rating for the EEEFP1C100AR cap.
@@Manawyrm I was asking, because I have a Voodoo5 5500 with lots of these caps. They all measure around 9-10 ohms @ 66KHz. I've measured some other similar caps I have, same ballpark. They are not Panasonics.
I have one of those really cheap component testers, but the ESR readings I get from it vary so wildly that I don't really trust it to be honest. I'd be happy to take some readings anyway if you like - just let me know.
Yeah the 2000's tiny heatsink gets VERY hot. When you say overheat, does the card start to artifact or crash when it gets too hot? Normally they should be able to handle very high temps. If you can find one of those slot coolers that you put in the slot below the V3 to provide some airflow - it would probably sound a lot better than a 40mm fan. Removing the V3 2000 heatsink is a risky thing to do, unfortunately. Prying at it can damage the BGA connections on the avenger chip, and I usually advise against it unless its absolutely necessary. Some people have had success with the freezer method as it weakens the epoxy and it can be pried off much easier. But this is still risky. I've heard of people using very thin wire to "saw" between the chip top and the heatsink, which cuts through the epoxy. It takes forever, but its definitely safer than trying to pry it off. Best of luck!
@@vswitchzero It's been a while but if I remember correctly the card would start to slow down after a few minutes of gaming to the point where it was basically a slideshow. Putting a tiny fan on it made it work fine again. I do have an old slot cooler but I haven't tried it. It's pretty loud too but not as high pitched as the fan so it would be an improvement. I'll give it a try. Thank you for the information.
Those atmels have a 20 year guaranteed life span. They die not long after that. I had e same problem with my nomad jukebox. Replacing it and flashing a new chip fixed it
Awesome thanks! The song at the end was a custom track composed for the channel by Bertrand Gagan and is called "Prelude to Space". He does some amazing work! You can find more of his stuff here: ru-vid.com/show-UCMrG-4sGyKJriiKIpFeuz5Q
This is the case with TTL families due to the way the gates are constructed, but absolutely not the case with CMOS families (which this ROM and indeed near on everything else on the board are manufactured in.)
Found one Voodoo3 from compaq on boot fairs maybe 10 years ago, missing capacitors, missing heat sink ,in bad condition overall i do not have skill to repair it or know anybody near me to repair it. Don't know if it safe to try it out without capacitors Everytime i clean my stuff i throw it away for recycling and next day i dig it up and put it in a box. Maybe someday someone can use it save it or use as spare part.
@@vswitchzero I kind of want to do this stuff myself, but I honestly don't know what I'd do with it even if I had your skills. I've never done SMD before. I have fixed a C64 and honestly I feel like I could do a better job now. It had 1 faulty ram chip (I knew which one, but I replaced them all because it's a known bad brand) and a faulty cartridge port. However because my home made power supply didn't have a thick enough cable, I had a huge voltage drop anytime I used my U2+ (the USB disk drive emulator) The computer worked, but barely and some programs were worse than others. But the voltages checked fine with the "naked board". One day I checked voltages with the cartridge. 3.5v. How the hell that computer still ran on 3.5v is beyond me. After that I replaced the power cable and it's worked great. But.........before I knew the real problem I replaced the PLA and the color ram. Both were fine. The color ram destroyed a trace in the process. So I had to bodge it. The cartridge port (definitely was bad) I lifted a trace. I'm still using it, but it's not flat on the PCB anymore. I have a hot air station now (I got it for 3d printing) I feel like that would have helped me avoid these issues.
@@awilliams1701 Great that you were able to do those C64 repairs! Its always a good learning experience - even if you replace unnecessary parts or cause damage requiring bodges etc. I was really intimidated by SMD stuff at first. I ended up buying some SMD practice boards from Amazon and just did component after component until it got easier. Eventually you discover techniques that work best for you. Getting a microscope was also a real game changer for me. Made all the difference. Cheers!
@@vswitchzero My problem is I barely have time for the C64 or anything else really. So if I got more parts regardless of electronics era, what would I do with them after fixing them?
