I'm really enjoying your videos, thanks for sharing. I did an electronics course as part of a telecommunications apprenticeship 30 years ago, mainly because 30 years ago there were no telecommunications trade courses. I got good marks but never had a true understanding and never did board level repair. I'm getting into it as a hobby and channels like yours are a goldmine. really appreciate it. 1.8v was missing and this AND gate which is on the 1.8v rail looks faulty. might not be a coincidence.
To confirm the fix while waiting a month for parts,,, Why not take any AND you can find w/ the correct voltages, convert the output if you have to with diode or transister. Its not high freq, its just reset, so solder a little circuit together with wire wrap wire etc. Tape it on the back. Check the test circuit works and check the card is fixed.
Yeah I could also probably try injecting 1.8V into the output of the AND gate and see if the card boots. At this point I guess I just decided to order some, as they also seem useful things to keep in stock, and get on to the next job.
@@LearnElectronicsRepair From a circuit design point of view they are nice. Because you can drop a gate anywhere and its easy to route. Vs trying to share around a hex package of gates around a large board.
I can see where ollie comes from that it aint worth for him... glad you bought it.. but we do expect to see the repair of it you changing the component and then testing !!!!
Thanks for sharing your trouble shooting. Super interesting. I'm new to electronics and am slowly beginning to understand your thought process. Please continue to do these videos. Thks
I also end up getting a lot of stuff from people where a 95% chance of fixing something 1 month from now is fair, but doesn't make financial sense for the customer. Then it becomes a personal project, meaning of gets deprioritized, and thrown in the pile of 300+ other unfinished projects.
@LearnElectronicsRepair So you're saying that the best way to complete all my unfinished projects is to start a UA-cam channel? That sounds like a fun new project! .... Wait a minute...
Amazing how you narrowed down the problem. It is a good card, have one myself using it for ai Fooocus and Comfyui. Looking forward to the final episode of this mini series 😊
Great troubleshooting, as others have said I would like to see the new part installed and the card working. A side note - your scope appears to have the horizontal display set to B Delayed. I would only use this setting after having set up the trigger on the main sweep and then "delaying out" to view an event occurring after the trigger. Using B Delayed with non-recurring signals such as random accesses of roms/rams can result in missed trigger events. Regards, David
Interesting as to start with a not all AND gates are sot23-5, not all graphics cards run on 1.8V logic, and not all 1.8V AND gates are 3.3V tolerant... so how can that be that I could use any? I did mention in the video I could possibly find a suitable part on a scrap graphics card but to me, considering I can buy ten for like 5 euros and I'm not in a rush, I'll save the time looking for a suitable one, just order some and get on with some other work in the meantime. Also then the next time I need an AND gate with these specifications i should have nine new ones in my component stock. But hey we all have different ways of working and that's fair enough 🙂
@@LearnElectronicsRepairIt’s not a problem if the AND gate can handle 3.3V; it will work at 1.8V. At the end of the day, it’s just a logic gate, and it doesn’t need to be precise in timing or anything else. I’m using VE AND gates, but when I don’t have any others, I use ones from RX 400 or RX 500, and they work for me. Most of them use 3.3V. (The output of the AND gate equals VCC, so don’t worry-it will be 1.8V at the output.) An interesting thing I noticed: the 1.8V line actually has 1.85V. On some cards, if the voltage goes higher than that, it will kill the core. I don’t remember the value of 1.8V in the RTX 3000 series, so maybe it’s normal. I also had an RTX 3060 Ti Palit where another AND gate (not the one you mentioned) was shorted. I replaced it, and it worked normally.
@@LearnElectronicsRepair I think you are right. No need to bodge something together that might end up causing more problems if you make a mistake with the bodge. Working in tight SMD spaces like that with all manner of wires, doesn't sound practical. Much better idea to wait for the right part 😉
@@mohammedb.8456 You are not wrong and neither am I really, it's just a different way of working, if it comes to the same result in the end then we are both winners 🙂
You would have to be extremely lucky to have found that the reason it doesn't work is that little logic chip. Personally I think the main issue is something else totally unrelated to that issue. Modern gfx cards are extremely complicated and very few dead cards are successfully repaired. But we'll see in 3-4 weeks ;-)
I agree that 60eu seems like a deal. I work a lot on games consoles for a buddy who owns a game store and I give him 1/2 on most things. Nice find on the AND gate. Not sure why but those fail somewhat regularly even if the card hasn't been abused. Doesn't make since to me since it isn't high voltage, high amps, or outside facing. I still wonder why the card decided to power on when you were poking that regulator chip. Maybe a bit of smutz or something bringing one of the pins down or maybe something was "latched closed" and you shorted a couple of the pins together when you were poking around and it livened the thing up.
Why you don't shunt the 1.8 v with the output for testing the graphic card ? And if give picture then order the and gate... Thanks for sharing your knowledge... Greatings from Portugal 🇵🇹 🌟
Id be seeing if i could find the and gate on a scrap board or an equivalent . Hate waiting for parts. Sometimes ill have something which will "do the job" on a scrap board , to prove the fault and then order parts after
I did mention that possibility in the video. I'm not sure I have 1.8V AND gates with 3.3V tolerant inputs. Also I had to make a quick decision as it's Friday and this either goes back to the vendor on Sunday or it doesn't
Yeah you could do that. It could just start or there may be some timing issue between PCI_RST and the GPU_RST. Personally I'm just happy to order some parts as I am confident enough of a fix but I get that some will have a different approach and that is not wrong either IMHO. IN the end it is the result that counts
I think if I could find a part on a donor board I would put it on to see if it worked or if something else is causing that part to go bad so I dont ruin a new part. I can't tell you how many times I've done that because I didn't have a doner card to get the part from just to be sure.
If it's a logic issue and a signal is low when it needs to be high and the chip is prime suspect, it's a simple matter to take that pin high manually with a bit of wire and prove it there and then, why wait ? there is a possibility of other faults as you well know ! or even remove the chip (your going to anyway) tack it to a piece of veroboard and test it perhaps ?
can the 1.8V be manually injected just to prove that it will work once the replacement AND gate chip becomes available? Nice video and detailed troubleshooting by the way.
@@LearnElectronicsRepair Yes, and the 5, now 12V... Yeah, I was thinking on the voltage rails part... Pushing a probe, specially a thin one like yours, might have been enough to close the circuit again...
@@aerogfs I've been thinking about that myself. It looks like, from the schematic, that Vin to the 1V8 switch IC has both a 5V and 12V supply. The only reason I can think why that would be is to get the GPU to return some response if the EATX-12V connector was not plugged in to say it needs plugging in. When I see 5V on the inductor that should have 12V it kinda suggests a bad connection from the 8way EATX conneycor. I'm thinking if this happens again just connect 12V to that inductor?
I would go back to the seller and tell him it's not working and not give him any more dollars for it. The 50 was if it's working. It's not so it's only worth the 20 already given to him. So give Ollie the 20 to refund his purchase and that's it.
Before ordering it just simulate a reset. Buying bits to fix a dead card is not the way forward. The correct solution would have been to tie the output to 1.8V through a 4.7K resistor then inject a low onto the output pin to simulate the reset then let go, the resistor pulls the pin high and the card should be working if that is your only issue. You could test that on the bench in 5 minutes. Come on Rich, use some more of those brain cells....
Heya, yes as you explained I think it's the reset. and as we didn't see any shortisches on the PCB the GPU and memery will most probeble not be damaged. 100 euro+ after repair for resale not sure if w'll sale it. but as you said for your own use it will be perfect
It's very possible and happens more often than you might think. I showed exactly how that happens on the first video ua-cam.com/video/_PAkfxbkmp0/v-deo.html
When there is a will, there is a way... Long time ago, a friend managed to install a SODIMM DDR on a DDR2 slot.... Then complained it was not working...
I'm really enjoying your videos, thanks for sharing. I did an electronics course as part of a telecommunications apprenticeship 30 years ago, mainly because 30 years ago there were no telecommunications trade courses. I got good marks but never had a true understanding and never did board level repair. I'm getting into it as a hobby and channels like yours are a goldmine. really appreciate it.
1.8v was missing and this AND gate which is on the 1.8v rail looks faulty. might not be a coincidence.
Roll on mid Feb. Let’s see this working 👍👍
Yeah I'll show it on the live stream either way
To confirm the fix while waiting a month for parts,,, Why not take any AND you can find w/ the correct voltages, convert the output if you have to with diode or transister. Its not high freq, its just reset, so solder a little circuit together with wire wrap wire etc. Tape it on the back. Check the test circuit works and check the card is fixed.
Yeah I could also probably try injecting 1.8V into the output of the AND gate and see if the card boots. At this point I guess I just decided to order some, as they also seem useful things to keep in stock, and get on to the next job.
The AND gate is fault, when not detected is the first chip to be tested after normal pcie tests
@@buug76 Thanks for your confirmation
@@LearnElectronicsRepair From a circuit design point of view they are nice. Because you can drop a gate anywhere and its easy to route. Vs trying to share around a hex package of gates around a large board.
I can see where ollie comes from that it aint worth for him... glad you bought it.. but we do expect to see the repair of it you changing the component and then testing !!!!
Yeah we will try it on live stream after the parts turn up
Thanks for sharing your trouble shooting. Super interesting. I'm new to electronics and am slowly beginning to understand your thought process. Please continue to do these videos. Thks
00:11:40 - The logo is Alpha Omega not A Open
If that fixes it it will be a great video editing gpu for you Richard.
Can't wait to see the results.
I also end up getting a lot of stuff from people where a 95% chance of fixing something 1 month from now is fair, but doesn't make financial sense for the customer. Then it becomes a personal project, meaning of gets deprioritized, and thrown in the pile of 300+ other unfinished projects.
True but I've promised to show the results on the following live stream, so I''m commited now 🙂
@LearnElectronicsRepair So you're saying that the best way to complete all my unfinished projects is to start a UA-cam channel? That sounds like a fun new project! .... Wait a minute...
Amazing how you narrowed down the problem. It is a good card, have one myself using it for ai Fooocus and Comfyui. Looking forward to the final episode of this mini series 😊
Great troubleshooting, as others have said I would like to see the new part installed and the card working. A side note - your scope appears to have the horizontal display set to B Delayed. I would only use this setting after having set up the trigger on the main sweep and then "delaying out" to view an event occurring after the trigger. Using B Delayed with non-recurring signals such as random accesses of roms/rams can result in missed trigger events. Regards, David
you can use any AND gate.
You can find them in any graphics card.
Interesting as to start with a not all AND gates are sot23-5, not all graphics cards run on 1.8V logic, and not all 1.8V AND gates are 3.3V tolerant... so how can that be that I could use any? I did mention in the video I could possibly find a suitable part on a scrap graphics card but to me, considering I can buy ten for like 5 euros and I'm not in a rush, I'll save the time looking for a suitable one, just order some and get on with some other work in the meantime. Also then the next time I need an AND gate with these specifications i should have nine new ones in my component stock. But hey we all have different ways of working and that's fair enough 🙂
@@LearnElectronicsRepairIt’s not a problem if the AND gate can handle 3.3V; it will work at 1.8V. At the end of the day, it’s just a logic gate, and it doesn’t need to be precise in timing or anything else. I’m using VE AND gates, but when I don’t have any others, I use ones from RX 400 or RX 500, and they work for me. Most of them use 3.3V. (The output of the AND gate equals VCC, so don’t worry-it will be 1.8V at the output.)
An interesting thing I noticed: the 1.8V line actually has 1.85V. On some cards, if the voltage goes higher than that, it will kill the core.
I don’t remember the value of 1.8V in the RTX 3000 series, so maybe it’s normal. I also had an RTX 3060 Ti Palit where another AND gate (not the one you mentioned) was shorted. I replaced it, and it worked normally.
@@LearnElectronicsRepair I think you are right. No need to bodge something together that might end up causing more problems if you make a mistake with the bodge. Working in tight SMD spaces like that with all manner of wires, doesn't sound practical. Much better idea to wait for the right part 😉
@@mohammedb.8456 You are not wrong and neither am I really, it's just a different way of working, if it comes to the same result in the end then we are both winners 🙂
Thanks for the trouble shutting tutorials 👍🏻
Hi Rich, another top video thank you. Would be intersting to see what price fully working ones go for if you can find one.
I had a look on ebay UK and used working ones are around 200 (UK pounds) or 235 euros
You would have to be extremely lucky to have found that the reason it doesn't work is that little logic chip. Personally I think the main issue is something else totally unrelated to that issue. Modern gfx cards are extremely complicated and very few dead cards are successfully repaired. But we'll see in 3-4 weeks ;-)
The card was not frightened enough of you the firs time.
Quite possibly lol 😆
I agree that 60eu seems like a deal. I work a lot on games consoles for a buddy who owns a game store and I give him 1/2 on most things. Nice find on the AND gate. Not sure why but those fail somewhat regularly even if the card hasn't been abused. Doesn't make since to me since it isn't high voltage, high amps, or outside facing. I still wonder why the card decided to power on when you were poking that regulator chip. Maybe a bit of smutz or something bringing one of the pins down or maybe something was "latched closed" and you shorted a couple of the pins together when you were poking around and it livened the thing up.
Why you don't shunt the 1.8 v with the output for testing the graphic card ? And if give picture then order the and gate... Thanks for sharing your knowledge... Greatings from Portugal 🇵🇹 🌟
21:26 top centre, cracked pcie?
Id be seeing if i could find the and gate on a scrap board or an equivalent . Hate waiting for parts. Sometimes ill have something which will "do the job" on a scrap board , to prove the fault and then order parts after
I did mention that possibility in the video. I'm not sure I have 1.8V AND gates with 3.3V tolerant inputs. Also I had to make a quick decision as it's Friday and this either goes back to the vendor on Sunday or it doesn't
Hey Rich, Could you short pin 4 and 5 to see if it works or has other problems? or does that not work like that?
Yeah you could do that. It could just start or there may be some timing issue between PCI_RST and the GPU_RST. Personally I'm just happy to order some parts as I am confident enough of a fix but I get that some will have a different approach and that is not wrong either IMHO. IN the end it is the result that counts
Sounds very fair
Can't you just bypass the gate on the bench just to see if that will fix it? Just to enable the chip and see if it works?
I think if I could find a part on a donor board I would put it on to see if it worked or if something else is causing that part to go bad so I dont ruin a new part. I can't tell you how many times I've done that because I didn't have a doner card to get the part from just to be sure.
If it's a logic issue and a signal is low when it needs to be high and the chip is prime suspect, it's a simple matter to take that pin high manually with a bit of wire and prove it there and then, why wait ? there is a possibility of other faults as you well know ! or even remove the chip (your going to anyway) tack it to a piece of veroboard and test it perhaps ?
See you on Dan👍🇨🇦
can the 1.8V be manually injected just to prove that it will work once the replacement AND gate chip becomes available? Nice video and detailed troubleshooting by the way.
Yes probably
110 is still a good deal for a card of that capability!
Couldn't any solder joint be cracked? As you started getting different results after probing this second time.
You mean on the missing voltage rails? 1.8V etc?
@@LearnElectronicsRepair Yes, and the 5, now 12V... Yeah, I was thinking on the voltage rails part... Pushing a probe, specially a thin one like yours, might have been enough to close the circuit again...
@@aerogfs I've been thinking about that myself. It looks like, from the schematic, that Vin to the 1V8 switch IC has both a 5V and 12V supply. The only reason I can think why that would be is to get the GPU to return some response if the EATX-12V connector was not plugged in to say it needs plugging in. When I see 5V on the inductor that should have 12V it kinda suggests a bad connection from the 8way EATX conneycor. I'm thinking if this happens again just connect 12V to that inductor?
You have donor cards, just take any and gate from other card :)
I would go back to the seller and tell him it's not working and not give him any more dollars for it. The 50 was if it's working. It's not so it's only worth the 20 already given to him. So give Ollie the 20 to refund his purchase and that's it.
Before ordering it just simulate a reset. Buying bits to fix a dead card is not the way forward. The correct solution would have been to tie the output to 1.8V through a 4.7K resistor then inject a low onto the output pin to simulate the reset then let go, the resistor pulls the pin high and the card should be working if that is your only issue. You could test that on the bench in 5 minutes. Come on Rich, use some more of those brain cells....
Heya, yes as you explained I think it's the reset. and as we didn't see any shortisches on the PCB the GPU and memery will most probeble not be damaged. 100 euro+ after repair for resale not sure if w'll sale it. but as you said for your own use it will be perfect
True but as I'm going to keep it then it just costs me €40-€50 if I supply my own time for free to myself
Tune in same bat time same bat channel.
Showing your age :)
@@andymouse people don't still like Batman?
@@1pcfred Well your quote is from very old TV Batman not sure if it's still screened !
@@andymouse I have not seen it in a very long time now. Still such an iconic line.
@@1pcfred Oh yes ! a classic :)
its actually impossible to plug in a mining adapter the wrong way round, as it wouldn't fit into the pcie slot the motherboard wrong way round
It's very possible and happens more often than you might think. I showed exactly how that happens on the first video ua-cam.com/video/_PAkfxbkmp0/v-deo.html
@@LearnElectronicsRepair well in that case those people must be absolute idiots
When there is a will, there is a way... Long time ago, a friend managed to install a SODIMM DDR on a DDR2 slot.... Then complained it was not working...
Could the other chip have fallen off rather than be missing?
The other one, U512 in the reset circuit?
@ yes
I think you will have a very nice 3060 card for 5 euros.
Yeah I am quite confident enough to take that chance