The Ultimate Short Circuit High Side MOSFET Tutorial Guide. 10 ways to find the faulty VRM. GPU CPU

hi guys welcome to a special learning electronics repair video in this video we're going to talk to very common problem we find with graphics cards and motherboards in particular and this problem is called a short circuit high side mosfet i'm going to show you in this video at least 10 ways to trace this sort of fault as to which mosfet is a short-circuit one some of these methods are quite novel and probably have not been shown before anywhere that i've ever seen for sure so whether you're a beginner whether you're intermediate or whether you experience with gpu repair i think you're going to learn something in this video i'm sure you're going to learn something in this video that you didn't know i'll divide the video into chapters so that you can skip through to the various different techniques i'm going to use and also if you want to refer back to this video later if you want to use one of these techniques it'll be easy for you to find that information that you want to watch again so if you like you can skip forward now to the various chapters in this video but i'd like to think you'd want to watch the whole thing and if you do watch the whole thing i'm sure you're going to learn a lot so to start with this is a graphics card that has a short circuit high side mosfet i originally started to make a video with this it belongs to a subscriber and he thought it may have been put into a mining adapter the wrong way around if you're connecting mining adapted the wrong way you're going to damage the gpu and i was making a video with this card i started to make the video and then realized this was an ideal opportunity to create this tutorial so first let's have a look at the graphics card it's an rtx 2060. let's have a look to see what is wrong with this graphics card although i've already told you and then let's have a look to see the various ways in which we can actually trace the short circuit component i have five graphics cards here which were brought in by two subscribers rubies and alvaro who live on gran canaria where i live and work there in the capital las palmas and they use these for currency mining they contacted me and asked me if i would look at these so i s yes i said i would and i made my usual offer which is to everybody i offer a no fix no fee service if i can't fix it cost you nothing i offer fixed price repair so if by the time i've diagnosed a problem i've basically fixed it then it's 45 euros if i diagnose a problem that requires parts or is going to be quite a big job of bga rework and this sort of thing then i'll negotiate a price and if you don't like the quotation back to the original i can't fix it you owe me nothing so that's how i work and that louise and i'll be like this so they've brought me these five cards this is the first of them um it says obviously zotac geforce gtx 1060 but it's not actually containing what it says on the box so we have the backing plate which is being removed and we have the graphics card in the anti-static bag yes and this is the card so it is a zotac but it's a g-force rtx and it's an actual fact if i can find the model number this for you this is an rtx 2060 six gigabyte graphics card so there's the card and the information i have with this one is that they think it's been inserted into a mining adapter the wrong way around if that's the case it's probably not repairable but it's very easy to work out if that's happened to a graphics card with a simple check before we check anything else so because we have an idea what the problem is with this let's go straight forward and see if we can prove that very very quickly this is a mining adapter and it runs in pcie single channel one times mode and this is the little adapter that fits into the pcie slot on the motherboard and it's very easy to insert these things the wrong way around and the problem is if you do that some of the signals here which is the first pcie lane and the reference clock end up connected to the 3.3 volt supply and that destroys the gpu normally so let's see straight away if that's happened and the way to find out is just to check on the pcie fingers we have the multimeter in diode test mode so i'm going to test from ground with the blackweed to the pcie on this side i'm interested in the third and fourth finger after the notch so one two three there's reading four but as i've settled down 5.6 5.7 normally they don't climb up on that and the fourth one interestingly the same normally they just steady i'll just check make sure my meter is connecting properly once more so for that in itself is a little bit unusual but i haven't looked at these uh 2000 series cars before so i can't say for certain that's what i have and we need to check are they both shorted to each other and the answer is no they're not shorted to each other so there's no shorts to ground or open circuits there but i find the increasing resistance slightly strange maybe you guys have seen that before the time it's completely normal on this side of the board i'm interested in two signals so the first one is on pins two and three and this is ref clock plus and minus or reference clock this is a 100 megahertz clock that synchronizes the transmission of data between the gpu and the pc so we'll have a look at these i'll just check the meter yeah it's good i'll go to the ground of the black leader let's have a look to see what's on pins two and three so here's pin two and that's reading 800 steady and pin three 800 steady again i can't say exactly what this should be they should be the same as each other and they shouldn't be shorted to each other which do not so that's those two and then the other signal you can probably see the trace here just going here again two pins pins five and six and this is the first pcie uh wayne if you like this is half of the way do the half is on the other side of the board on pins three and four so there's four pins make up one way so on this again we need to have one so we've got one two three four five now that means open circuit and pin six reads open circuit and this is actually okay and the reason it's okay is because these traps go to all these capacitors so i need to test from the other side of the capacitor but because of the capacitor isolating these pins from any voltage on here there shouldn't actually be a problem on these ones so we'll have a quick loop so we can just go i'll show you from pins one two three four pin five we'll go to the capacitor here yeah and pin six will go through the capacitor afterwards okay and then from the other side of the capacity to the ground we have that's how we expect it to read with a bleep yeah and steady yeah that's how i would expect it to be they're okay let's have a quick on the other side again so this is the side without capacitors that's more prone to damage so pin three strangely enough it does bleep now it didn't before but then it's climbing up and i find that a little bit odd yeah pin four let's have a look at the next lane so if you look along here i'm sure you can see where the lanes are so three and four have little traps going from well if i'll use the microscope it's more clear there that's a little bit clearer especially if i use the tool so you can see pins three and four have the two traces coming to them the ones that we were just looking at here okay and then you can see the next laners here these two so again let's have a look at this one and let's just see how it's reading this would not directly have connected via the mining adapter it wouldn't be going anywhere no matter which way i get the minor adapter what's on these two well i'll just get a good ground okay and that's doing the same sort of thing that's climbing up which is a little bit unusual but it's looking the same as the first layer i will say i'm sure i can just go across the floor i'm all down here here's another one doing the same sort of thing okay so i've not seen that directly before but let's now open this card up and let's have another bit of a look around it and see if all the resistances look okay if they do let's power it up and just see how far we can actually get with this well louis and alberto told me they have been watching my videos and they have looked at these cards but after a couple weeks of looking around they decided it wasn't quite for them or rather no that's not fair they decided that it takes longer than a couple weeks to learn this and i totally agree it took me with a good grounding of electrons to start off it probably took me about a year playing around before i was confident with gpu repair and with motherboard the same okay this is another one so some of the screws will be missing doesn't surprise me and we can take the board off and we can have a look so first of all there isn't really anything much in the way of thermal compound possibly they cleaned it off so we can get rid of the heat sink um there seems to be some corrosion inside here on the heatsink well this doesn't surprise me with gran canaria i mean las palmas the capital was by the sea um it's known by the way las palmas to be a the capital city with the best climate in the world and b it's also known as nomads city uh due to the number of tech startups there people living there um you actually got like you know all the infrastructure of a european country here with the prices of living in an african country basically so we have a very large cyber nomad community if you're interested search for cyber nomads las palmas yeah less polymers you might see it and you you'll find enough waffles so this is our chord let's see what we have in the way of resistance is on this so i'm on now normal resistance range and we want to have a loop so 12 volts coming in well there's no shorts on 12 volts now let's get to the 12 volts up here one side this connector is ground anyway i'm pretty sure the 12 volts is coming in onto here this current sensor resistor oh that was like another short yes we have a short on 12 volts do you see it on this current sensor resistor here so somewhere on here we have a shorted high side mosfet that's the first thing we have that probably explains that increasing reading i was i was getting so let's see if we can find out where that actually is i'll just check the 3.3 volts as well which is the fourth finger in that's okay no short there so there's two or rather large book regulars here which are probably going to be the ram phase as i'm not familiar with these cards at all yeah so 70 ohms and that looks normal for a ram phase there's two here that was reading 3.8 k right this one 12.5 is probably the pex rail that's the pcie express or it's the voltage for the memory controller but there's another one up here which might be memory controlled what was this one reading again well it looks like ram 69 so that this is almost certainly just a filter on the 12 volts input now we can check that just go and find the other side of it here yeah we can check here yeah we've seen the short here as well so this is on the 12 volts coming in ah something's broken here oh i know it's not it's just a little rubber pile it looks almost like a broken capacitor but it's not just like a little rubber pad to bounce the backing plate um okay that's all the supplies as i can see oh there's one more here i don't think i found this one before what's this one really seven kilograms that looks okay what's the v core reading so that'll be these i was reading very low but this is likely to read very low anyway see 0.2 of them shorted and 0.4 with them not shortage so that doesn't surprise me that's probably okay so what we seem to have here is a short circuit on the 12 volts coming in from here through this current sensor resistor through this is this connected to it as well yeah it's there as well but this 12 volts doesn't have a short so i'm fairly confident we say that this is powering the memory control or the pex and the ram and this is probably powering the four quills here so the short is probably going to be a high side mosfet this doesn't have individual mosfets which has these driver chips here so let's zoom down on these in fact let's use the microscope and let's see what these actually are you can see that we have a part number qa3111 [Music] so that's the part number there's some more numbers i'm assuming the first part is actually the part number and then here we have this chip yeah up9511p so i'm fairly confident this is going to be the vrm controller and these may be dual mosfets or they may be drivers they're not single mosfets because there's one by each coil so they're at least dual mosfets so let's have a look to see if we can find the data sheet for these i couldn't find a data sheet for the qa3111 but i can buy them but i did find a data sheet for the up9511 and that tells me all i need to know about these so let's have a look at this data sheet this is the data sheet for the up9511 you can see it's one to eight phase synchronous rectified book controller we have a pin out here and if i just scroll this down a little bit we can see a typical application so it shows for each of the one to eight channels you have a pulse width modulation output and a current sense input and then you have an external driver chip which drives mosfets or quite possibly the mosfets are inside the driver chip which is what we have i'm sure so we can see that each of these chips are actually drivers and the mosfets are going to be inside them because they're not separate and this is where we have a short to work out which vrm which voltage rail has the short circuit high side mosfet is actually quite simple the way you do this is to have a look first at the resistance of the short to ground that you find on the 12 volts so in this case it's 0.2 of an ohm i'm on normal ohms range now we need to go to each of the coils that we can see and again measure resistance to ground so we have one here and what we're looking for is a resistance that reads the same as the one we found here something the same as that point two so we can see this one is 7.5 kilo ohm 7.3 kilograms 3.8 k that's high 12.6 ohms or whatever one and we have two more here 69 69 and then we have these coils down here which we can get to on the bottom of the board so we that's these four and that's that's the one it's very close he's actually reading point four point four point four so that one is basically reading the same i mean there will often be a slight difference with your meter as that one okay so that's where the short is it's on these coils here yeah that's this phase down here this is where your short circuit is if we're lucky we'll find that the vrm which has the same resistance to ground as we're reading from here is one with a single phase like this one so there's just one coil and one driver there's another coil and the drive is probably over here we know they're only on a single phase because the resistance reads difference on the two of them these two up here actually both read the same so here we have two phases on one supply this is almost certainly the supply to the memory and you can see we have one driver here and we don't have one driver here resistance is 69 ohms so we know the problem isn't here the problem is here and this has four phases and this is where it gets more difficult in this case or in this case to actually find where the short circuit device is this board in fact only has four phases down here but it's common to have graphics cards and motherboards with eight phases 10 12 16 phases maybe on some very high powered ones and when you get to these sort of boards with many many phases it is important that you can determine which high side mosfet is the 41. one other thing i'll just mention the resistance of these coils is extremely low almost to zero so in actual fact it doesn't matter which side of the coil you measure to ground you'll get the same results regardless of which side you're actually reading from so don't think you need to measure from one side of the coil all the other they will both give you the same readings on the multimeter now let's have a close look at these circuits i'll draw it out and i'll show you exactly where in the circuit the short actually is this is not a short to ground this is a short from the 12 volts coming in to the voltage rail that supplies whichever component whether it's vcore vram pex memory controller supply it's a short from 12 volts to there it's not a short from that rail to ground i've just moved to the other side of the screen because i have a nasty habit of drawing underneath the insect picture where you can't see it okay so to stop that that mean i've just shifted over this then is your 12 volt supply coming in and from here we have a number of vrms each vrm is quite simple so we have two mosfets in series connecting directly to ground the two mosfets go to a controller this is your vrm controller i can't fit the whole way the vrm controller okay and the controller switches these mosfets on very rapidly one after the other okay from here the junction between the two we have a coil and on the other side of the coil we have capacitors usually lots of capacitors not just one as i've shown to ground and now coming out of here is the supply to the load whether it's the vehicle or the vram or whatever this circuit is called many people call this a book regulator but it's actually a synchronous regulator but they're sort of synonymous this is a synchronous regulator call it whichever you wish that's the correct term for it so the way this circuit works the 12 volts does not supply power to the load when this mosfet turns on the 12 volts flows in here into the coil but the coil has inductance and it will resist the flow of current and what happens is before the current can come out the other side it has to charge the magnetic field inside the coil so the all the energy coming into here is being used stored in the magnetic field that builds up and once the magnetic field is saturated that means it can no longer magnetize any more it's saturated then the current will come through into the load all the currents so at the point of saturation the 12 volts will also appear here and we don't want that to happen this is welders may be running on for memory 1.35 volts so this controller only switches the mosfet on long enough to magnetize the coil before it's fully magnetized saturated it switches off again so no more current is flowing in it will not magnetize any further what happens now is this one switches on and it effectively discharges the magnetism in the coil and it's that magnetism collapsing is what generates the voltage to charge this capacitor and supply the load and by controlling the amount of time this one has turned on in relation to the amount of time this one has turned on you can control the amount of magnetism stored in the coil and you can control the voltage there that is a simple vrm synchronous not a book okay so that is a normal single phase but on hours we have some multi-phase so we have a a two-phase supply we have a four-phase apart i'll draw the four-phase supply but obviously from this you'll understand the two-phase supply is basically the same thing so here's the drawing a bit of dickie card yeah dickie card copyright 2022 yeah okay what we have here then are the four phases this is the controller chip and each of these is the high side mosfet for each phase and they're called high side because they connect to the 12 volts and the other ones connected from the core to ground on low side four of them and the signals here would usually be called upper gate zero because we start with zero because we computer minded yeah and the lower gate zero yeah upper gate one lower gate one and same for two and same for three upgrades two lower gates two upper gate three lower gauge three and so over however many phases we have and the way this works is just like the single vrm but with added functionality and that's it has multiple phases so basically in this case the control will switch the upper gate of this one on so this draws power into here magnetizing the coil and before it's fully magnetized it switches that off and switches this on and the magnetism collapses and goes down to v core which is here yeah this is your v core your load and this is where all the capacitors are lots and lots of capacitors from there to ground once that's happened while this is still collapsing it then switches on up a gate one powers this one on and this charge is this cool with magnetism and the same thing switches off switches on and again that discharges into there and that one's for all four phases you can think of this a bit like either a four-cylinder car where each cylinder has got a different phase it has a phase so 90 degrees around the rotation 360 degrees each cylinder will fire in one order of another so each cylinder fires one after another and that's what gives the smooth rotation of the motor and it's all works fine and it all can generate more power than just one cylinder cut alone another way you can maybe think of this a bit like a gatling gun you know those especially in world war ones where you turn the handle the barrel spun around and it fires the boys out of each barrel and turn and the reason for that is basically you can get the much higher fire rate so it rotates the barrel for each bullet that's each discharge of one of these and by firing the bullet down the barrel you generate heat friction and this effectively generates electric friction with the resistance of the coil so they get hot yeah the mosfets go up get hot their clothes get hot and he fires them in turn or in some sequence which he uses so that each one is effectively handling a quarter of the load it's only fiving if this thing was running uh 200 000 times a second each one of these only one at 50 000 times each and it shares the load so that's what this is doing this business with the upper gate and the lower gate is called the drive to the mosfets so here we have a pair of mosfets high side low side okay and each one has to be fired and turn this will be controlled by a pulse with modulated chip pwm and this basically sends out of it four sets of pulses so phase zero phase one virtual on pwm zero pwm one pwm two pwm three so this is sending a pulse on each one and turn yeah well this pulse has to be converted into this switch this on then switch this on switch this on and switch this on so normally the pulse width modulator the four of these go to the controller which is off my piece of paper so i can't show you exactly because it wasn't a very good it was a dodgy cardio i was a dodgy cat drawing we can see that this is just effectively sending a pulse and this chip is controlling the switching between the two for each time it's post yeah so you often find motherboards and graphics cards where you have a pulse width modulator then you have a driver chip one driver chip for each pair of mosfets so four driver chips eight mosfets one pulse width modulator you'll find that just to add on to this i will just show you normally there's another mosfet here for each of the low sides so normally each low side mosfet has two of them and the gate of this one is connected to the gate of this one so this switches on alone to charge the magnetic field and these two in parallel switch on together to discharge it normally this is one type of mosfet normally these two are a different type of both the same so now in actual fact what you'll normally find is one pulse width modulated four drivers one for each set of mosfets and three mosfets in each phase that's a very very common topology that you'll find to make things more integrated and smaller everything's getting smaller these days and more modern graphics cards and motherboards you tend to find that you have the pulse width modulator with however many phases and then the drivers are built in with the mosfets so each of these pairs of mosfets is replaced by one chip this is the chip inside the chip you have the two power mosfets connecting to ground connecting to 12 volts inside here then you have some logic circuitry this logic circuitry will have a power supply often 5 volts or 3.3 volts but it could be any suppliers on the board it could be 12. but often it's 5 or 3.3 and the logic has coming into it pwm so in the case of the first phase pwm zero and then we have another one of these driver chips just the same thing inside exactly same as this one and coming into here will be p w and one and two and three so you have your four of these chips this effectively replaces the mosfets and the driver all in one chip and these are basically called drivers or vrm drivers coming out of this chip there will be a connection to the coil let's go into the load so this is what we're generally faced with when we're trying to diagnose the problems with these circuits now we know what the circuit looks like i'll show you where the shortest so the short is in one of these mosfets say it was in this one so what happens is this mosfet internally goes short circuit from the drain to the source and we've now got a connection from the 12 volts directly through the mosfet to the coil so when you power up all 12 volts comes down from here in the first few microseconds it will magnetize the coil as soon as the coil is now fully magnetized this can't turn off so the current now goes down here into these capacitors into the load and this will draw a massive current if there's a fuse in the 12 volts which they're often is on some on higher end graphics cars but not always if there's a fuse the fuse will blow if there isn't a fuse and you've got a good power supply atx power supply that's monitoring the current then the atx power supply will shut down so you'll find your pc or switch off and this is gonna switch back off again and it often won't come back on unless you unplug the mains and do it again which by the way is not a good idea if it's shut down the first time we did it for a reason every time you switch the atx on you'll send another boost of 12 volts down through here more stress on the load which is probably the gpu core or the cpu core so that's the second thing if you have a graphics card where they kind of cut corners and didn't fit fuses and you have an atx power supply that doesn't monitor the current very well what will tend to happen there is this mosfet he will set on fire burning a hole in the pcb possibly the inductor will actually inductor won't melt but tracks can burn out all this current is going to cause damage if you're really unlucky the graphics card will set on fire so that's what you need to watch for and this is where your shorts are so you now know the short is here into this coil short circuit high side mosfet this is why when we measure from the 12 volts and i'll just refer back to the previous diagram when you're measuring from the 12 volts the resistance from here that is your is your test meter i always draw an analog meter it could be a digital music when you measure to ground from the 12 volts the resistance you see is the resistance of this load because if there's a short here the current is flowing through the short i mean when i say good your resistance meter injects a small voltage to measure the resistance and it measures the resistance by working out how much current is flowing so the test meter does send comment into the circuit the current flows through the short circuit mosfet through the coil charge up these capacitors and you see the resistance of the load it's exactly the same thing as if you connected your test meter and would make sense if you connect one end to the coil yeah and you're connecting the other end to ground looking i've drawn the meter backwards yeah but sometimes you'll either you'll have to use your your digital multimeters backwards yeah red and black wires so measuring here is the same as measuring here this is why the resistance you see from the 12 volts to ground will match the resistance of whichever vrm has the shortest mosfet and with a single phase vrm it's very simple to find up shorted mosfet or driver in the case of this it's still the same it's just that the mosfets are no longer separate components the inside this driver chip the shorter is there okay capacitors the same thing applies when you measure from 12 volts you will see the resistance of the load and this is where the problem starts diagnosing faults in these circuits which mosfet is short and the reason is a problem is this you're measuring from the 12 volts to ground the current's going down through one of these okay if you then measure the resistance from here to ground you'll actually measure the resistance of the load as i've just explained to you if you measure from here to ground or measure the resistance of the load if you measure from the top of this mosfet it's the same thing as measuring from the top of this mosfet it doesn't matter if your meter paper is put here or it's put here or it's put here always put here you are still measuring the same resistance so you need a method to find out which of these high side mosfets is short and that's what i'm going to show you in the next step of this video i will just show you a few physical examples of what i've just explained on paper so this is graphics card has uh six phases here six coils and these have driver chips so there's one square chip sitting next to the coil and this is a driver chip okay above it we have another this is a single phase and this has high side on low side mosfets this actually has two high side and two low sides which is a little bit unusual it's more common to find ones which have one high side and two websites in fact i'll just show you that now so here is another graphics card this one's a 800 gt a bit older and and you can see with this one we have two coils two phases one high side mosfet and two low side mosfets one and two you can see that the two low side mosfets the same target each other four eight three five n and the single hard side mosfet is a different type for a four one n this is very very common we can see another single phase here and again three mosfets the easiest way to find which is the low side of the two that are the same four eight three five ends again and this would be a yeah the 441 yeah yeah 4841 so that's very typical and if we take a closer look at this one this actually has just to prove the rule of thumb i've just told you is not always correct this has four mosfets four c one zero n and if we look quickly you'll see that the top two are connected in parallel so these are one side of them and this is the other side draining source as you the source and they connected together and then the source of these two will go to the drain of these two if i'm sure you can see the large circuit tracer comes down here okay and then the other two ends will connect to each other as well so we have two pairs in parallel here's an example of a driver vrm driver 5 ir3553 m one drive it on each phase and i'll just show you the data sheet and you will see that each of these has two built-in mosfets this is a typical application diagram for the ir3553 the one we were just looking at there are many many other drivers you will find on vrms on motherboards graphics cards but they all work basically the same as this one so we have v in this is your 12 volt supply going to the high side mosfet which is integrated inside this chip this then connects to the low side mosfet again integrated and it connects to ground from the junction of the two we have a connection to the coil which is then driving the load v out in our case it was v core and these are the capacitors which charge up on that rate to smooth the pulses and give us steady supply you can see we have a signal here pwm pulse width modulation coming in this effectively controls the duty cycle of these two mosfets therefore generating the required voltage through the coil there's a few more signals but these are basically to do with enable signals and current monitoring and such there's nothing that we need to really worry about too much in this tutorial we will also find on the board the portsmouth modulator controller which in this case is this i see so this is sending pulses each of the drivers in turn and then the driver is controlling the on off switching between the pair of mosfets on that synchronous regulator and finally this is the rtx 2016 the one which has a short circuit high side mosfet this again uses driver chips port numbers qa3111 and i've not been able to find the data sheet for this particular chip but i can find some information on the driver the pulse width modulator which is here and this up9511p just has four pulls with modulation outputs which tells me that each of these contains the driver and the mosfets integrated so our short circuit high side mosfet is in one of these four driver chips but which one so let's have a look now at 10 different ways we can determine which is the faulty one and let's see if each method actually works here is my fluke multimeter it's a very good multimeter this will read resistance to a tenth of an ohm so if i short these two together we can see we've got 0.2 ohms let's see if this meter can tell us which of these phases has a short circuit high side mosfet so i'll go from 12 volts coming in to each of the coils that's really not 0.2 this one is written in 0.2 this one is reading 0.2 and this one is reading 0.2 so they all be exactly the same and it doesn't matter which side of the coil i go to either they all read exactly the same so that multimeter is not going to tell me which of the coils has a short circuit high side mosfet which phase but this multimeter can only read to 0.1 of an oh i have another multimeter which will be to a hundredth of an hour point zero one so let's have a look to see if that can tell us which phase has a short circuit high side mosfet this is the an end a an8008 and this will read the resistance down to a 100 of an ohm i'll just short these together and you can see it's reading 0.22.21 okay 0.20 basically with them shorted so let's try this one so we'll go from the 12 volts coming in here to each of these coils again and let's see if we can determine which has the short circuit high side mosfet so that's reading point zero two one one nine two zero basically the same as always shorted together that's really the same that's really the same that's really the same so even with a multimeter leads to a 100th of an ohm we cannot determine which high side mosfet is short circuit so with the low ohms meter i'm going to measure the resistance from 12 volts to each of these coils the idea being that the driver which has the short circuit high side that's closest to the coil will give the lowest resistance so this one's reading 0.2 0.027 that's on the 2 ohms range let me get to a more sensitive range 28.7 micro ohms what's this one really um 29.8 not much difference just one medium 30 what's this one really 301 so it seems that this first one 28.3 is actually reading lower than all of the other ones how much resistance is this oh yeah the coils are reading very very low 0.029 it has very very light resistance but it looks like this one is the one that has the short to 12 volts the one nearest the bottom of the card to use the esr meter this relies on the fact this is 100 kilohertz which this uses it will see the coil as a resistance so let's have a look at what we have so i'm going effectively all this end of the calls are connected together i'm going to the end of each one and see where the lowest resistance is 0.14.16 do you see it so i'm measuring across the coil yeah and when it comes to the one with the short i'll see the way we so 0.164 0.15.13 yeah 0.12 0.003 so that is also showing me that this is the one that has the short on it so we have two methods one using the short circuit finder one using the esr meter that both points to this chip being the one that has the short circuit but it's possible although the multimeter cannot see the difference in the resistance here it might be able to tell us which is the faulty one if we look at the drive signal coming into it from from the controller so let's have a look so this is where the the pulse width modulation drive comes in i've tracked it from the chip so that's really 8.6 k that's really 8.46 k 8.57 k and this one's reading 3.5 ohms method 3 40 chip okay there's three ways to find it about one month ago i published a very very popular video and this video described how to convert a standard multimeter into a very very sensitive short circuit finder we tested the short circuit finder on the video and it worked very very well i then went on to publish a second video a full a week or two later that dispelled some concerns that people had that this would damage circuitry that runs on oil voltages such as gpu core cpu core and so like and we did some tests proper tests on graphics cards and proved that it wouldn't damage anything at all so i've now reconstructed my short circuit finally let's see if this is actually sensitive enough to determine which of these four phases has the short circuit high side mosfets this is set to volts range and you need a multimeter that is quite sensitive so this reads down to 100th of a volt we're now going to connect the positives to the 12 volts coming in i'm going to try on each of these coils to see what reading i get so this one is giving me point zero zero zero one this is giving me yeah the same this is giving me the same so he's basically saying this is a dead short does this one read any difference no so in this case the short circuit finder is not going to tell us which one has the short i was just about to give up on this method but i've realized i may be able to increase the sensitivity of this further so i'm going to go to the millivolts range on here which is now showing overload because the voltage reading is so low this may now help me so i can now see 11.6 millivolts on this one next coil 11.6 basically very similar this one 11.9 slightly higher and much higher ah okay yeah so this is definitely really much lower on this one okay we can see seven millivolts yeah 11 millimeters another very common method of short circuit finding that people use is voltage injection so via this method we inject a voltage into the short circuit to see what component gets hot now we know that all of the current is flowing through this mosfet the faulty one we notice this one we've already determined that via well a few methods now but let's see whether this method is going to work so in this case we need to inject 12 volts into here connect to ground that's what we normally do 12 volts to ground and see what gets hot i have my bench power supply now i need to set the voltage on this to be lower than the supply voltage for the gpu that's the only way to be safe it needs to be aware and then i can set the current to flow whichever i wish so i can now effectively short the two leads together so we have the two leads we short them together and i can now dial up whatever current i want so i can get into this maximum just make sure the weeds are sitting properly i can get five amps okay i can get five amps into the load you would think that was probably plenty enough to warm something up so let's connect this to our board and let's see what happens just to prove i have the polarity right i had the polarity right previously i'll just connect an 8-way connector into the plug that it fits this way and you can see that the yellow is on the top so this is your 12 volts which goes to the inside end here which is where i was measuring on the previous tests so we know that's right so i can connect the 12 volts to here and the ground can go to ground somewhere on the board so i'm just going to use uh i'll try the metal bracket first see if i can actually get a good enough connection onto it no i don't get a good connection they're only one point something i'm so let me find a better ground point than that one with the weed shorted together you see i can get about five amps but this isn't a dead short it's about 0.2 of an ohm which is the resistance of the gpu so if i connect my power supply to here i'm seeing about 0.8 of a volt that's 5 that's 5 for this gpu and i have about 1.6 amps 1.7 now is the faulty part getting warm and the answer is no it isn't i'll just leave it on a little bit let's get a little bit of time to see whether this actually gets warm the gpu isn't particularly warm either that's cold really probably the reason it's wandering around so it was probably a slightly bad connection on the end of the wire here okay any very slight difference in the resistance of the connection or where i'm connecting to the ground or here will obviously cause a slight fluctuation like as well be a little while nervous not getting more now you might says yes using the infrared camera i don't have an infrared tamper but i do have a temperature meter which i have a thermocouple and i can certainly use this to see whether the short circuit device isn't any warmer than the next one so this is the short circuit device this one and that's reading twenty seven point six twenty seven point seven what's the other one really it's actually it's warmer twenty nine point eight twenty nine it's really the same 28.9 28.5 28.4 okay 28.2 and that's reading 28.1 the temperature's just dropping a little bit that's probably because the thermocouple was in contact with something warmer doors that one read just a little bit higher 28.7 28.2 20.7 i won't see if i can actually get a definitive answer to this will it see the difference in the temperature 20.5 i think we have to say no this this is not not generating enough heat to give any definitive answer sorry guys but that method is a note this is a slightly different variation on the method i just used so in this case we're injecting current again into the short from the 12 volts and what we're going to do now is try to determine which of these coils the current is flowing through the idea being that the one which has the shore which is passing current through the coil will read a voltage across it so we can go to the millivolts range on here and we can actually have a look so if there's no current flowing through the coil and the coil has some resistance all be very low we shouldn't see any voltage drop across it so that was reading zero this one's reading zero this one is reading zero does this one show a voltage drop yes it does so we know the current is flowing through this coil so this method has determined which coil which phase has a short circuit high side mosfet this method is a little bit more esoteric we could say what i'm going to try now is c if i can find the short circuit phase by ring testing the coils the idea of this is that the phase that has the short circuit on one end of the coil is going to vehicle and there's lots and lots of capacitors and because we have a short circuit the other end of the coil is going to the 12 volts where there are also what sorts of capacitors to ground off this rail now if i even effectively pulse the current through here which i'm going to do very simply by just rapidly just touching it touching the crocodile clip against the 12 volts if i do this and when i put an oscilloscope on this end of the coil i'm expecting that the one which has the short circuit i'll just effectively just see the voltage going on and off but on the other ones this end of the coils effectively open circuit because both these mosfets are off and i'm expecting to see like some sort of spike or pulse here a ringing effect on the oscilloscope on all the good phases and not on the path i have no idea if this is going to work but the theory seems good so let's try it and let's see if we can determine which phase has a short circuit high side mosfet by ring testing the coil so with this method i have my digital oscilloscope set up there i'm going to put the scope probe on the coil so i'll start with this one and i'm now going to touch the power in instance to the the 12 volts so the uh power supply is set to 0.8 of a volt maximum current so it can't do any harms or anything so i'm on the first one clue well let's just very briefly just touch the connection yeah that's what we'll see what's the on this telescope you see i see like a pulse with like a lot of ziggler's ugly stuff at the top of it that's a technical term by the way zigglery's ugly stuff let's just turn up one more sensitivity that one probably went off the screen yeah big pulse with ziggly zagly stuff on the top and what i'm interested to know is is the one with the short behaving differently i'll just move it down a little bit so we can actually get on the screen a bit better than we good yeah let's go to the next one pulsing with ziggler exactly stuff does this won't make any difference and i'd have to say no it doesn't so this interesting method doesn't actually appear to help okay so that one was a failure here's another idea which is more fun than use but you know let's have some fun so i have here just an inductor coil connected to a scope probe and you'll see my oscilloscope with that on the screen so i'm going to hold this coil close to the other one and touching it and see when that magnetizes the coil because the current's flowing through this one this has a short do i see anything induced into this one so we'll just uh touch against the 12 volts i have my scoop on the most sensitive settings and this doesn't work okay this doesn't do anything i do know that these casings on these are designed to keep the magnetic field inside so i'm not totally surprised just for that reason it doesn't work but let's sort of go underneath because just other side of the board it's there and we might see something from this side all the way down to it let's just try it and there we go i'm just touching it but nothing's happening so that method although it was interesting didn't do anything maybe with a sensitive enough pickup but this doesn't work many people ask in the comments to my videos on short circuit finding if i remove the coils one at a time until the short goes away will that tell me which mosfet assures and the answer to that is yes clearly that will because we can see on the diagram the explanation that the short is feeding current through a coil one of them now if you remove that coil and break the circuit then yes that will tell you which high side mosfet is yours however if you've never tried or if you have tried to remove these coils this is not easy to do i have tried and it's very difficult to get off i mean i've taken off for salvage purposes in the past but these are not easy to desolder they really aren't there's a lot of heavy planes in here i think the best answer to the question can i remove the coils is if you've got the equipment good enough to remove these so we're talking a preheated a good preheater strong hot air something like a quick 861dw if you've got that sort of kit to do that then you have no excuse not to own the other kit that'll do it a lot easier for you okay i think that's enough said about that one and in a similar vein can i remove the mosfets until the shot goes away yes you can but as pointed out with removing the coils they're not that easy to remove a lot of these drivers which are not mosfets will have connections underneath and they're not that easy to replace either so once again if you've got the equipment good enough to remove these which is going to be a pre-heater it's something cheap like the little card eight five three b he pulls out air yes you can do it but come on guys i've just shown you there's much easier ways to do it than that having said that now we know this is the short circuit device we're going to have to remove this because it needs replacing so i'll show you now how i actually go about removing one of these the first trick about removing these is to use the preheater which i'm going to use so we need to get a lot of heat in from under the board such that when we come to use the hot air on the top we don't need to heat it up so much bear in mind there's a lot of capacitors around here and if you get too much heat into these they'll bulge in which case they'll fail or will go back and make a mess and don't smell very nice so the first thing i'm going to do is just cover over these a little bit of caps on tape aluminium foil is also quite good for this so i'm going to try at least to protect them from the heat i'm also going to come in with the heat from this direction as best i can so so i'm trying to keep the heat away from these a little bit having said that i may still end up having to remove at least these ones at the end um the second point as i just made then is to use a preheater so this is mine you've seen this before what i've called it eight 853b and this is a hot air preheater you can see it just pushes hot air from underneath the pcb so get the board onto this and then we'll get this thing removed i often desolder components by adding some wedded solder to the pins first to work with the melting points but in the case of these these there are pads underneath the device so that really isn't going to help you what i would still suggest you do is add a little bit of forks so we've got a little bit of flux around here now these have the buy the driver some small surface mount components normally i take a photograph before working so that so that i can refer back to that afterwards if i've knocked something off the board or just load something i can see where it goes but with these because they're all the same and it repeats the same circuit the components next to each one is the same as the components next to the other ones so if i do dislodge something in this case i can easily work out what it was now let's switch on the preheater so it's on and it's on pitch i've turned the dial up there's no temperature setting such on this as it does the dial which goes from cold to hot so there's a little bit of gas work but i'm going to warm this up i'm not going to use a thermocouple because i'm used to working with this equipment but a thermocouple is a good idea if you're not and you want to try and get us up to about 120 degrees or something similar to that that's been on for a couple of minutes so it's warming up quite nicely now i'm now going to turn the preheater up to full i'm going to try to do this right handed which is a bit awkward because the brackets here that might actually burn the back of my wrist yeah i can get it from this direction let's get the hot air so hot air on 450 degrees maximum airflow using the quick 861tw and i'm now going to try and remove this oh well that's taking a lot of heat and it's actually overheating the capacities and i still can't dissolve it i actually got this unsoldered before that will give up so i'm gonna have to cool this down again put it onto low and remove these capacitors before i'll come back in and have another go to change this as you can see and as i suspected it just biscuit pasta just bulged yeah the bad smell came from like that was the first thing i knew he was a bad smell and i thought he's going to go bang in the minis um you'd probably see on that one the bottom of it's all bulged outwards so i've now removed all three capacitors and this one's not quite so bad and we can try again and you can see how much heat it's taking to get into this see that one's gone on the base of its bulge that was so this should answer your question well can i just really remove the coils or the mosfets until the short goes away and the answer is yes but do you really want to okay so we'll get over here again let's have another go with removing this driver chip those were the ten different methods i could think of to find to locate a short circuit high side mosfet in the multi-phase vrm i'll just go them through them briefly a little resume and then we can all have a thoughts in the comments below very interesting to know what you guys would think about that and also any favorite methods i use and anything else you can think of that i didn't think of so method number one measuring resistance from each coil to 12 volts with a multimeter doesn't work method number two measuring from each coil to 12 volts with a millimeter it worked method number three using the esr meter i think was a dramatic success and i think actually some of you guys might have been quite surprised by how well that method works method for measuring uh the resistance from the gate to ground if the mosfets yeah this will work pretty much every time but it does depend on the internal damage to the short-circuit mosfet in the case of drivers like these it works dramatically well on this board but i wouldn't guarantee it would work on every board it depends on what's happened inside the driver chip method five using the one dollar short circuit finder i think that was quite an impressive success again considering how cheap that thing was to build and i will link the video again at the end credits of this one you'll see a link to that video voltage injection to see what gets hot i know a lot of you were very strong uh proponents of this one but it didn't work nothing got hot not even using the thermocouple which could make the 0.1 of a degree could i say that that one driver which at the shop was getting any hotter than the other ones voltage injection with a millivolt meter to measure the voltage drop across each coil i think that was very very impressive the results we saw on that one and i hope you guys would like to agree with that um the more esoteric ones pulsing the voltage and seeing if we can make the coil ring on the ones that open circuit at one end basically that didn't do anything uh same with using effectively inductance coil and the oscilloscope to see if we could induce some comments from the one that had the short so you're passing coming through that core you can induce into another and see some voltage possibly would work but certainly not with the equipment i was using remove the coils that ain't easy guys i mean you've just seen the end of that video i i i could get the coils to be soldered removing the mosfets wow i mean that was even more difficult in the case of this board i need to get that mosfet off because it's short and yet it appears to have welded itself to the board even the inductor that was close to it wasn't dissolved with it so removing the mosfets yeah but come on guys there's got to be much easier ways to do it than that in fact i've shown you much easier ways to do that no yeah okay so um to end that one i hope you enjoyed that and i sincerely hope that my first comments that no matter whether you're a beginner whether you're intermediary or whether you are experienced at this sort of work i'm sure you've learned something today i'll see you all very soon on the next word electronics repair video ciao for now guys

Channel: Learn Electronics Repair

Views: 75,179

Keywords: electronics repair, electronic repair, lessons, course, training, free, gpu repair, gpu not detected, gpu not found, graphics card not detected, ebay gpu, school, BLOWN FUSE, short circuit mosfet, short circuit vrm, short tracing, short finding, vrm repair, motherboard repair, graphics card repair, short tracing technique

Id: WMhpKR9x7Lg

Length: 75min 8sec (4508 seconds)

Published: Sun Apr 03 2022