Gadgetory

although we'll probably have a fire extinguisher and the thumbnail will refrain from the bomb emoji and the hashtag FBI involved hashtag fire aspect to the title what we're doing today is looking at basically how to blow up your motherboard vrm so the point of this is to illustrate that it's actually depending on board and what you're doing an overclocking it can be quite easy to exceed or nearly exceed the safe temperatures for vrm components on a motherboard in this example we're using a board with no heat sinks on the VRMs and we're also overclocking it as high as we'll be able to so this is an example of where a user who has just seen overclocking done online on a particular CPU or product gets the idea that hey it's not hard to overclock this thing and they go about doing it without necessarily thinking out the consequences and when one should overclock to what frequency and voltage before that this video is brought to you by the new cable mod pro series cable mods a new pro cables come with pre-installed closed combs for clean builds accompanied by a revamp to color style and vibrance the cables are now using thicker wires and they've also added right-angled internal USB 3 extension and right angled SATA data cables by the pro series cable kits at the link below or customize it your cable set with the configurator also linked below so this should be pretty fun it's going to be partly a what happens if anything when you severely overheat the vrm components we've done MOSFET testing in depth like this in the past with the EVGA ACX cards but now we're revisiting it on a motherboard that has no heat sinks and what we're doing is measuring the SOC vrm component temperatures while we do some live overclocking I have two thermocouples set up right now on MOSFETs and they're just going into a thermocouple reader which right now is reading fairly high at thirty five ish degrees Celsius for each one room temperature is about twenty twenty to twenty one right now degrees Celsius and part of what we're testing here is also what happens when you have no airflow so this is a worst case scenario we have more of this testing coming up inside an actual case as well if you're interested but if you think of a case like maybe a bit Phoenix ENSO which I face the zero airflow unless you create airflow manually up in the top of the case has knows no airflow really so you have a case like that maybe you put a closed-loop liquid cooler on it and the radiator is mounted in the front of the case so it's not even close to the V RMS this is a real scenario that can happen there is no airflow and also there are no heat sinks on the VRMs so right now just ambient we are 10 degrees above ambient just by the V RMS running sustaining windows with nothing going on and with no case even but it has no air flow over it so this is our scenario we've setup it is a worst case scenario to show again what happens when you push things too far and what is too far to push the V our own components all right so right now we're just gonna set up some basic overclocks and step through them and watch as the temperatures rise just for fun this is all really kind of for fun right now it's not necessarily like a guide or anything so I'm doing here is I've set a really super achievable clock 1500 megahertz we can go higher than that and 1.15 graphics voltage I've pushed this particular chip up to 1,700 sort of but it's stable at 1650 on a different motherboard so we're gonna see how high we get on this board and I'm just gonna run fire a strike loop in while it sorts itself out and we'll look at temperatures as we go and keep pushing it so I'm just gonna push this like a user would when the user is trying to see what kind of clocks they can achieve because a lot of people don't really know that the V RMS exist or the MOSFETs exist or that they can overheat or what's a safe temperature for a MOSFET so I mean I was there a couple years ago so this is going to sort of show that even if your CPU temperature is in check you might have other things to worry about so right now we're running at one point one ish for the SOC voltage and up to one point yeah about one point one and that's got our MOSFETs at 53 degrees and rising if we if we show that so here's our temperatures this one is going to this one's going to the MOSFET closer to me on the SOC VRM and this smaller number down here in the corner that you may not be able to see is going to a MOSFET that's I've determined as the hottest one that's closest to the memory also part of the SOC vrm SOC drives the APU graphics so APU graphics voltage is SOC voltage more or less different motherboards tweeted different ways the amasai boards don't really do anything currently when you put an AP graphics voltage you need to use SOC for that case but the point is both APU graphics voltage and SOC voltage go through the SOC PRM so by overclocking the apu we are pushing everything through with SOC PRM and the same would be true for memory on your cpu so right now we're already at 65 degrees and we really haven't done anything and it's still climbing so we're not thermal runaway or anything but we're definitely getting there and just while this Rises is to give you an idea of what kind of safe temperatures are for VR Emmas it depends on which component you're talking about it depends on what component they've used specifically so when I say which component you're talking about I mean choke capacitor MOSFET just crash we'll get back to it MOSFET capacitor choke and then after that which version of each one's a which which specific model of MOSFET so generally speaking kind of safe guidance would be for capacitors you don't really want to exceed about 105 degrees Celsius a lot of the lower end capacitors like on these boards are often rated for a 105 degrees Celsius at 5000 hours so and the higher you go over that you will have lifespans aggradation so if we're at 105 C on a cap and let's say we run at 24/7 rendering or mining or whatever something that you can do 24/7 by the time we've hit about 5,000 hours you can start expecting the ends of pop and that will depend on the capacitor some some are at 10,000 hours and so forth but a lot of be 350 boards use the cheaper capacitors so this crashed which means as I use or you'd say okay time to increase the voltage a bit so let's just go ahead and do that we'll push to one point one seven five and let's start it again and we'll also have to make sure it's actually doing something different yeah there we go so as you'll see in our killing your IMC video which is coming up this voltage number is not necessarily what's actually going through the socket it's just the closest estimate we can get through hardware info after accounting for things like resistance between the pin and the pad and the distance traveled you might actually have a higher voltage or a lower voltage in reality than what's being shown in software but that's board dependent and something we'll talk about separately load line calibration also comes into play we're currently on a high load line calibration so not quite extreme but high will help us stabilize the voltage provision a bit all right so we're coming back to it I've let it sit for about a minute and we're at 73 degrees on the main MOSFET again no air flow on these that's kind of the whole key factors we know there's no air flow that's intentional if you have this board in a crappy case see all of 2017 then it's entirely possibly of no air flow we're climbing pretty fast at 75 so we've gained five degrees since I started talking my AP graphics voltage input is one point one seven five one point one SOC voltage and we're at 1.15 ish right now for the current voltage or I should say the present voltage means confusion it's bouncing around a bit and either way we're a bit more stable than last time we're still only at 1500 megahertz so much.we stabilize there we'll go a little further but first we need to see how hot the MOSFETs get so 77 degrees right now is where we're at so 77 see the other MOSFET which is closer to me or sorry let's see which one is this this is the okay so the other MOSFET down here is reading 76 right now that would be the one closer to the memory and then which is blocking a lot of the air flow and 79 degrees is the one that's closer to me those are both SOC MOSFETs and the liquid cooler is doing actually pretty well right now we don't need to pan the camera over there but I'll check what the CPU temperature TDI is only 42 degrees which is completely acceptable so TDI is totally fine we're not even overclocking the CPU by the way I'm just messing with graphics so this would be a lot worse with CPU overclocking or a cpu workload even but yeah eighty degrees now so let's talk about saves time we're talking about 105 C for low in capacitors and being at about five thousand hours of life and with chokes you can get into one hundred twenty five hundred fifty degrees territory with MOSFETs about the same power state as MOSFETs also looking at one twenty five really beyond there you start D rating kind of hard and eventually you have some kind of failure depending on which specific MOSFET it is so the big concern with these components is as a user if you're not aware these things exist which is totally fair or you're not aware that they need to be really decently cool if you're overclocking they don't give you signs of failure it's not like you start seeing artifacting if your MOSFETs are too hot you'll just keep running up the temperatures and eventually something fails now hopefully there's a and s VI D connection somewhere between the CPU and the component that says at some point hey I'm 110 degrees now this is still within spec but we should throttle back a little bit to prevent a runaway thermal scenario that happens on a lot of boards if you've disabled SPID or it just doesn't have that kind of protection there's no over temperature protection on some MOSFETs some VRMs then you could definitely run into a scenario where it runs up until something fails and that will typically be in a puff of smoke won't probably won't start a house fire but it'll certainly kill at least that component on the motherboard 83 degrees now and we might be kind of hitting steady-state at this point and how much power is going through here let's see how much power is going through so I'm just going to current clamp it zero it out current clamp which is directional so I need to clamp it that way 5.7 apps right now 5.8 5.7 5.8 let's call it five point eight so five point eight by twelve point three we're watching about 71 Watts right now down the EPS 12-volt cables with our current configuration and we're stable so at this point again not monitoring temperature a user might say you know this is pretty good let's see if we can push up to 15 50 megahertz and get that stable I've seen all these reviewers GN included hit 1600 or 1650 maybe I can do that so let's give that a shot I just heard the heard the coil line stopped on the motherboard which indicates a crash so there's actually some slight coil wine from it running the settings we have and it stopped and now we've crashed so we need to increase the voltage which is the first step to going into an out-of-control scenario alright so we're back let's uh let's just push it up to one point to from one point one seven five we're kind of getting to the top end of what's comfortable here and we'll talk about the safe voltages for SOC and a separate content piece because there are definitely unsafe voltages especially depending on motherboard and LLC and everything LLC tables but that'll be a separate piece more or less branded as how to kill your IMC or how to kill your integrated memory controller so keep an eye out for that but anyway pushing up to 1.2 ap graphics voltage which on this particular board does seem to actually do something so we took some time to stabilize the slightly higher overclock 1550 megahertz and 1.2 for the AP graphics voltage 1.15 SOC which hardware info not not a DMM but harder infos times one just crashed okay so there we go this time it didn't blue screen so we're getting closer it's just you know what let's let's increase this you really should not go beyond like 1.2 for an for SOC voltage you might hear that 1.3 is okay like AMD has said that I believe so the thin is there's a difference between typing 1.3 here and getting 1.3 into the CPU and that difference on boards that are more aggressive on voltage could be deadly so if you type in 1.3 and you're sending 1.4 into it that's very bad but you know let's let's just increase SOC voltage a bit talk it's all the same honestly let's do that see if we can stabilize the go into BIOS I feel like it's actually in BIOS right now if I know how you can tell you NIT f10 and then I could hit enter yeah it's in BIOS it's going to BIOS it's not sending display out you just throw it back over there literally throw it so you'll notice in a minute that our monitor changed that's because the piece of pixie or monitor that I have which I really really dislike for a lot of reasons that monitor wasn't giving us display in BIOS and also they have scaler issues so after wasting like an hour of our time tonight every place the monitor and that I do not recommend that particular pixel monitor although I'm sure they make one or two things that are actually good that is not one of them so let's get back to doing things the way I actually wanted to do BIOS now that it works what okay that's CPU polish okay now we're pushing high voltages this is kind of get into danger territory especially on this motherboard so let's pull back a little bit this because I made adjustments in BIOS now let's just see if like one point two four is stable something like that should be it's getting awfully close to the limits of what we would recommend for SOC voltage I'd recommend if I possible stop in 1.2 1.25 somewhere in there so we are pushing 1600 megahertz now we're gonna wait a minute for the MOSFETs to warm up but I do mean almost literally one minute because that's how long it will take to get to ninety degrees we're at 82 already 83 this isn't the most scientific way to take a measurement it's just an infrared like it's just a laser so problem with this is things like emissivity and reflection we've talked about in the past but I'm just measuring something that's a blackbody component so that's actually fine for this thermos or our chokes which I'm not actively measuring all right like 80 our MOSFETs there we go 94 degrees that's what we want to see 95 is climbing okay there's a ha well we'll get to that later maybe okay so to recap we are beginning to enter thermal runaway scenario we're getting close a reminder of what we're doing here there's no air flow on the MOSFET they have no heat sink and even if they had a bad one it would be irrelevant because even with a bad garbage heat sink if there's no air flow over the heat sink then it doesn't matter because if if all you're doing is soaking the heat but you can't get rid of it all it's doing is like delaying how long it takes to get into a dangerous territory 99 degrees now so again we'll see these things might throttle some of them throttle it like 110 what happens frequency dials back voltage drops and it'll spike both of those numbers will spike up and down as they throttle the control I think I've pushed these to 120 already pre-film and then I stopped or actually it was technically 117 so we're gonna see if we can go a little higher than that this time and cause like a catastrophic failure we are stable at 1600 megahertz I was sent degrees not there yet going for it and this one with an SOC voltage of about one point two three so again this is a completely possible scenario where someone's like I'm stable at 1600 and you won't see any signs of it being unhealthy for the vrm to run at this temperature until something goes terribly wrong and if you push the SOC voltage too high the opposite problem occurs where if you push out the C voltage too high you can just slowly degrade the IMC to a point where you'll have to dial back your overclock or increase your voltage to maintain the same overclock and worst case scenario you might kill it if you really really push like like say one point four volts you kind of degraded over time for SOC voltage and not not that much time either one point five kills things but yeah we're pretty stable now 104 degrees not even really trying I mean this is all completely reasonable and so about the motherboard part of this blame kind of goes it's hard to blame Asus here too much because this is not a board that they're marketing at overclocking and this just gets into territory where like if you're buying this kind of motherboard and you're overclocking it's kind of falling onto your own shoulders there's a reason that warranties are void with with overclocking often so I don't know that I really fully blame a soos because not having erm heatsink certainly I would prefer that they have erm heat sinks but you could run this processor stock and the vrm is not gonna overheat so if their goal is to provide a cheap motherboard that you can plug a CPU into and let it go they've done that obviously they're not marketing it for overclocking and if they were that would be a serious problem because it's not good for that with no vrm heatsink but we are at 107 degrees and climbing quite rapidly you're starting to get into like runaway scenario with the I don't I need like any like a black sticker on one of the caps actually there's a black corner on it so all the other black body components around the - I've probes are slightly lower because pretty determined that those were the hot spots 107 108 let's push this up to 16 50 megahertz and see if we can get stability on it overclocking only goes into increments of 50 megahertz by the way so we crashed so that means I need to increase the voltage a bit that's a 1.3 okay so here you go this is where again depend on which vendor you ask and we'll talk about this a lot in our other killing your IMC video some vendors will tell you one point three is okay one point three once again to repeat if we're saying one point three is okay it's important to know is that when I type in the number one point three or when I measure the back of the socket with a multimeter and see one point three which one of those is acceptable is probably the latter so either way hardware info when we're actually running an application thinks about one point two eight after resistances and things we might be a bit higher than that and I've configured it to one point one SOC well no actually it's one point two this is wrong BIOS has it at one point two and a bit over one point two for AP graphics voltage will see that stable and this is going for 1650 Hertz CP is still under 51 degrees Celsius so you would have no indication at this point that something's wrong 116 degrees okay well we were away for a minute getting everything set up so this is with an SOC voltage of about 1.2 nine to 1.3 this is where depending which manufacturer you talk to you they might say this is actually a safe voltage to hold but they would probably suggest not to go beyond this we're pushing that down it's oh it's actually slowing down it's not crashed we're pushing we were pushing nine hours we just went down to two right when I took that measurement no it's back okay maybe that was a throttle that might have been a throttle on the PRM so we're back to pushing nine amps if you want to do the math that's nine times twelve point three to get the watts which is 111 so we're pushing like 30 watts more than we were previously into the EPS 12 volt rail or down it and 117 see if it throttles again at the same yeah so it's throttling when this reads 118 which might suggest that the vrm is trying to prevent catastrophic failure so that noise I just took a recording of is coil whine and every time it stops we know that it's throttling let's see if it does it again it'll be out like 118 should be around here if it does throttle yeah it's getting a little higher each time Oh hit 120 this time starting to get towards runaway what kind of power are we pushing down into this poor motherboard it's actually less power it does smell like maybe the tape the thermal tape I'm using might be starting to burn 123 degrees mm-hmm SOC voltage still saying about 1.3 so you're following the guidance from vendors who've said 1.3 then you're still within guidance and you think about this if you don't have a probe on it and you don't know any better that a vrm can even like that it even exists or that it can overheat you look at your CPU temperature like man 53 degrees that's really good and I have a 16 50 megahertz overclock stable and everything's looking pretty good on the screen it's not even slowing down anymore because I increase the voltage enough 130 degrees people from the thermocouples this is a real thing that can happen in use cases where you have no air flow so MOSFET one 130 mosfet to 128 TDI only 54 degrees SOC voltage 1.28 to 1.3 133 oh gotta shut down so we might have just had an over temperature protection shut down oil wine stopped everything's hot welcome to hell alright so fire strike I think corrupted itself in that crash so we're just using fur mark now which is a power virus for VRMs basically so eighty degrees I'm gonna demonstrate one more thing here I don't know that we'll get a pop tonight because it looks like there's an over temperature protection to force a shutdown doesn't mean it'll protect you from degradation or anything but it will shut down the system so what I'm gonna do is take a capacitor measurement before we finish and see what temperature those running at and how quickly would they die although they're not as choked out for air flow as they could be there we go 100 degrees already that's quick so alright so I've jammed the capacitor in there it's not exactly where I want it but it's decent this number is what we're looking at can you read that sorry MOSFETs at 100 14.4 our capacitor is currently at 94 and rising so here's the thing if you're able to get everything let's say you don't get a thermal runaway scenario at 130 degrees like we just did where it shuts down at 136 external temperature is probably 150 internally thermal shutdown to protect you let's say you're sitting in stead of something like 120 where there's no thermal shutdown and things look actually okay in Windows so at this point you'd say you know what I'm operating within spec and there's no visual issue there's no artifact and there's no crashing there's no actually the coil mines pretty much gone away you might think that this is actually okay it looks stable but our capacitor is rapidly approaching 100 105 degree marker we're at 97 now and that's without even being an internal reading I that's an external probe and I'm not even really happy with where I've positioned it so I think it would be worse in reality potentially so where we are right now is with just fur mark we were we were hitting 130 plus which 550 internal because that's that just happens to be a magic trick point so say where let's call it 140 heading 140 on the mas but I was measuring primarily we currently have one capacitor at 91 degrees and this is a completely stable overclock the framerate is not dropping any more we don't really have coil whine anymore and we're at 1650 megahertz and things look fine so what isn't fine in fact is that we're at 91 degrees externally for the capacitor we already know that the Delta between external and internal can be quite large because we were reading like 135 136 and we had a thermal shutdown at that temperature for the MOSFET that's not really a normal thermal shutdown temperature 136 internally it was probably 150 and you can look up the datasheet I'm sure and get the exact number but either way we've got at least 10 degrees of headroom there which puts us pretty close to 105 this is in open air if I put this in a bit Phoenix and so and that's the GoPro if I put that in a bit Phoenix and so or some other closed-off case it really wouldn't be hard to get the extra 5 degrees if we're already at a hundred plus and this is with not perfectly accurate external measurements anyway so what I'm saying is although we're apparently not going to get a catastrophic pop today because there is in fact a an over temperature protection what will happen is in about maybe five thousand hours of use in this condition your motherboard could potentially blow a capacitor and if it doesn't cool you're lucky but it will eventually and that's kind of the that's the problem that's where if you're overclocking on a platform like this it's not really meant for it a lot of be 350 platforms or like this one even though they have fake garbage heat sinks on I mean like literally basically plastic hunt some of them like the top of the heat sink is plastic so when I say fake I mean actually it's it's a fake heat sink in fact so they have bad heat sinks on them and you're running in these conditions you can blow a capacitor and and it will take a while so this is this kind of response like anyone who thinks they're doing fine on a closed off airflow case with a cheap motherboard as bad vrm or bad vir I'm cooling and they think that they're okay because you're fine today but see how it isn't a year of that kind of use maybe one and a half years depends on how much you gave under those conditions or whatever you're doing so yeah I think we can we can stop this one here but basically we didn't get a catastrophic failure which I really wanted because it'd be cool on camera but the board has some kind of over temperature protection and the MOSFETs but you still got to see the whole process and you got to see something hit 136 degrees which is pretty damn cool for a computer component well actually it's the literal opposite of cool you know what I'm saying so yeah it's just kind of a fun workshop basically there wasn't a whole lot of point to this except hopefully there was some stuff I said in between the beginning and now that gave you something you can take home and learn from even though we didn't get it cool like catastrophic flame out from it but if we left this set up for a while it would eventually degrade and or die depend on if the IMC takes the damage or if the capacitor takes the damage first so yeah that's it as I said no major conclusion here or at least not like a bombastic one but just kind of workshop it from beginning to end for fun with high temperature stuff so subscribe if you like this type of content we're gonna do it killing your IMC one in the next few days go to patreon.com/scishow Andrews Nexus tops out directly or store dock gamers Nexus dotnet to pick up a shirt like this one which is the Jian Graf logo shirt that we just restocked thank you for watching I'll see you all next time