Gadgetory


All Cool Mind-blowing Gadgets You Love in One Place

EVGA VRM Test Planning: New Thermocouples

2016-11-10
everyone so we are updating our EVGA vrm testing I thought I'd give you an inside look as to what's going on with the first video we basically looked at a validated Tom's Hardware des testing of the EVGA FTW VRMs for the 1080 FTW we have this was also a problem with other cards 10 60s and 10 70s included in some cases that's all in the first video and then we also did a thermal pad replacement video to cool down the V RMS to an acceptable temperature now here's the thing that testing was done with thermal imaging I'm going to talk about why that's problem in a second but we were validating what Tom's did and use their methodology second thing I want to put a direct thermal couple probe on the trouble spots specifically the second and the seventh MOSFETs to see how the temperatures look but before getting to that project this coverage is brought to you by our patreon backers who helped out by providing some extra funding every month and in return we give some inside looks and this is one of those if you want to help out with future videos like this hit the link in the description below to patreon and you can check out our page there I'm not going to recap the entire issue here and all the testing that we've already done it's on the website it's on the channel if you want to look at the thermal imaging we did but we are expanding on that quite a bit I've done some endurance testing as well I'll talk with that soon first of all thermal cameras so these are pretty popular these days because they can plug into smartphones now but there's a lot of problems with them they're great for rapid prototyping they are great for quickly figuring out where the hotspots on something and have temperatures actually improved but they're not necessarily that accurate they're fairly accurate except for well I should rephrase that it's not the accuracy that's the issue it's where you use it and how you use it and in the case of these when you're dealing with something like a tempered glass side panel on a case or an acrylic window on a case these are pretty much useless because of the emissivity of the surface that's true somewhat for these as well for video cards looking at a PCB you're okay it's not quite as reflective shiny and it's a bit more accurate in that regard but when we take a thermal pad on here there's two problems with a thermal imaging device and it's not going to be resolved with the more expensive one either the first is the civet e of that surface how reflective it is things like that interferes with what's being read on the camera the second problem is just that it becomes an insulator so when we put a thermal pad on here yes we saw a pretty big improvement but if you think about it you're measuring the temperature of a pad that's sinking all of this heat and we don't necessarily know what the underlying components are measured at even though they're being synced now directly to a backplate in the case of EVGA is this is a metal back plate so it actually will dissipate some of the heat so it was a good first test we definitely saw an improvement overall in performance of the card I did stick a thermal couple down there thermocouple probe like one of these down into the heatsink for that first test and I validated the temperatures so we know that these vrm pads are a definitive improvement but we don't know the specifics of how it works with the hot spots that are way low down on the card which if you look at photos of the 1080 FTW s and even the 10 70s that folks have posted online where they've had trouble the sort of explosions pots where the the fetes burnout are the second mosfet and the seventh or the eighth mosfet which is sort of well just above the middle so I want to test that and the way we're going to do that is by using a thermal probe and the problem here is these are the ones we have they are k type thermocouple but they're not flat they're pretty damn flat but they're not perfectly flat it's relatively flat that's not the main problem though the main problem is that it's exposed metal by nature of being something that's reading temperature so you can't just stick that into a card because it's electrically conductive the problem we run into is if I stick one of those in here I can't really just put it on top of the MOSFET because now you're interrupting the flow of heat from the MOSFET to the thermal pads and so now your temperatures are going to look way hotter than they actually are so our thought was well let's put one in the middle but now you've got a new problem now it's touching all kinds of electrical components including the feet for the feds sit ups on the MOSFETs feet touching the capacitor all the tiny components in there that can bridge something and cause a short so we don't want to do that because then the cards dead or there's damage to it and we don't know what's going on anymore even if it appears like it's working you kind of lose control the situation when there's electoral issues so couldn't quite do that the next thought was we could use captain tape this is ka pton captain tape is not electrically conductive it's pretty thermally insulating it can withstand far higher heat than these are going to put out so it seemed like a perfect solution the idea would be you put some tape around one of these like this and then it would theoretically not conduct electricity and jump any components and it should be resistant to melting the tape itself that is so that wouldn't be a problem the problem that had becomes well it's thermally insulating so we don't know kind of how much of our measurements getting lost in that tape so here's the solution I came up with I spoke with a couple thermal engineers in the industry we spoke with Bobby kanstul you may have seen a video with him he was the guy we spoke to at Corsairs lab previously basically told Bobby hey we're doing some tempt some some very specific temperature testing I need a thermocouple that I can stick to a surface and not worry about one conducting electricity and shorting something and to ideally something that I can stick to the top of a service between thermal pads and not worry about destroying the heat transfer from the device to the pad and Bobby helped us out he sent me in the direction of buying some Omega thermocouples and they were pretty expensive but I picked him up and as far as thermocouple goes you normally spent a couple couple bucks for one but these are $65 45 and then I paid a lot for shipping to get them here in time so these are pretty damn cool the cable may look like it's exposed here this is actually housing it's got a really thin film housing on it so it's not exposed copper that's good that means we won't conduct anything they're short anything or stuff like that then the end of it here the actual thermocouple itself is buried within it's very thin that's important and it's within a an adhesive pad that can withstand pretty damn high heat more than we need to worry about more than 100 Celsius hundred 10 Celsius I'm worried about measuring and as a final benefit it's not going to cause the same issue where if we have one of these between the thermal pad and the device being measured the MOSFET we're not going to have issues with insulating the thermocouple or destroying the transfer of heat there will be some change it's not a perfect solution but it's a pretty damn good solution considering we're not a validation lab with a hundred thousand dollars of equipment so I'm pretty happy with that now the problem with these when i got them what huh so there's always problems never easy this is why it takes a while do these sometimes problem of these was these came without any kind of connector on them it's just two wires so you've got a positive and a negative wire and these are I'll give you a backdrop here so hopefully folks has been better so there's a positive and negative wire these have to go into the housing for our k type thermocouple zand then that'll connect to our amprobe thermocouple reader which is what we use for all of our ambient testing and so I've already built one as a prototype we're going to build another one now basically I'm taking these existing thermocouples I really like these for ambient testing for validating temperature for case tests and things like that we take a measurement actively with two of these connected to one of these and it'll tell us the temperature see if maybe you can get a reading on this let me switch these different spot so they just plug in the top there we've shown this before but you get a reading of the temperature so it's about 22 Celsius here ambient temperature and just show you that it's kind of working I'll go ahead and grab the end of it and it will jump up so that's going to be a skin temp so pretty cool stuff this is fantastic for validating the ambient temperature we subtract it from our diode readings you can see it takes a bit to update but it does update pretty well and it's a second a second measurement so that's all stuff that we want this is all good news but what I have to do is I have to take apart a couple of these and and put the housing for the end the actual connector around one of these and it's not too hard but we're going to go ahead and do that on camera here this great tool but it's not the perfect solution so I do want to do these thermal couple measurements will take direct measurements of the vrm take some of the vrm without thermal pads some with it and that will tell us how hot is the thing and its hottest of hot spots and are they actually dangerous temperatures the final video of this series after this video will be after I analyze all the data and I'm going to be working with build Zoid from actually hardcore overclocking will get on camera together and we'll talk about what our findings were for all this EVGA vrm stuff going around I know there's questions about vram I've seen your comments I'm focusing on the VRMs for now because it's enough work already vram will come next if at all but I can tell you from our quick look at all the devices i have here which is for i've not seen any issues with Vee around contact but that doesn't mean it's not an issue I just haven't seen it but we'll talk about that later let's focus on the VRMs today so building the thing this is the easy part the next part of to tell you about what are our limitations and concerns when actually putting a thermocouple a probe on the device and see if this is right yes that will work so there's three screws in these there's a positive terminal here on the inside that's connected actually to the screw and a negative terminal or ground the positive is the thinner of the two prawns so let's take these out this will it's sort of open the housing and there we go yeah third hand here so open the housing once they're fully out and at that point we can wrap the copper exposed wire from our new thermocouples around the screws and get a better thermocouple out of it at least the better one for the specific purpose ok so these screws just hold the the housing to the connector I don't need to take those out or do anything with them these two screws are the ones that we've got wires around so if you look at it you'll see that this is the thermocouple if you look at the end of it hopefully that shows up there's a red and a yellow wire one of those is ground and ones positive or negative and positive and they terminate in the metal probe and touching that is what gives you a heat reading or even just an ambient it gives you a reading of ambient temperature so from looking inside here we can see the yellow one the smaller instrument the point the yellow wire right there at the end of the screw driver is the positive and that's connecting to this screw it's just copper wrapped around the screw and the ground the negative wire is the red one which also terminates in bare metal wire and goes to this screw so we're going to take those apart and I believe based on my testing I did earlier we can even take the existing thermocouple out without needing to cut the cable or or strip the wire to the cable housing which is nice because I'd like to be able to reuse those they're good and I like that so there you go that's the easy part it's got really nice housing but we're not going to keep it for the new cable your thermocouple basically you've got your two ends there and those wraparound the screws let me know when you have that that's good ok as so now we're going to move this aside don't need that anymore get one of these new ones unwrapped just enough so that I can work with it and I have to remember now which I believe it's the same I think it's still yellow and red and I think yellow was positive so I think yes that is correct so we're going to do the same thing with these really thin ones these are just flat a like straight cable the other ones were bent nicely so I'm going to have to replicate that I will be bending these do that need just some pliers or something kind of fine to work with so I'm going to bend these one at a time and a lot of this is housing it's just the very very end that's actually exposed wire that's decent enough to get started I'm going to go ahead and stick one of these screws in there and make sure it lines up yes ok so this will work and it's going to be very hard to see this on camera so a plan here is to what's the easiest I think it would be easiest if I get like one thread in there and then hook it yeah ok so that's in there so let's hook this around the screw and all it needs to happen is this metal wire needs to make contact with the metal plate that's down here so that's this plate right here so we're going to be making contact between that and the screw as long as it's touching that plate then it feeds out to the the prawns and that goes into our thermocouple reader and gives us the measurement we want so this is wrapped around I'm going to look in closer here okay that's pretty good so we're making good contact this will be pretty easy to test and validate once we're done alright that's pretty good so I think I'm happy with that I'm not going to put the enclosure back on we need to validate it for us to make sure that it's even reading a temperature and then hopefully it's reading an accurate temperature we've got some other challenges there too which I'll talk about in a moment so let's put this and I'll put this in slot wanna put it in slot 2 okay this is the other one I know these are accurate already now they should be like it may be a degree apart i think is about their accuracy they're a little different because this one is this one right here is reading temperature again through adhesive so it's not the same measurement as you'll see what the other but let's take a look 22.7 on thermocouple one that's this yellow known good one I have the one we just built so to speak 22.7 that is pretty damn close so we're at times looks like we're about point four Celsius difference which is perfect that's that's what we want so I'm happy with that that means this is going to be pretty representative of what we're normally used to working with and obviously there will be more validation we're not going to show it on camera but i'll go ahead and put two of these on the same place on a hot component will see what the heat temperatures the load temperatures are but so far it looks pretty promising since the idols about the same or i should say the ambient maybe alright so we've got two of these off camera i'll build two more of these we're going to use for total i have to a thermocouple readers the same ones so we've got four total I will be placing those all over the card the plan is to do second mosfet seventh mosfet up here and then that'll be before and after thermal pad mods and also be formed after the bios mod that EVGA pushed do one on the back side to the hot spot is right around where the inductors are inductors are going to be dead center here this channel with kind of nothing these are the capacitor banks the other side as the capacitor bank and then this is looks like our faites so the hot spots right around here and I know where it is pretty well because if we put the back plate on the card which i'm not going to fully mount it but you'll get the idea it's always right around here in this first ventilation port I'm not really ventilation but you get the idea so the first hot spots always right there so I want to be testing with one thermocouple there for sure so that's that's pretty good setup now what are the concerns here so the zoe there's always a problem this problem with the thermal camera there's gonna be a problem with the thermocouples problem here is we've resolved the electoral conductivity concerns we've resolved the thermal contact concerns where i'm basically i'm going to peel this you can see it bends like that so on appeal it will have an adhesive thermocouple we can apply to the surface which will go right around here and then i'm going to cut it to size a bit not too much but just enough so cut it to size so the next question becomes before mounting it obviously how do we route the cable well the problem is whenever you're near inductors is a chance a pretty high chance of I electromagnetic interference so with electromagnetic interference your concern is if we run this cable let's say we came in from the other side for some reason wouldn't really want to but so you coming from this side you might now by crossing an inductor be introducing EMI to the system and that's just going to depend on the device you're measuring and on the quote Bob you can still on the black magic involved with the particular thermal probe and the card so we don't want to cross an inductor the way EMI manifests itself in a thermocouple reader is basically you have a thermocouple reader like this and you'll see the temperatures on the screen bouncing between let's say it's under load let's say it's maybe 50 Saul's ease and you'll get a jump 60 celsius then we'll go to 50 again so there's a lot of jumps that are really big ranges in a very short period of time that's EMI you can tell it could also be a bad brew but probably EMI so how do we resolve that well like I said we avoid going over the inductors directly for one and then two you got to look out for traces but with a PCB like this we don't really have much much choice so we're going to have to cross traces and the ideally from what the thermal engineers have told me you want to cross the traces at a 90 degree angle but in this case I'm really not going to have a choice so we're going to come in with the probe from my thought is from the PWM section of the card and land on the fetes that we want to measure after cutting them this stuff to size so we'll land right around here and I'm thinking I'm going to route this cable just short of the stay away from all this stuff stated from this stuff I'm going to try and route it just like that straight out the bottom of the card and then it will feed into the thermocouple reader then we'll do the same up here seventh fat or eight and this one's going to be a bit trickier but I'm gonna have to decide do I want to ride out the bottom of the top and we need to do it in a way that the cable for the probe doesn't touch whatever thermal pads are on the original solution shouldn't be too hard really because there's not a lot of it that's all we're avoiding here is is that strip on the fetes so I think we'll be in good shape the next step is actually running the test so that's what we're going to do pretty cool stuff I am happy to get the process started here with this vrm testing I am sure that will run into continent some kind of complication and if we do all obviously take time to buy whatever I need to buy or rearrange whatever I need to rearrange that the tests are accurate so that may mean you're waiting about a week for the update but hopefully it will be the sort of end all be all of this vrm issue from EVGA so we know definitively a few things one how much of an issue was it with the stock card too does the bios update alone fix it and three how much do the thermal pads really do when you're looking through the lens of a probe rather than through the lens of a lens of a literal camera lens because again we've got the concerns of fem a-- civet e and of the reflective surfaces and just of the thermal pads insulating the backside of the PCB and if you're taking a thermal image of a device a weed just like toms did because we were validating toms methodology we took an image of the back the PCB that's great but and it's good for for validating finding issues what it's not great for is a PCB itself is sinking heat so we may be getting temperatures that are either slightly higher or slightly lower than whatever component were measuring depending on what it is and which side of the PCB it's on because you're going through a couple layers of PCB there so that impacts the results but that's the theory that's the plan thank you for watching as always subscribe to follow up on this because there will be at least one fault probably one follow-up hopefully the final hun and patreon link in the post roll video if you want to help us out directly there will be a link below thanks to those of you who do support us already because it does help with things like this like buying one hundred dollars of thermocouples and should be fun test I checked by soon I'll see you all next time you I'm not stuff later
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.