Gadgetory

hey guys build Zoid here from actually hardcore overclocking and today we're gonna be taking a look at a gtx 1070 TI PCV this is colorful volcán x edition and well has quite the PCB and you know it is also a lot taller than normal before that this video is brought to you by thermal grizzly makers of the conductor not liquid metal that we recently used to drop 20 degrees off of our coffee leak temperatures thermal grizzly also makes traditional thermal compounds we use on top of the IHS like cryo not and hydro not pastes learn more at the link below so let's get right into it as there's no really much else to say about the card you have a minor supporting voltage well you have the pecks of erm down here which is normal for basically all Nvidia cards this vrm exists and this runs on one this outputs one volt and it powers the PCIe interface as well as some internal PLL's of the GPU core very important to keep the card working completely useless for overclocking until you're well below zero so only really worth worrying about if you're on a liquid nitrogen and even then it's really not a high power output vrm so it's you know you're never gonna see anybody do bother with doing a two-phase design of that it's always just a fully integrated buck converter like we have here up here we have what I suspect is the 1.8 volts rail which and video cards require for some internal PLL's on the GPU core as well as their bio system on cards with gddr5 X this would also be used for power for the VPP line of the gddr5 X chips but on a GT the gtx 1070 TI uses gddr5 so basically this is here just for the bio system and the GPU core now then we're gonna move into the actually interesting VRMs even if they do just as much as the other two in terms of what we're clocking on air cooling and water cooling this right here is the V curve erm so that provides the bulk of the power that your GPU uses and behind that we find the memory v RM right here so that power is the gddr5 memory chips located around the GPU core so the V curve erm is a one two three four five six seven eight nine ten eleven twelve phase the RM design and so naturally it is using a doubling scheme the doublers are located on the back of the card and they are not just doublers it's actually dual drivers with the phase extension functionality so basically you have one chip controlling uh basically turning the MOSFETs in this VR I'm on and off well you have four MOSFETs hook it up to one drug driver IC which drives the four MOSFETs well drives two MOSFETs at a time so if fective lee it looks like a twelve phase this is the only real way to do a four twelve phase vrm design I'm not gonna go into the details of how exactly it works but yes this this is the right way to do a twelve phase there's a way to make something that looks like this and isn't actually running out of like isn't running in twelve phase mode but this is the proper way to do it the chips for that are located on the back those are fed the PWM signal from this chip which that is the completely typical for NVIDIA cards you p95 eleven voltage controller this is an eight phase-- chip so here it's being only used in six phase mode so they're not using two of the phases available on the U P 95 11 and that's because you can't like if you're using a doubling scheme you have to double all of your phases you can't double like four and then use for more natively from the chip that doesn't work so that's why you know this is being and in six phase mode here and the u p95 of eleven supports switching frequencies up to 600 kilohertz which considering that and I ran out of space there up to 600 kilohertz and basically that's how quickly it turns the MOSFETs on and off in the vrm now since this is using a doubling scheme this 600 kilohertz would actually go to the driver chips on the back of the card 600 kilohertz and once it comes out of the driver chip to the phase like the these MOSFETs up here would be switching at 300 kilohertz and the MOSFETs down here would be switching also at 300 kilohertz because basically the way the doubling schemes all work is that they take the 600 kid they take the input pwm signal and just split it in half so every other pulse goes every other pulse goes to the other V our other phase so that's basically how this works now then let's talk about the actual MOSFETs use these are dual N fats from Alpha and Omega semiconductor these are AoE 69 30s these would have you would have colorful really likes using them I've seen these on a gtx 980ti from colorful but the first time you would have probably seen these is all in the GTX Titan XP as Nvidia uses it that these for their reference card these are actually really nice MOSFETs very low RDS on on the low side fat at zero point Wow no weight 1.05 milliohms RDS on for the low side fat at 5 volt gate-to-source and i'm using 5 volt gate-to-source because i have no idea what drive voltage colorful is actually using on this card I assume it's 5 volts because they do have this V arm here which the u p95 11 and a bunch of logic on this card actually requires a 5 volt rail so it just makes sense to reuse it for driving as well and then the high side MOSFET built into the a OE 69 30 is 7 million RDS on at that same 5 volt gate to source voltage luckily we're like this is pretty high but the high side MOSFET is all about being able to turn on and off quickly and not all about its RDS on and in that department this thing can turn on at speeds of as high as 4 nanoseconds turn on in 3 nanoseconds turn off so this is like this is really fast you find a lot of high-end MOSFETs out there that are sort of in the 10 nanosecond range for both of these and then when you start looking at the really cheap MOSFETs that are also available you start looking at 20 to 30 nanoseconds switching speeds so well turn falling times and rising times and well those are basically atrocious but these are really really fast definitely in line with some of the top MOSFETs that you can find out there so the end result is that with this you know with this high-end mosfet and the ridiculous phase account this vrm is absolutely insane overkill for a gtx 1070 TI so under normal operating conditions you'd be looking at this vr i'm pushing to the core about zero 1.09 volts which that's the maximum voltage you can set through something like afterburner and you're gonna be looking at currents between 150 and 200 amps for these kinds of current levels at 300 kilohertz switching for frequency because you know I'm doubling scheme and 5 volt gate to source voltage for 158 you're looking at about 8 watts of heat output for 200 amps and it's looking at around 13 watts of heat output so this is really really an efficient massive overkill on power colors I mean colourful messing up the name is Hieu sorry about that it's colorful not power color less colorful names would be good but uh yeah very very efficient we are I'm a ridiculous power capability very overkill I really think you know some people would probably prefer it if this wasn't as tall and the vrm was a few faces less but as far as the arm designs ago there's nothing to complain about here if for some reason you decided that you really want to run this gtx 1070 TI on liquid nitrogen which to be completely fair since no gtx 1070 TI is actually specifically designed for liquid nitrogen you might as well run across a run whatever you come across so if you decided to run this one you'd be pushing probably voltages around 1.3 5 volts some cars will be slightly lower some currents will be slightly higher and about 250 amps and also basically the reason why you don't go higher on the core voltage for liquid nitrogen is because jp104 really doesn't work well at high voltage it basically doesn't scale and I've seen like I've already overclocked at 1017 liquid nitrogen and everybody who's worked with 10 80s and 10 70s before can confirm this basically the jp104 chip really doesn't like high voltages even at very low negative temperatures so you end up getting stuck around that sort of voltage range and current draw which means that the VR I'm even under liquid nitrogen which would be a kind of extreme usage scenario ends up only producing about 19 watts of heat so again you know the it's ridiculous overkill like this vrm wouldn't look out of place on a gtx 980ti but here it is on a 1070 TI so props too colorful for splurging on components here now then the memory of erm is a two-phase design this is better than what you would find on most cards in sort of well just in general a lot of cards come with single-phase memory power and two phases can help with memory overclocking a tiny amount not anything huge but it can help so that's nice to see and the MOSFETs used are again the same Alpha and Omega 69 30s for gddr5 you're gonna be looking at an output voltage of around 1.5 volts maybe 1.5 5 volts on some cards so it depends but I'm doing the rating at 1.5 volts the controller is au p90 now not 90 1658 it integrates both while the driver circuits for both phases so there's no actual like basically there's no other driver chips anywhere around this VRM and this chip supports two phases up to 300 kilohertz switching frequency and I ran out of space again I you you'd think I'd learn after the first time wouldn't you evidently naw so 300 kilohertz 1.5 volts again using that 5 volt gate to source voltage because they're going to use the same drive voltage everywhere you'd be looking at maybe 25 amps of current output on two-phase memory on the gddr5 because we do only have eight memory chips here and at that current output level you're only looking at about 1.5 watts of heat output so again great job colorful with the vrm here and you know it's a good thing that these VRMs are both so ridiculously efficient because well the heatsink is kind of just a bit like the vrm eat sink where this card is just basically a sheet of aluminum and that like I mean I guess it has more surface area than the MOSFETs but that's not really you know that doesn't that's not exactly confidence-inspiring in the vrm cooling department but with the heat output levels that these the arms are putting out like the heat output on these VRMs is so low that I think colorful can get away with the anemic looking base plate the rme cooler so that covers the vrm portion of the card one cool feature that colorful decided to add on to the card is you actually get voltage read points off here so you can check your n VDD so that's your core voltage FB VDD which I am not sure what that that's gonna be the memory so that's gonna be the memory one there I'm not used to you know and I don't want overclock and video cards that much so I'm not used to the terminology used here so that's gonna be V core pecks VDD that's the v RM I pointed out down here 1.8 volts is the one I pointed out over here and you also get a ground tab so that's really nice they don't have a measurement point for the this VR I'm down here which I assume is 5 volts which is kind of interesting but this is not the first card to like include all the other voltages and not bother with the drive voltage and this can be handy if you know your Experian you aren't able to RMA the card and are having issues with it running these voltage read points are really convenient because you basically don't have to like figure out how to stab the right capacitor for this one if you if you think that there's an issue with this rail or that or any other rail on the card so you know it's a nice feature to have for overclockers though I think for most daily users this is completely pointless still a nice addition and it doesn't really cost anything to do as it is just a PCB you know small PCB design tweak now then if you want to really let the card fly basically take and videos power limitations off of it the shunt resistors for doing that are located down here and I wrote over one of them so you have three shunt resistors because you have three power inputs you have two eight pins and you have the PCIe as well and so each of these shunt resistors monitors the current going from one of those sources I am not sure which one is for which source if you're planning to short these out the best way to do that is generally to apply a thin layer of liquid metal across them because if you try to short them out by like soldiering over them or something you run the risk of shorting them out too much and if the GPU detects that the power the power reading provided by the shunt resistors is too low it will put the card in safety mode and basically the card is a safety mode on a gtx gtx 10 series card is basically the card is stuck at 139 139 megahertz core clock which is you know which basically makes the card completely useless so you know thin layer of liquid metal is your best bet for getting a ez power mod if you want to do something more advanced the ia32 to wanna is up here I've done a advanced power mod guide for like completely remove all power limitations on these cards guide on my own channel so actually our core overclocking but for most users it you know if your although if you're at the point that you're modifying but you don't have a soul during iron really just thin layer of liquid metal and you don't want to go heavy on the liquid metal because liquid metal does eat solder so if you apply too much of it you're you are running the risk of the shunts basically that well the solder dissolving and the shunts falling off so you know be careful when modifying the cards because you you can basically end up killing the card and well RMA is not gonna be okay with it when you send them a card missing components so that's it for this video thank you for watching like share subscribe leave a comment down below or any questions you have if you would like to support what we do here with gamers Nexus there's a patreon link down in the description below and if you would like to watch you know see more extreme overclocking kind of content like the well more of erm stuff and other overclocking ridiculousness I have my own channel called actually hardcore overclocking you can go check that out I imagine there will be a link down in the comments below for that so once again thank you for watching and see you next time you