How to PERMANENTLY FIX high motherboard VRM temperatures throttling performance
Вставка
- Опубліковано 1 жов 2024
- This video is intended to showcase how you can reduce VRM temperatures on any motherboard that lacks factory fitted heatsinks.
I was just going to make a video and moan about how bad not having VRM heatsinks on some motherboards is (i say some because things like socket FM2/+ exist) can be and why you should really put some thought into your purchase so you don't accidentally hobble your CPU.
Mid way through I decided to do something about it and see if i can unlock the full potential of what is a decidedly mediocre board, but one that should still be able to extract a decent amount of performance from an I7 9th Gen CPU.
The last step of adding a securing element on the top of the newly added heatsinks might not be applicable to your specific motherboard if it does not have through-holes, but apart from that all steps are the same.
Music:
Track 1: Sweet Talks by Limujii
Track 2: We Are One by Vexento / vexento
If you install heatsinks on your (maybe A320 Board?) VRMs use a big top blow cooler, not a watercooler.
It helps tremendously
I did - big ol' fan aimed straight at them that only comes on when the MOSFET temps are going over a certain threshold
Assuming your PC case has a MB tray cut-out for cooler installation (most do), cut a fan out of its frame and stick it to the CPU cooler backplate with RTV. Use a splitter cable from the CPU fan header (or the 'CPU opt.' header). Most cases will have space for 15mm depth fan (80,92,120mm), some even for a 120/25mm. VRM mosfets are actually intended to sink heat to the PCB ground plane, i.e. the motherboard (NOT via the plastic package seen from the top). This mod, which is much easier than fiddling around with little heatsinks, using a 92/14mm (sic - odd size Noctua) dropped VRM temps on an Asus Prime Z390A (probably a very similar to this board but with marginally effective heatsinks) with a 9900K, 160w package draw from c. 78 °C to c. 50°C, UK summer temps. You can barely hear a fan in this location even at 2 or 3 1000 RPM)
Hmm I'm going to have a look at your small Noctua fan idea, funny enough I've got a A320 I've been meaning to look at
@@ESHW Give it a go! I actually went further and made a baffle out of heavy black card with a cutout and shallow cone or venturi around the fan, attached with 3M magnetic strips, which seals off the MB cut-out and directs all the fan's output into the 6mm gap between MB tray and MB (kind of resembles a 'blower' fan now). I've since never seen VRM temps over 50°C even with unrealistic Prime95 loads of 200W+. This is with a Noctua NF-A9x14 (it pained me to cut it out of its frame but it was all I had to hand at the time).
I have a 12900k, I haven’t OC it yet and i only use it for gaming. The VRM has measured 44C max temp when running cyberpunk on 4k ultra. I’m using a 3090 with a slight OC and the gpu is at 70-75 but the memory is at 90. I’ve already replaced the pads. What should I do to squeeze out the extra performance.
By the sounds of it, you can try and experiment with a bit of a higher OC/undervolting. You are probably already getting the most out of it on a normal day to day basis. If your GPU (I am assuming this is edge temperature) is at 70-75, what is your hotspot? It is usually 10-15C higher, but it can be more, especially after a thermal mod - that is literally the only thing I can think of that might be limiting you . GPU memory at 90C is perfectly acceptable in gaming though :)
I have an MSI H-310 paired with i7-8700 and why would anyone pair a power hungry CPU with a cheap mobo ... well, BUDGET ....
I managed to squeeze a little bit more performance by undervolting which was a surprise but thinking about it again it makes sense after installing (intel XTU) there is an indicator that tells you when you hit (power limit throttling) so what i found for my particular mobo is that the throttling stops completely at ( -60mv and when i tried again benchmarking the CPU on (XTU & CB R20) it gave me better scores with lower temps on VRMs & CPU.
do you power limit at 65w?
An MSIA520M Pro VH ($60) paired with Ryzen 5 5600x ($190) and VRM Mosfet heatsinks ($7) = extreme budget gaming build and a 😁
how did you stick heatsinks, thermal tape or glue ?
I have a "b450m a pro max" and for the cpu I have 3400g apu -has a built in somewhat ok gpu- games usually run ok but when the VRMs get hot, I get 5-6FPS for 10-15 seconds then another 30 seconds of good gaming and the cycle repeats, I tried to direct a fan against the VRMs but that only helped a little. Do you recommend your method for my motherboard? do you recommend thermal paste or tape paste or something else, and if you have a b450 video I would love to watch it. thank you.
Fun fact - I've done this modification on both B450 and A320 boards, but never recorded it - I'm sure I mentioned it in a few videos (identical workflow, just without the strap across the top). With a 3400g I am very surprised you are getting this behaviour (sounds exactly like VRM throttling, but it must be really hot in your room or the CPU is way OC'd). I suggest using thermal glue (make 110% sure it's the right kind) and glueing heatsinks on; then let it set for a day or so - DO NOT add any thermal pads pads or paste if doing this.
Alternatively if your motherboard is still in warranty and you want to make the changes reversible use stick-on heatsinks, but if doing that I very much recommend beeping your computer horizontal so they don't detach (once the VRMs get hot they do have the tendancy to fall off - hence the very permanent and very warranty breaking thermal glue). Because you are using an APU make sure you add heatsinks to the SOC VRM (the bit on top) as that regulates the GPU side of things. Last thing - just adding thermal paste or thermal pads won't do anything without the heatsinks ;)
@@ESHW Thank you very much for this detailed reply. are there many types of thermal glue or did you mean the right brand? Thank you again
@@ferasa8394 There are many types of thermally conductive glues - make sure you use something that is meant for fine electronics (if you just google it the first option is JB weld - don't use that as it is heat resistant, not thermally conductive glue!).
The description should say suitable for heatsinks.
I use something called Easycargo ES910 that I can definitely say works well enough for VRMs (there are better out there that can be used on more sensitive components such as bare dies - you do not need that for your application, plus the price also jumps considerably). There types of glues come with many alternatives such as the Halnziye HY910 and other brands (Toaiot and a few others) - they all look identical and I believe they are the exact same thing from different vendors. Alternatively I also used (a long time ago) AG termoglue AGT-116 that says it has over 1W/mK thermal conductivity....but it was rather expensive as it came from Poland so I switched to the more inexpensive stuff.
Also make sure that any glue you buy DOES have a high enough insulation coefficient so that it is considered an electrical insulator (you don't want to short out anything - 99.9% it will be fine if it is silicone based, but it does not hurt to check the technical spec sheet on the manufacturers website).
@@ESHW Sorry for being late but thank you VERY much.
I am using Ryzen 5 3600 on a A520M D3SH motherboard, the VRM temperatures when gaming is like 56°C, is this temperature high? I am not planning to overclock or do anything
That temperature is ideal! You really should not be worried (like at all) unless you hit the 90sC constantly...
ive been have pc restarts like 5 minutes into some games. I've done everything else possible to fix the issue and now I'm thinking my vrm is overheating and causing a restart with no blue screen. my mother board is a asus tuf z370 plus gaming, my cooler is icue h150i rgb pro xt and my CPU is an i9-9900kf. I've made sure my temps are CPU temps are good and they are. I've updated bios, and other drivers. changed windows setting for the auto restart and some other things. the problem only happened when I put this CPU in the rig.
It's possible, but the z370 isn't a crappy board with bad or overheating VRMs. It should not restart if the VRMs are overheating, it should just throttle the CPU (massively) as any VRM of any worth will have that functionality.
@@ESHW dang i was really hopong that might be the issue. its very annoying. some games work fine but like sea of thieves will restart 5 min in, there's no visible throttling prior to the restart
And what about the chokes? On my motherboard the chokes are 110 celsius, can i install heatsinks on them?
I don't see why not, I usually just glue the heatsink touching them. It's the MOSFETs (or equivalent power delivery circuitry) that I personally always want to take care of and keep cool - those chokes/inductors are just a bit of coiled wire at the end of the day, not much to go wrong I guess.
I am using gigabyte b460m ds3h motherboard with i7 10700f pl1 and pl2 limit to 125 watt without any heatsink on mobio vrm the vrm mos temp reaches upto 112 c during cpu z stress test and the cpu frequency suddenly drops below 1 ghz from 4.6 ghz then again within seconds it comes back to 4.6 ghz and this strange behaviour always happens when i stress test the cpu or putting too much load on the cpu can u tell me what should i do to prevent the cpu from down clocking during gaming the vrm mos temp stays between 95 to 105
And the cpu temperature during gaming never go above 70 c and during stress test it goes upto 80 c
First of all try to limit the CPU power usage and re-test. See if the frequency behaviour changes and it stops seesawing and decide if the performance regression is something you can accept or not. The 112C on the VRM is certainly high, but it usually goes a bit higher before throttling (board dependant though, but VRMs throttle the CPU exactly as you described!) - there is a possibility the value you are seeing has some sort of delta or compensation factor applied to it for whatever reason and the true temps are higher (for instance if gigabyte knows that their sensor isn't at the hottest point and they compensate for that). Adding VRM heatsinks will help, but it is not the best solution as you have to smother everything in thermal glue (kinda irreversible) and it only helps if you are on the edge, usually by adding just enough heat dissipation surface to stop the throttling but the temps will continue to be sky high and most likely there isn't much room for improvement. Alternatively 10-series boards are becoming rather inexpensive today and you don't need to add that much over your ds3h for something with a more powerful VRM
not vrm ! this mosfet
I did both heat sinks and small fans, it really helps
How degrees you go down?
I already have a big cooler on my vrm but these will give higher temps right?
VRMs usually run quite hot - mid 90s is more or less the norm on mod-range boards - as long as you have some sort of heatsink on them it will be fine...
What thermal glue do you use? Is exist any chance of short circuit when heatsink connect with chokes and mosfets or other components?
I just used some thermal silicone based glue. A popular one (and relatively cheap) is the Easycargo es910, there are a lot of variations, but they are all basically the same product, sometimes marketed as thermal cement. It's completely non-conductive and safe to use, I've done about a dozen motherboards with it, just make sure you don't order one with added metallic powders and such as those might actually cause problems. Thankfully most of them are adequately labeled as conductive. That being said I do make absolutely sure that I don't accidentally short any components, if in doubt just leave a small 1/2mm gap and fill it with the thermal glue. The top of the modules themselves are non-conductive, but do have a proper look for any stray pins or globs of solder that might pose an issue.
@@ESHW thx man
Can you recommend best budget mobo with h410/h510 chipset and b450/b550 chipset?
@@koreanfanofficial4667 On the Intel side, i must admit I don't have that much experience with 10th/11th gen chips so not really sure what to recommend until I've tested them out myself. All is know is stay away from the Asus Prime series as they cheap(ed) out massively to the point they don't even have heatsinks! On the AMD side though I would strongly advise for the B450 Tomahawk MAX (it's awesome and unless you need PCIe 4 go for that). Only drawback is that it does not have aRGB support, only 12V RGB. For just a bit more you can find the Asus Prime X570 which has both aRGB and 12V RGB support + PCIe 4 support, but overall is not as good in my opinion (memory support is a bit worse and the VRM/OC potential is nowhere near that of the Tomahawk). An OK option is also the B550 DS3H from Gigabyte as it is relatively inexpensive and it gets the job done; it has decent connectivity + both aRGB and 12V RGB, but it won't OC that well . The Aorus ones are good, but also on the expensive side.
Hi can i ask a question? If the vrm gets burnt is the cpu damage aswell? Can i still use the cpu on a new mobo?
It kinda depends how the VRM failed, if you are unlucky and 12V went to the CPU it will have killed it, but it's quite unlikely. I would definitely test the CPU on a new motherboard, chances are it will be fine.
How to do it in laptops?
Best I can think of is you add thermal pads and link the MOSFETs to the laptop chassis...but that depends if it's feasible on a case by case basis
@ESHW sure I’ll try it
im not buildzoid lol
125c VRM temps holy god.
Is that on your card?
@@ESHW I heard it in the video. I did have the same issue with a shop bought OC, motherboard would hit 110c with a fan cooling it. Its not good for other components. 125c is the maximum some VRM chips can take.
I would be adding heatsinks and a fan to VRMs at 125c and then looking for a new motherboard.
@@GodKitty677 did you fixed it? I'm dealing with same kind of deal and just ordered 2x 40mm fans for cooling vrm since i need sustainable system for long rendering sessions
My VRM in gigabyte B560 without heatsinks are 97c while playing games my cpu i7 11700
eh - that is on the high end but still *technically* acceptable temperatures
@@ESHW but my cpu and gpu have good temps 60-65c and i have throtting in cpu , i think is vrm protection because sometimes reach 99c
@@Centuriom it shouldn't throttle at 99C (might be that the sensor is a bit off or there is a built in reporting deviation for some reason, I've seen a lot a boards doing that, mostly MSI so when the temperatures are reading 97C it might actually be 110C) - boards usually start throttling when over 110C or so and if the temps keep climbing past 120C then it throttles very had because most good quality VRMs have a 125C thermal ceiling above which they will start degrading. If the CPU suddenly drops to below 1Ghz on all cores for a couple of seconds, then comes back and sea saws like that...then its probably the VRM protection kicking in. In any event considering you are running an I7 chip on a heatsink-less VRM board...I would consider adding them (fair warning warranty goes out the window)
@@ESHW yes bad idea pairing a i7 11700 125w TPD in a 6+2 VRM no heatsink motherboard, my throtting is no so much hard 4600mhz go to 4200mhz
Whats the temp at the end
If memory serves it plateaued around the low 90s when running cinebench which is firmly in the safe zone.
Bro Im using a gigabyte b560m ds3h motherboard that comes without a vrm heatsink and I plan to add the same heatsink as you.. I know it's still OK to use without heatsink since it's a budget board that comes only with a 65watts powe limit in both gaming. But during rending in programs it able to increase power from my i5 cpu non k from at least 100watts and it goes 65 Celsius during rendering. My main concerns is in the Adobe premiere during video export my vrm increase to 79celeius and I plan to decrease it by following your tips here.. The question is will it really lower the vrm temps??
Adding heatsinks will no doubt decrease the temperatures on your board, but unlike the one in the video, yours is not overheating at all (depending on the exact make and model the maximum allowable VRM temperature is in the region of 125C for most common used ICs) - your VRMs are under no risk at only 79C. That being said it is a low cost and fun afternoon project - make sure you use the correct non-condictive thermal glue and if your heatsink is not anodized make sure you aren't shorting anything.
If you don't mine me asking why are you exporting video in premiere using your CPU? I find it excruciatingly slow if I forget to set the task to the GPU encoder
Ok thanks for the advice if 79c are under no risk I find that actually noticeable too. I'M just makng sure my motherboard will last if what if I keep going using adobe premiere like this during the exporting after finished project the vrm temps are so high every time I convert the video...
@@JLgaralde you cab literally leave the VRMs at 79C 24/7 and its going to be 100% OK - no degradation and no lifespan reduction....
I have same b560m ds3h paired with i7 11700 125w , and VRM on games reach 97c and i have Throtting so less mhz in the cpu...i put a 12cm fan in front of vrm but only temperature down 5c .. need heatsinks i think
its common for most motherboards with out heatsink and right now the day 12th and 13th gen were release right now all of there budget motherboards in the B series have heatsinks now. They probably notice this and to compete with MSI and ASUS noticing that there motherboards have heatsink even for budget boards they went putting some in the leftside only..
Can a thermalpad work well as alternative to thermalglue?
Only if secured in place via mechanical means (i.e. the strap across the top)
@@ESHW means only has effect with downforce pressure?
My mobo doesn’t have any hole :/ It’s B365m D2v board and im intending to swap my 8400 to 9700 so ive just bought a pack set of heatsink cubes with multiple sizes like yours to fit in my small mosfet area.
@@minhnguyennhat9556 the thermal glue will be more than enough to hold them there - I went overboard as I usually sell a build once I finish playing with it so I had to ensure that they would not detach in shipping
@@ESHW Im not sure if the glue is not too super-adhesive that the hearsinks would not bumpup my mobo mosfets when i attemp to uninstall them.
@@minhnguyennhat9556 Err - I would not try to uninstall them to be fair - there is a good chance the board will be damaged. If you need to have the modification reversible (i.e. for warranty purposes) - then just use something like 3M 8810 thermal conductive adhesive tape - it will work, but it won't be nearly as secure as proper thermal glue, but if you only use the tiny individual heatsinks it should be fine.
Do not forget the inductors. They will be hotter in return too. Seeing that the chokes are of budget quality, they tend to be MUCH hotter.
I see no issues with my h410m-e and i9-10900 non k
If you do not lift power limit durations you most likely won't have any trouble whatsoever, your prime board will do just fine. If you do set the PPL 1&2 to their max and the board doesn't have a heatsink (or a ridiculously overspec'd VRM) then I would check temps just to be safe. I initially thought that it was a reporting issue, surely they can't allow it to go that high without throttling...
@@ESHW I just do gaming so it’s fine
For people who always use all core load it’s a problem
You are using an i9 10900 on a H410m motherboard? is it good for gaming? No throttling? How do you keep it cool? I have an H410m with an i5 10400f and i was thinking of getting an i9or i7 but I've heard that the VRM gets pretty hot with such powerful CPUs on the H410M MOBO
@@imnotatimetraveler84yearsa39 room temp 28-30c
Was using a normal tower cooler snowman m6 but now using Deepcool L360 ARGB no issues with temp in both cooler while gaming cpu throttles down and temps stay under 60c on aio and 75c on air cooler when running al core stress test gaming 0 issues boats up to 5.1 ghz easily
@@imnotatimetraveler84yearsa39 VRM gets hot only if you lift power limit to too high stock 64w up to 150w no issues in my Mobo
How are you measuring VRM temps? Does that board have an inbuilt sensor that can be used with HWinfo64?
The board has a built in VRM temperature sensor and HWinfo can read it. I initially thought it was inaccurate so i used a handheld surface thermometer and measured the MOSFETs themselves and that showed over 100C as well confirming the ridiculous operating temperatures.
@@ESHW weird, my ASUS b450-f strix doesn't have an inbuilt sensor that shows in HWinfo.
@@fablearchitect7645 your only option then is to take a handheld thermometer and point it at the VRM if you want to determine the temperature - BUT - considering the massive heatsinks that the B450-f Strix comes with, unless you have an overclocked 3950x/5950x on there i really would not worry about VRM temp at all as it is a beefy 8-phase VRM with adequate cooling. My guess is that a temperature sensor was not included because ASUS determined that no matter the (normal user) scenario the VRM will not overheat.