Safe VRM Temps/Fix For GTX 280?

A

AtramentiA

n00b
Joined
Oct 14, 2008
Messages
53
Been seeing this around a lot recently, seeming to come up due to the bad aftermarket cooling for the GTX 280, specifically, to which I just switched.

What is the highest safe temp range for the VRMs?

I'm having the same problem as I'm seeing others around having... my VRM temps are literally skyrocketing under load, and I've seen my VRM temps hit up to 80C using the HR-03 GTX. Apparently, there are a lot of others getting the same results.

I have more than adequate cooling in my case, and the temps will drop back down quickly, but when gaming I range between 65C-79C, depending on how graphically demanding the game is I'm playing. Idle VRM temps aren't even any different than the stock cooler offered.

Does anyone know what the actual max safe temp range is for the VRMs on the GTX 280? EVGA's site states over 70C is "normal", but like many other people I'm seeing more recently on a few different forums, 70C-80C seems way too hot.

As for a possible "fix"...

I checked another post here from just recently, and it made me think of a similar thing: what if you removed the pushpins from the plate, removed the cheap thermal pads and actually used Arctic Silver thermal adhesive to adhere the plate to the VRM chips?

AS thermal adhesive has worked wonders for me in the past on memory chips, so I cant see why this might not help...?

The plate that comes with the HR-03 GTX is very light, and though the chips are small, I think between all the chips it might offer enough over all surface room for the adhesive to hold, and should cool a hell of a lot better. I've a feeling this might work, but am afraid to chance it if it doesn't, because then I'll never get the pads back onto the plate.

I'm tempted to try, but before I do, if someone could tell me what is considered a safe temp for the VRMs and if pushing 80C is too hot, I'll know whether or not it's worth experimenting with.

Thanks in advance.
 
80ºC under load is actually a very good temp for any card's core/memory, so you don't need to worry. You should only worry, when you reach temps near 100º on the core, which is very near the threshold temp for the core to throttle down.
 
Silus said:
80ºC under load is actually a very good temp for any card's core/memory, so you don't need to worry. You should only worry, when you reach temps near 100º on the core, which is very near the threshold temp for the core to throttle down.
Click to expand...


+1..
 
Silus said:
80ºC under load is actually a very good temp for any card's core/memory, so you don't need to worry. You should only worry, when you reach temps near 100º on the core, which is very near the threshold temp for the core to throttle down.
Click to expand...

I believe he is talking about the VRM's, not the core/memory .

OP,
I would try the AS-5 and see where that gets you...just make sure you get good contact, as the pads are rather thick...be sure the HS actually touches the modules.

My card is water cooled and I have a 40mm fan and heatsink assy mounted right on the card for the VRM's..this works out very well for me.
 
A VRM is nowhere near as temp sensitive as any kind of processor.

They do need to be cooled, but IMHO I doubt you will see any stability increase doing it. Best you will see is a more regulated Vcc for the GPU core and memory. Possibly longer card life by extending the life of the MOSFETS that make up the VRM. Water Blocks for GPUs generally ignore the VRM and those guys hit some serious clocks, so you are probably fine.

 
Most MOSFETs are engineered to operate around 100c continuously. However, they raise the overall temp of the card due to heating the PCB, so cooling them is important.
 
Crosshairs said:
I believe he is talking about the VRM's, not the core/memory .

OP,
I would try the AS-5 and see where that gets you...just make sure you get good contact, as the pads are rather thick...be sure the HS actually touches the modules.

My card is water cooled and I have a 40mm fan and heatsink assy mounted right on the card for the VRM's..this works out very well for me.
Click to expand...

Exactly, I am indeed referring to the VRM's and not the core (which is nice and cool with the HR-03, btw, just not the VRM's).

As for your mention about the pads, that's exactly my point, and I'm glad you understand what I mean, as that's my concern.

I don't have any extra pads around right now that I could cut up and put back on the sink if this doesn't work, so I hope it will reach down far enough if I attempt this.

jfb9301 said:
A VRM is nowhere near as temp sensitive as any kind of processor.

They do need to be cooled, but IMHO I doubt you will see any stability increase doing it. Best you will see is a more regulated Vcc for the GPU core and memory. Possibly longer card life by extending the life of the MOSFETS that make up the VRM. Water Blocks for GPUs generally ignore the VRM and those guys hit some serious clocks, so you are probably fine.
Click to expand...

The VRM's are not as sensitive? Why does it seem that so many are so concerned (even sites that review GPUs etc.) about the VRM temps? A lot of issues being made over the problem with this aftermarket cooler right now due to the VRM temps, but if they can handle that kind of temp without drastically shortening the longevity of the GPU, I wont worry too much about it. Obviously, the cooler the hardware the longer it's lifespan, so I do want to try something here.

vick1000 said:
Most MOSFETs are engineered to operate around 100c continuously. However, they raise the overall temp of the card due to heating the PCB, so cooling them is important.
Click to expand...

Of course, and that's exactly what I'm trying to figure out, without screwing this up big-time ;). The overall temp of my GPU right now is at least 10C+ cooler all around since moving to the HR-03, with the exception of the VRM's, of course, which are drastically hotter under load.

I should have my new tubes of AS adhesive early today, but I'm still debating on what to do, though I don't think I really have much of a choice. If the VRM's/MOSFETs can handle those temps for the long term, then I'll leave well enough alone. But that doesn't seem to be the general consensus regarding such high VRM temps.

Not really sure what to do here, though I personally feel that 70C+ under load is too hot, so if I give this a shot and it works, that would be great.

Well, unless anyone has any more info/advice regarding this, I suppose I'll have to assess the sink height again (little hard to tell; have to really eyeball it) to be sure and give it a shot and hope it works.

Thanks for the feedback. Any more is always appreciated.
 
P.S.

If you've seen the set-up of the GTX 280 and the way they did the sink with the HR-03 GTX
(see step 12 on this page), the only thing that's covered is the VRM chips. There is nothing on the MOSFESTs, for which there are holes cut out in the plate for them to stick out of/above the plate.

Should I put any sinks on the MOSFETs as well?
In the original GPU housing, they had nothing on them either, only the VRM's had thermal pads. Don't know if it'd make a difference in that case.

But you can see what I'm thinking of doing here; just putting the adhesive in place of where the pads are (in between the MOSFETs), removing the pushpins and hoping the sink sits low enough to secure it in place.

If this doesn't work, I've no idea how to cool this thing down.
 
Those are not the FETs. The little chips with leads that sit flat against the board are the FETs, a.k.a. "voltage regulators" or "VRM" (sometimes the leads are under the chips).

The bigger parts sticking thtough the holes are inductors, they don't get that hot.
 
If your that worried get an Antec Spot cooler and put it right on the VRM's. It has a three speed switch for fan speed and works pretty well for this exact issue.
 
vick1000 said:
Those are not the FETs. The little chips with leads that sit flat against the board are the FETs, a.k.a. "voltage regulators" or "VRM" (sometimes the leads are under the chips).

The bigger parts sticking thtough the holes are inductors, they don't get that hot.
Click to expand...

Oh, so mosfests are the vrm's? I thought they were two different things.

Copyright said:
If your that worried get an Antec Spot cooler and put it right on the VRM's. It has a three speed switch for fan speed and works pretty well for this exact issue.
Click to expand...

I've thought of it, though if Thermalright had done a better job, I wouldn't even have to consider it. Either way...

It's not just I'm "paranoid" about it, as it's a problem that is so far being backed-up as being a big issue. I just don't want to burn up my GPU.

For all I know, these things can take that kind of heat forever, or it'll fry itself tomorrow. I'm not really getting anything totally definitive in terms of numbers (even from EVGA) or lifespan with the VRM's running this hot, but in general I can infer from what I've heard and getting here, that hitting temps of 70C-80C on the VRM is a very bad thing and I should do something about it.

Using the plate as the sink isn't going to work. There are other things on the card in between the vrm's that I think are just a bit too high, so I'm screwed for that. I could fit normal mem chip sinks on there, but it would be precarious, as only half the sink would sit on the chip and not centered. Maybe the AS thermal adhesive would be strong enough to hold 'em on, but I'm not sure that's such a good idea.

I can try spot cooling it. I held a 120mm fan to the sink last night and didn't notice the temp dropping, so I'm not sure if that would even be enough. Maybe because the pads on the sink are just too thick. So, not sure that would even work, and I noticed today after work that pushing things in games such as Fallout 3, Crysis, Left 4 Dead etc. that I actually hit 79C at one point. Cant be good.

I've no idea what to do at this point. They do make smaller sinks, but I've heard people haven't had great results with them, and I cant seem to find them anyway. I can try to spot cool it, but again, holding a fan to it for about five minutes didn't make the temp drop at all last night.
 
How would I go about monitoring the temps of the VRMs? The only temps I'm seeing in RivaTuner are Core and Ambient.
 
The problem with using small sinks is that they don't dissipate heat well, essentially defeating the purpose of having them. They are fine for memory since VRAM does not get really hot, you could probably do without sinks on most VRAM anyway.

I still don't see why you can just dremel off the VRM portion of the stock sink and mount it the same way it was (but without the back plate), using some plastic washers. This way proper contact is assured and with a littel airflow over the back of the card they should stay just as cool as they did with the stock cooler before.

Cut it right here (red line)...

my.php
 
Parja said:
Ok, downloaded GPU-Z. I'm seeing four temperatures listed like this...

GPU Temperature
PCB Temperature
GPU Temperature
PCB Temperature

What's what here?
Click to expand...
Hmmm, I'm using GPU-Z 0.28, and under the sensors tab it looks like this:
gpuzdo9.jpg

"VDDC Slave" are the VRM temps if I remember correctly. Do you have something similar?
 
Mr. K6 said:
Hmmm, I'm using GPU-Z 0.28, and under the sensors tab it looks like this:
gpuzdo9.jpg

"VDDC Slave" are the VRM temps if I remember correctly. Do you have something similar?
Click to expand...

The Nvidia cards probably dont report as many temps as the ATI cards...Im using a 260GTX and only have 4
I have
GPU Temperature
PCB Temperature
GPU Temperature
PCB Temperature

vick1000 said:
I still don't see why you can just dremel off the VRM portion of the stock sink and mount it the same way it was (but without the back plate), using some plastic washers. This way proper contact is assured and with a littel airflow over the back of the card they should stay just as cool as they did with the stock cooler before.

Cut it right here (red line)...

my.php
Click to expand...

The problem with that is if you need the cooler fr an RMA you are screwed....:)
 
vick1000 said:
The problem with using small sinks is that they don't dissipate heat well, essentially defeating the purpose of having them. They are fine for memory since VRAM does not get really hot, you could probably do without sinks on most VRAM anyway.

I still don't see why you can just dremel off the VRM portion of the stock sink and mount it the same way it was (but without the back plate), using some plastic washers. This way proper contact is assured and with a littel airflow over the back of the card they should stay just as cool as they did with the stock cooler before.

Cut it right here (red line)...

my.php
Click to expand...

Well, I can try it, and had actually though of cutting it just about right where you placed the red line.

Question, though... are the VRM sinks on the stock portion tall enough, unlike the HR-03's, to reach the VRMs without pads? They're actually not much taller.

I'd also have to saw off the two screw mounts, which are taller than the sinks. No problem, I guess. Unless I cut further to left in your picture, just to the side of the original screw mounts on the plate.

The pads EVGA had on there were actually a joke, and flattened all to hell, though it did keep the VRM at 43C under load. Although, the stock fan is actually right over the VRMs, which probably helped keep it much cooler as well.

The plastic is a little thick, but as long as you feel the VRM sinks are tall enough from the stock cooler, I'll do that today and hope it works. I also have a small fan coming that I'm going to use to spot cool the VRMs, such as is the setup on the stock sink itself.

I hope they're tall enough to reach, because even though the original thermal pads were flattened all to hell, they were still taking up a little room. I guess I'll find out later today (friggin' 4am right now, so I can be waking the wife with a dremel, heh) and hope it solves the issue.

p.s.
Does anyone know the actual safe temps for the VRMs on the GTX 280? Because if the above doesn't work and I continue hitting upwards of 70C-80C under load, it'd be good to know whether or not they'd eventually fry, heh.
 
Whats strange is I contacted thermalright a while back asking when this cooler was going to be released. They informed me they were having issues keeping the VRM's cool. Now that the cooler is out I dont see any changes from the sink they orginally had issues with and the one that has been released.
 
You must log in or register to reply here.
Back
Top