VRM related question as far as heat goes

[bburrill2012]

Just wonder what causes the VRMs to get so hot?

Is it a raise in FSB or CPU/NB Voltage or a combination of both? My heatsink on my northbridge is pretty hot to the touch and I expect that but when I touch the heatsink over my VRMs its just as hot.

They are so hot, if I leave my fingers on them for more than 10 second it almost burns.

 

[bluezero5]

VRM is basically a transformer, fed directly from the PSU controlling the right voltage to feed it into your components, as such they dissipate a LOT of heat in that process. If you are serious about using higher Vcore and VDRAM, then you should look at VRM cooling as well. For my RIVE board, the VRM overheating was the first bottleneck that I had to overcome during my overclocking over 1.5Vcore.

 

[Archer0915]

Lots of current flowing through there. What is the phase design of your board and can you control them?

 

[bburrill2012]

12+2 and I have no idea how to control them if i could.

I am going to do a custom WC setup in the future so maybe Ill add that part of it to the list.

 

[bburrill2012]

Can't find any water cooling for the vrms

 

[bburrill2012]

I read in another forum by googling, some guy had the same problem, that his vrms are all of a sudden getting so hot that he cant leave his hand on the heatsink either. He then found out that he didnt have all the screws in his mb to every standoff/post. He said that fixed his problem. Why could that cause them to be more hot?

I will see if I have all my tight.. I did take the motherboard off a week or two ago.

 

[BeepBeep2]

That wouldn't fix the problem...
Are you talking about at idle or at load temperatures?

It is not uncommon for MOSFETs to hit 80-100c under load before any type of over-temperature protection kicks in to save your house from fire.

 

[bburrill2012]

not even under full load, about 60%. Was playing BF3 for about 2 hours. Even after 30 minutes of not playing the heatsinks were still super hot.

I backed it down from 255x18 1.425v and 1.28 cpu/nb to 215x21 1.4v and 1.2 cpu/nb and have been running prime now for 15mins and they only warm now.

I understand they are supposed to get hot but i try to be cautious. I've had a motherboard blow up and almost catch fire before, luckily i was home when it happened. I have a big black burn mark on my cpu cooler from it lol.

Although, i may put it back to 255x18 but might just add some fans, i am using an open air test bench and only have a fan on my cpu heatsink and gpu of course, no other fans.. so maybe they were just getting real hot cuz of the passive cooling. However I do have a 140mm fan right now blowing between my cpu and gpu

 

[Automata]

Considering 100c will boil water and 23c is normal room temperature, I wouldn't be surprised if 60c+ would burn your hand. Does your motherboard report what the actual temperature is? Going on feel doesn't say much.

 

[bburrill2012]

Its usually around 35C.. right now I cant tell because I have issues with this motherboard. I need to RMA it. My VRMs make a whinning noise under load for one. Also, i think some sensors are defective because in bios it says my cpu temp is 24 and mb temp is 45 and it never changes. Any temp monitoring program says the same thing and it never changes even under load. My 12v 5v and 3.3v readings in bios are SUPER low for examply 11.2 4.4 2.6. My cpu voltage in bios or in any temp monitoring program is SUPER low as well. I have it at 1.4v in bios and under load in prime right now open hardware says 0.84v and bios hw monitor reads about the same 0.8-0.9v

 

[bluezero5]

for VRM cooling, I have some experience, hope I can help.

first, if you can't find a water block for it, don't despair. You will be surprised at how well air cooling can do for VRMs. Just buy copper heat spreaders, there are small copper squares with long extensions, very cheap, some comes with adhensive heat tape as well. stick them on the VRM individually, and just have a fan blowing directly on it. This will likely solve most of the Overheating problems he be facing.

 

[bburrill2012]

Okay so I put a fan blowing between my cpu and gpu (open air test bench) and a fan blowing toward my cpu cooler/vrms from the top of the mobo.

The NB heatsink is barely warm after playing for several hours and the one over the vrms is cold. But the one over the vrms isnt really touching the vrms so i thinkk it was hot due to the NB one being hot cuz its connected. I touched the only vrm i can and gets a little warm.

Thanks

 

[Bobnova]

They don't really put off a lot of heat generally speaking (on good motherboards anyway), it's more that they don't have a lot of cooling.
By the time CPUs were putting off the sort of heat levels your MOSFETs are putting off at full load with that OC the CPUs had dedicated, decently sized, heatsinks with fans on them.
The mosfets get dubious aluminum jobbies designed mostly for looks, with no fan.

Most MOSFETs these days (decent ones) are rated for 150c, that is really hot. Like nearly instant burns kind of hot. Admittedly starting somewhere in the 100c to 140c range more heat gives them more resistance which generates more heat. Positive feedback loops only have one end point, BOOM!

Anyway, if you can leave your hand on a MOSFET heatsink without getting burnt, it's a perfectly acceptable MOSFET temperature.

As to why they get hot, your CPU is drawing something over a hundred amps. At that level of draw even a little bit of resistance equals a fair amount of heat.
The amazing part is that they don't explode on startup, really. MOSFET technology has come a very long way.

 

[bburrill2012]

Thanks

 

[Buddywh]

for VRM cooling, I have some experience, hope I can help... Just buy copper heat spreaders, there are small copper squares with long extensions, very cheap...

Bluezero: any links to those copper heat spreaders? I got some aluminum spreaders... lots of fins, but still rather small for the amount of heat coming off those VRM's.

The VRM model isn't a transformer but a switch. When a switch is off the current is low with high voltage across the switch; when it's on the current is high and voltage low. I*E=P (power, heat) so when either I or E is close to zero P is too. But during transition from off to on and back, both current & voltage is high, so is P and so lots of heat.

The faster it switches the more often in a given time period it's in transition and processor VRM's switch super fast because a high switch frequency enables the use of physically smaller coils and caps, much more practical on space-constrained motherboards.

Those small coils (along with the caps, they integrate the pulsing output of the switch and turn it into a nice steady DC voltage) have naturally high I2R losses in the winding (I(avg) can be like 100 amps someone said), so they get hot too since I2R=P (power, heat).

That's why the VRM's and coils get so damn hot.

And they will buzz because the windings act like speaker voice coils, being agitated with the pulsing output of the VRM, and move if not tightly wound or encapsulated.

 

[bluezero5]

http://www.performance-pcs.com/cata...54_197&zenid=6d75437b03608fe696300e6f7eebfa80

see a good list of them here.