vrm and mosfet cooling

it will depends on actually vrm mosfet, most vrm package has an insulator casing, but some do have metal contact for better thermal dissipation. also be careful not to accidentally short the legs of the chip.

Personally i would use a 1mm non conductive thermal pads in conjunction with the heat sink.
 
There shouldn't be a short circuit or any electrical issues if the paste is not electrically conductive...in this case you are probably safe.

However...why are you doing this in the first place? VRM components can handle much higher temps than the cpu can (well over 110 C). Unless you are pushing LOTS of voltage, the VRM will be just fine.
 


I'm having a downclock problem and voltage imbalances

 


that might have to do with bad controller? which mobo do you have?
 


970a g43
 


when i run prime95, temp is 57 degrees but cpu downclock itself to 1.8ghz from 3.5ghz
 


It would be highly, highly unlikely that 57 degrees is your peak temp while running any version of Prime95. Can you report the temperature during a 15 minute run? (Like the actual temp during the run, not right before or right after).
 


it drops in 2 minutes to 1.8ghz from 3.5 ghz the temperature drops with it. its max temp when test runing is 57 degrees
 
I can all but guarantee that the throttling and downclocking is NOT VRM related.

If you are reporting accurate Prime95 activity, if the MAXIMUM temperature you see during a 15 minute run is 57 degrees, your problem is NOT thermal throttling of any kind. If you said something like "at 3.5 it hits 100 degrees, THEN downclocks to 1.8 and stablilizes at 57 degrees", that WOULD indicate thermal throttling.
 


I dunno...do MSI boards throttle the CPU when the VRM's overheat? My Asus M5a88M did; but I was running a 6300 at 4.4G at the time too.

Putting heatsinks on the FET's can't hurt (just make sure not to short leads of the FET's or adjacent components), that's what I did for my M5a88m. But that won't be enough: you'll also have to put a fan on the VRM to move air around the heatsinks.

3.5G seems like an awful low clock to induce that though, even for an 8 core in Prime95. But maybe.

Are you running an 8 core FX? I don't think i caught that.
 
1. What cpu?
2. Don't use paste on VRM's. Many of those 970 mobo's have offset VRM's, it's only usually about 1mm difference, but that's enough where the paste will become more of an insulator than a transfer medium. Paste needs to be as thin if not thinner than a sheet of paper to be effective. Best to use thermal pads as they'll squash in tight spots but still make contact on the lower VRM's (it's every other one) and still do the job intended. The thermal tape included with those heatsinks may or may not be a good fit, but it's closer to the right idea.

What you can do is actually look at your mobo at the VRM's area. Many of those MSI boards will already have holes next to them as upgraded models use the same board. The upgraded models have heatsinks already, so shop around eBay or Craigslist or similar sites for a 'for parts only' or busted mobo, or one that's dirt cheap and has a factory VRM's heatsink.
 


yes i have fx 8320. it doesn't matter overclocked or downclocked. when i overclock to 4.1 ghz it happens again . when i downclock to 3.4ghz, it dropping itself to ~1.8ghz.
 


The coolers I bought contain a thermal pad. I just want to know if these aluminum plates touch each other short-circuited.
 


Have you disabled APM?

Also check CPU heatsink mounting. It just doesn't sound like the VRM has a chance to warm enough to throttle the CPU.