VRM thermal paste or electrically insulating thermal strip?

[megatron]

http://i57.tinypic.com/e7igwh.jpg

I had a powercolor 280x which had metal cased vrm (like the 290x does above) and I put thermal paste on it instead of a thermal pad. Would this be the cause of its failure? Is it common that the tiny surface mount resistors near the VRM would be the same height and therefore require the use of a thermal pad?

I've got an Asus 280x DC2T 3GD5 and it has different beefed up vrm with a plastic case but from pictures looks like it could have these surface mount resistors in between the chips which would get covered by the VRM heatsink.

The VRM temps are good but I had to downclock the memory via bios edit since it comes with 1600mV memory voltage as standard with no vram heatsinks.

I'm going to fit a kraken G10 + VRAM heatsinks but I didn't buy any thermal grizzly minus pads when I ordered a big tube of Kryonaught.

Is it possible to break the card due to using non conductive thermal paste? I would have thought the pcb design would allow heatsinks to be placed on the vrm without the fear of a short.

 

[thebennyboy]

non-conductive thermal paste shouldn't do any damage even if it's in contact with the electronics. That said, thermal paste shouldn't really be used for VRMs. You should be using thermal pads for those.

 

[megatron]

non-conductive thermal paste shouldn't do any damage even if it's in contact with the electronics.

That doesn't answer my question.

That said, thermal paste shouldn't really be used for VRMs. You should be using thermal pads for those.

I've used thermal paste on my 6850's vrm's as the vrm cases are plastic and there are no small components in the path of the heatsink. I fitted it with a zip tie or 2 through the holes for push-pins of a non equiped factory heatsink.

Aluminium is conductive ( I checked with a multimeter ), and copper is obviously, the most common heatsink materials. Thermal paste is designed to fill the gaps in contact not isolate contact as a thermal pad does.

My question relates to specifically metal packaged VRMs and the height of surface mount resistors vs the VRMs. I find it a poor design decision to design a heatsink for fitting to an area which has a chance of a short.