Artic silver 5 as heatsink paste for MOSFETS

[NeilP]

I have seen a few discussions about the use of AS5 on CPU's etc

AS5 manufacturer say that it is not conductive, but should be kept away from tracks/ traces IC pins etc as it may be capacitive.

Well without thinking I used it yesterday on the heatsink/ tape on the IGBT's (power output transistors) on my BMS battery alloy shell 2kw charger. I jsut grabbed the first tube of heatsink paste that came to hand

All readings I can make with a Fluke meter seem to show it is non conductive, and I have stripped and rebuilt the unit to check for stray wires but see none.

But the case gets a voltage on it with the AS5 paste.

Just stripped and rebuilt with normal white generic heat sink paste, and no voltage present on the case.

I only found out there was an issue when I got a jolt off the case. I plugged it in ot the mains power and as I did so I touched the outer case of the brushed stainless steel wall socket, while the other had was n the alloy shell case. I got a nasty jolt.

I still not 100 % sure it was the paste, but cant see any other cause.

 

[dnmun]

your case should be grounded for this very reason. you should never have received a shock even if the IGBT was shorted through the collector to the heat sink. the heat sink and the case should be grounded. something wrong when you put it back together.

there is 340V DC on the hiside collector.

 

[NeilP]

The case, from the factory is not grounded. there is ground going to the board from the main side, but no connection to the case. typical chinese production I guess. I did consider adding an earth to the case. but not actioned that thought yet.

I could not determine what voltage there was on the case. I put a multimeter in both AC and DC mode between the case and the stainless steel wall plate of the mains power socket.
In both AC range, it gave an indication outside the 750VAC range on the meter. . I did not touch it again so try and work out AC or DC by touch.

As i stripped it, I did a screw at a time, and could not see any indication of any of them causing a problem. Always possible I missed something

Double checked the film behind the IGBT's all good, no obvious short to the case any where.

After rebuild, with the AS5 removed, all seems good

 

[Punx0r]

AFAIK the paste is never used an insulator, so even if the Artic Silver paste was an excellent conductor, you should still not have continuity between the IGBT tab and the heatsink.

I agree that it seems very odd for a mains device to have an unearthed metal casing. I can't recall ever seeing that before and suspect it may not be legal.

 

[dnmun]

the ground wire in your power cable may not be continuous to service ground.

i don't have my EMC-1000 open right now but i will look or the ground eventually.

it usually is a ring terminal under the screw or nut for clamping bar that holds the transistors or diodes onto the heat sink.

 

[NeilP]

On the EMC-2000 the earth from the mains goes directly to the main PCB. , no ring terminal or any other wiring, just the three core cable in from the mains, with all three wires going to the PCB.


With regard to the AS5 , even on the first build with the AS, I used a multimeter to check continuity between the IGBT tab pins and the heatsink case..there was none.

When powered off there was nothing I could find that made continuity to the case, but when it was powered on there was definitely a potential between it and real mains earth

 

[wesnewell]

AS5 is made with silver in a fluid base. That should tell you something. The best, cheapest, thermal paste I've used is wheel bearing grease. I've used it for years on cpu's.

 

[silviasol]

Silver is conductive but only if it crosses with another of the tiny microscopic silver particles that are in the paste. You can put a gob of it crossing a positive and negative connections and it will not conduct since there will always be spaces between the micro particles. Maybe when it is compressed between the heatsink and mosfets they are connecting due to the compression but there is nothing to conduct electricity ontop of the mosfet.

 

[bowlofsalad]

The best, cheapest, thermal paste I've used is wheel bearing grease.

Proof?

 

[NeilP]

For the unknowing even 'wheel bearign grease' could pose a problem. Some have graphite and other additives in them that could also make them conductive.
Intuitevly you woudl expect that a compound (AS5) made of 99.9% silver particles to be conductive..but it supposedly is nto according to the manufacturers.
As soon as I foudn the problem of th elive case, that is instantly what I though too. which is why I removed it.
Manufacturers state it is not conductive though they do say to keep it away from pins / traces etc due to potential capacitive issues.

I only used it 'by mistake' I have three or four tubes of different heatsink compound ont he bench and it just happened to be the first I picked up.

 

[wesnewell]

25 years of using it on cpu's starting in ~1990. And AS5 may not show continuity with an ohm meter, but put enough voltage to it and those little ions will indeed jump the micro gaps between silver particles. If you've ever seen dc voltage jump through air then you know what I'm talking about.

 

[NeilP]

I'll have to dig out my Mega insulation tester and put some kilovolts across a tin of it and sse what happens.

 

[silviasol]

There are other metal based thermal pastes as well, I have some copper based and aluminum based stuff. Arctic ceramic is ceramic based and you get a huge 20gram tube of it for under $10.