Enough heatcompound on output transistors?

[MaccAu]

An externally hosted image should be here but it was not working when we last tested it.

Hi All😱

My first time at doing output transistors, a friend helped me with a video of how to do the bias etc on my Technics SU-V8

I fixed and smoothed the scratches and appreciate the help from that thread I made 🙂

I am a bit concerned , i may have used too little heat compound on the ICs. I got myself some Wakefield 120, and a source for Mica washers for MT200.

I know too much is not good either for heat transfer but too little could be a problem also 😕

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.

 

[ostripper]

MT-200 devices have much better thermal resistance than most other
packages.
With 2 screws torqued I'm sure your good. I just use a line of compound
between holes , torque till I see a small bead all around the device.
Wipe away the excess , you are just concerned about the few uM on both
sides of the mica.

OS

 

[Richidoo]

That's plenty. Heatsink grease should be thinnest layer possible, only thick enough to fill microscopic voids that would otherwise have only air to conduct the heat. Any thicker will reduce heat transfer unnecessarily. Use very thin layer, then squeeze out as much as you can. The applied layer is so thin it looks more like clear than white.

 

[Ian Finch]

You seem to have applied puh-lenty of compound there - witness the bead of compound exuded from the washer/heatsink and washer/transistor case gaps as you clamped them up. As you can't see any residual air gaps under the transistors, that's all you can hope for and that is why it's important to have the faces as flat as possible so that the film of compound is as thin and uniform as possible and the amplifier can meet the design specifications safely. Though it's necessary as a kind of gap filler with some benefit to conduction, thicker layers of it just reduce heat transfer. Compared to metal, it doesn't really conduct heat well at all.

You can measure the temperature in operation by using a thermocouple probe with your multimeter etc. applied to the centre top of each device and so check that all devices are shedding heat evenly as they should if they're correctly fitted. There will be some variation due to each transistor's location relative to the others but 10% higher than others could be a warning that more attention is needed.

BTW, what was the source of the transistors.....Chinese Ebay store?

 

[MaccAu]

Thanks OS, I torqued the screws nice and tightly, and the concern was just 2 transistors, and the mica, not looking like they had enough compound

Thanks Richard, its good to know,that thin is fine and I have enough on mine. I screwed them in tightly

Thanks Ian, Good to know, i have enough compound there. They were screwed in tightly to spread the compound. I spread it thinly across the heatsink, then a thin layer using a spare mica insulator on the back of the sankens to cover up any metal (just a thin coat). Variation also in the heatsink compound, but all is fine

No, I dont like using Chinese ebay stores. Mine came from Wagner online here in Australia. The code is YC90, they came in a pack of 10
http://www.wagneronline.com.au

Sam

 

[AndrewT]

.......Heatsink grease should be thinnest layer possible, only thick enough to fill microscopic voids that would otherwise have only air to conduct the heat................

Yes.
exclude all the air in the interface and only just enough to minimise the thickness of the remaining goop.

If the mounting allows, I bolt up gently and rotate the device back and forth to spread and squeeze the goop out as I continue tightening. This seems to bring centrally located goop out to the edges.
Capillary action should hold the goop liquid in the ultra thin gaps.

When I come to remove the devices I find what I would describe as "too little" and this must indicate that the goop really was that thin.

 

[MaccAu]

Thanks Andrew for the reply. What worried me a bit, i may have tightened the screws too much, but i am sure there is a tiny bit of goop

I lined up one heatsink, and I am a bit concerned, maybe it was already like that with the Technics SU-V8, the heatsink shield doesnt cover the black transitor or whatever the tiny things are. Again a thin layer was used, however the screw holes lined up on the heatsink

An externally hosted image should be here but it was not working when we last tested it.

An externally hosted image should be here but it was not working when we last tested it.

Thanks Sam 🙂

 

[mt490]

The best general purpose technique is a small splotch or line in the centre of the surface(s) and then assemble down without re-attachement and minimum side to side motion. You may want to initially perform this on one transistor/surface and detach your work to check you have the right metering of paste to provide adequate coverage - clean the surfaces off and reattach with the correct metering, don't reuse the paste!. This will also tell you if the surfaces are adequately flat and free of protruding defects.

You want only a minimum to squish out as this also means minimum chance of entrained air in the paste, and the dot technique generally results in the thinnest layer possible out of simple application techniques.

The above is just a good technique to use in DIY, but is it actually that critical? Not really, you could easily just dump paste on like a nutter and squish it in and you won't tell the difference without looking for it the majority of the time.

Getting a consistent, pocket free layer is a big thing for power electronics fabs where they're dealing with much larger surfaces - factories use a pasting machine to ensure a good initial metering and distribution to produce a quality product.

 

[roger2]

MaccAu, what is your source for MT-200 mica insulators?

 

[MaccAu]

MaccAu, what is your source for MT-200 mica insulators?

Hi Roger,

I get them in Australia from wagner net, they might post internationally

Wagner Online Store | Wide range of products and accessories

code to search is YC90

Cheers
Sam🙂

 

[Struth]

Hi Guys

Thermal compound is almost universally applied too thickly.

The compound is supposed to fill in microvoids and variations of surface flatness - not to be used like peanut butter in a sandwich.

The correct amount is VERY small. It should be levelled using a stiff flat scraper and the remaining grease should be almost invisible.

Thermal grease compresses over time, so using too much results in loose power devices after a short number of weeks or months. The Hewlett-Packard Journal reported on this back in the 1980s and silicone grease has not changed much since then.

Have fun

 

[MaccAu]

Hi Struth,
Yes i used enough to I think, didnt spill out much, and on the tiny black transistors, just thin, the V8 runs perfectly, in fact it appears to have slight more power than the other V8 of mine 🙂

 

[JonSnell Electronic]

The size of a grain of rice is adequate for most uses. Too messy otherwise.

 

[jneutron]

Hi Guys

Thermal compound is almost universally applied too thickly.

The compound is supposed to fill in microvoids and variations of surface flatness - not to be used like peanut butter in a sandwich.

The correct amount is VERY small. It should be levelled using a stiff flat scraper and the remaining grease should be almost invisible.

Thermal grease compresses over time, so using too much results in loose power devices after a short number of weeks or months. The Hewlett-Packard Journal reported on this back in the 1980s and silicone grease has not changed much since then.

Have fun

Totally totally agree with you. I remember reading that back in the day, but not sure if it was the HP journal or not. I remember using a razor blade to scrape the compound (not cut, but tipped towards the direction of travel. I prefer the blades that are not the scraper blades, they are too flexible.

As you say, it's to fill voids. having any actually in the way of metal to metal contact only reduces the thermal transfer.

The good part of that package is of course the total copper area and screw config. Most manu's consider the baseplate to be isothermal, and by using pressure to squeeze compound out, users do indeed go isothermal. Not the best, but at least it's consistent and usually good enough.

btw, good mention of the ductile squeezout behaviour and loosening.

John

 

[MaccAu]

Thanks John and Jon
The V8 is running super cool with bias set at 40.1 and 40.2. Layer isnt overly thick, and clamped down well 🙂