heat conductive paste

[seavote]

a quick question. i'd like to install the regulators in my D-aoc tomorrow. is the heat conductive paste used between regulators and heat sinks in audio electronics the same stuff used in computers? i have some of the pc stuff already. do i need to buy something different?
thanks

 

[Emperor-TK]

Odds are that your PC heatsink paste is silver loaded, whereas many general electronics pastes aren't. In other words, you already have the good stuff, use it. I don't know the D-OAC design though. I am assuming you don't need the heat sink electrically insulated from the regulator. In that case you would also need mica or a sil pad.

-Chris

 

[seavote]

thanks emperor. 2 of the 3 regulators in this build need to be isolated so i'll have to wait till i aquire the correct paste for the job. i do have this sleeve(see photo) that came with a heatsink i salvaged from a switching power supply.

i'm thinking it's for insulation, but do i need to use paste with it? there was none in the previous use. i'm thinking i'd try it with the heater leg of the psu. 9V down to 6.3V. sound ok to you?

 

[Emperor-TK]

I can't really make out what the material is, it almost looks like heat shrink tubing. Was it between the heat sink and the regulator in the other power supply? In the DOAC you haver 9v to 6.3, but at what current? How many heaters are being regulated? If it's anything like Analag's Poorman, that's not an insignificant amount of wattage being dissipated (that would be about 15W if it were a Poorman).

Without knowing exactly what the material is and what it's thermal conductivity is, I wouldn't use it, especially when sil pads are so cheep. Sil pads are more thermally conductive than mica, and don't need any other type of grease with them. Despite the fact that CPU greases are loaded with silver flake, they aren't very electrically conductive (stick an ohm meter in some Arctic Silver and see what I mean). You can probably use silver grease with a mica insulator without much worry about electrically shorting from a sloppy install, but I won't promise that it won't get leaky with time and temperature if the "grease" has any high boiling volatiles in it. Regardless, a sil pad is neater, will probably still give you better thermals than mica/silver anyway. If you are only dissipating a couple of watts, then dry mica would be fine too, forget the grease.

-Chris

 

[seavote]

9V to 6.3 at 2.5 at 2.5A. i just checked out sil pads. they're available and cheap so i'll go with them. seems like the easiest and best(reliable) solution. so i'll go with them. all these little things that you guys know and take for granted have to be learned at some time. every step of the way with every build(all 3 of them, this is 4)i continue to learn and better understand what i'm doing. this forum is great . thanks emperor and thanks to all those who know what there doing for helping those of us who are trying. sorry for the rant. it just hit me again what a great forum this to be part of.

 

[Svart]

watch out with sil-pads..

They have a tendency to "squish" down around the screw holes (if you use them like that..) and make contact anyway..

ask me how I know..

:roll:

I try to buy insulated silicon now instead.. much easier for the negligible price increase.

 

[tv]

the "white" paste would be O.K. IMO.

like the ceramique http://www.arcticsilver.com/ceramique.htm - or misc. OEM stuff.


the "silver" stuff OTOH - I wouldn't use it!

 

[Emperor-TK]

[quote author="tv"]the "silver" stuff OTOH - I wouldn't use it![/quote]

Why?

 

[tv]

http://www.arcticsilver.com/as5.htm

Features:
Contains 99.9% pure silver

:!:

 

[KingKai]

TV,

Are you allergic to the silver, or the high thermal conductivity? :?:

iirc, my large jar of white Wakefield heat sink compound contains a large percentage of silver. I believe this is the reason it is white. :thumb:

Peace - Irv

 

[Emperor-TK]

For the sake of discussion, here is my quick background. I've been making electrically conductive silver inks for the past 10 years. I also used to work for the company that supplied silver flakes and powders to many of the thermal grease manufacturers.

The performance of silver is probably not necessary in this application but again, the silver is "the good stuff" with no real down sides (including cost, check Newegg for some comparative pricing). Someone would need to explain to me why a ceramic based grease would ever be preferable to a silver based equivalent formulation. Compare the performance specs and pricing per gram of Artic Silver 5 and Ceramique. The Ceramique is more expensive with poorer performance.

King Kai, any white heat sink compound would not contain large amounts of silver. It would be an oxide based product. Silver pastes with large particle sizes have a metallic sheen to them. As the particle size decreases, it takes on a matte gray color, eventually turning brown to purple to black at <100nm particle sizes.

-Chris

 

[tv]

on the same page that I linked before:


Not Electrically Conductive:
Arctic Silver 5 was formulated to conduct heat, not electricity.
(While much safer than electrically conductive silver and copper greases, Arctic Silver 5 should be kept away from electrical traces, pins, and leads. While it is not electrically conductive, the compound is very slightly capacitive and could potentially cause problems if it bridges two close-proximity electrical paths.)


It's only me, but I'd sleep much better with a ceramic, nevereverevenslightlyconductive paste on my stuff. Just a hunch, though YMMV


http://www.computing.net/answers/cpus/arctic-silver-5-destroyed-my-mobo/13783.html

google for more stories ...

 

[seavote]

funny but after my third post. i clicked send and immediatly thought it would be funny to make another post immediatley after. it would have read:

OK now its time for someone to post that they disagree with emperor, he is totally wrong ang really confuse me!!


i should have known. but these discussions are always informative and in the end we all develop our own preferred methods. thats what makes this forum great. thanks for all the input.

 

[Emperor-TK]

Once again, I've taken a simple topic and turned it into a long winded discussion. :green: Sorry.

Capacitive issues don't apply here, we are regulating DC. Any AC noise transmitted from the Vreg to the heatsink through the very poor capacitor wouldn't be a high enough voltage to worry about. Also, I did caution about theoretical issues with silver greases and sloppy installations in my second post. I wouldn't recommend spreading it across the top of your CPU socket like the guy in your link. I also wouldn't smear it across the high impedance section of a microphone PCB. Used in it's typical applications without being unduly messy, I don't think that there is anything to worry about. I also fill my car up twice a week with a highly volatile, explosive material. So far, so good. You also can find plenty of stories on the internet where this hasn't worked out either. They won't keep me from driving (but I might cut down on smoking weed at gas stations ) :wink: . I'm also not going to switch to the more expensive, less efficient horse because there is a minute chance that I will have a gas station fill-up accident.

So my take on the options discussed:

1) Sil pads - They are super convenient and work well enough for their intended application. I've used them without problems, but after reading Svart's post, I'd be careful not to crank down on them too tightly. I'd also make sure that the plastic screw insulator is seated properly before tightening.

2) Silver grease/mica - buy name brand grease. Don' be sloppy and everything will be fine. If you are sloppy, you will still probably be fine. There are a few places not to use it (like between a PGA socket and a processor). Most applications should be worry free. If you already have white grease, don't bother unless you need high performance (like an 80W P4 processor).

3) White (oxide) grease/mica - just as inconvenient as silver grease, costs more, doesn't work as well. Most applications that we care about won't require it. If you already have silver grease, don't bother unless your application is near very sensitive components.

4) dry mica - super convenient, excellent insulator, probably good enough thermal conductor (including the trapped air) for low performance applications (like this one). I'd still use grease unless I ran the calculations with the three thermal resistors in series and was sure that I would dissipate enough heat (1 micron air/50 microns mica/1 micron air). So much for the "super convenient" part.

Note that neither the silver grease, nor the oxide grease will be a sufficient electrical insulator without an interface material like a mica pad.


-Chris

 

[tv]

tis your gear, do whatewer you want..

 

[KingKai]

[quote author="Emperor-TK"]

King Kai, any white heat sink compound would not contain large amounts of silver. It would be an oxide based product. Silver pastes with large particle sizes have a metallic sheen to them. As the particle size decreases, it takes on a matte gray color, eventually turning brown to purple to black at <100nm particle sizes.

-Chris[/quote]

Chris,

There were two mistakes in my failed attempt at humor.

1) I was equating "silver" with "silver oxide".

2) I thought the oxide in the Wakefield 120 compound was silver. It is zinc oxide as listed on the MSDS from Wakefield.

Sorry TV, no insult was intended toward you nor misinformation.

At least I learned something

Peace - Irv

 

[PRR]

> I've been making electrically conductive silver inks for the past 10 years.

My understanding is: that's not easy. You grind silver to powder, mix a binder, apply, and you get a LOT of voids, poor conductivity, even with all your diligent work.

Or at least that's one guy's theory how to improve solar cells. He says the collecting stipes are blocking light, and "need" to be that wide to carry the miniscule current. He's found a new silver-ink (you got something new?) which can be built-up instead of wide, block less area.

(Why can't they extrude silver ink around a hair-fine wire and lay that on the Silicon? The ink just joins the Si surface to the copper. The copper carries the current lengthwise. Too many steps for a product which already costs too much?)

My take is, you need insulation, use insulation. You need grease, use grease. But don't use so much grease it could possibly overwhelm your insulation. Not only wasteful/costly, but is a royal pain for the poor tech who someday has to replace the under-specced transistor.

> it almost looks like heat shrink tubing

I seen stuff like that which was NOT heat-shrink, was pliable, like sil-pad. Sil-pad would be costly to use all-around the part, when we only need the sil-stuff on the pressure area, and can use cheaper stuff for finger protection. But maybe the one costly part is total cheaper than a costly pad and a cheap finger-guard and two installation steps.

I'm a mica and zinc-oxide and BIG bolt guy myself. I think the joint should be very good before the grease. (I don't like TO-220 packages, think even the mighty TO-3 could stand a few more bolts.) Grease helps, but is never the solution, unless you are dying to get 2% faster overclock in your shoot-em-up PC.

For general DIY use, if you have any worry about heat transfer, don't cheap-out, use a bigger sink or multiple devices. In production, you'd want to do your sums before committing to a big greased sink, but you also have to factor worker slack-off which may force a better cooler just to cover Monday Mistakes.

I trust oxide greases around ignition systems and CRT supplies; I might be leery about 1,000+V on a metal grease. I'd test before using thermal grease in very-hi-Z systems (but they rarely have any need for thermal conductivity, and spark plug grease is cheap enuff). I'm probably wrong.

> Capacitive issues don't apply here, we are regulating DC.

To be pedantic: we are regulating-out the variations, so it must be an "AC amplifier" in the regulator, and capacitance matters.

> Any AC noise transmitted from the Vreg to the heatsink through the very poor capacitor wouldn't be a high enough voltage to worry about.

Agreed: a 6 ohm 60Hz-1MHz regulator won't be bothered a bit by any picoFarad added by a glob of metal grease; or if it was, then stray layout changes or dampness would also cause trouble.

> White (oxide) grease/mica - just as inconvenient as silver grease, costs more, doesn't work as well.

You say White costs more than Silver??? (Or do you mean Name Brand white versus PC-commodity Silver?) Impossible to tell at my level, a few grams a decade, since the $2 selling price is all packaging, dealer cost/profit, and marketing. Clearly most folks here will never need to care about the cost. Still it is amusing to think the PC gang is sold Silver, not because it is pretty, not because it works better, but because it is cheaper per barrel?

I'll tell you what ruins all big-sink and super-grease considerations. One good speck of dirt (or burr) on the interface makes the thermal joint nearly useless. I changed a CPU. The chipset controlled fan by CPU temp. Normally it starts fast and goes slow in a second, stays slow until I run a stress benchpark. This time, the fan went to FULL speed after 5 seconds. Huh? That would be how long it takes an UN-cooled P4 to overheat. Shut-down, cleaned the joint better, the thing hasn't got hot since, steady 51C, "half hot" for simple silicon.

 

[Emperor-TK]

>>>You grind silver to powder, mix a binder, apply, and you get a LOT of voids, poor conductivity, even with all your diligent work.

Close. Pigments are ground for printing inks. However, conductive inks almost always start with a metal that was chemically precipitated as a very fine power from a metal salt (20-nm to 5-microns). Often this powder is "flaked" by rolling around in an attritor or ball mill with milling media. One key difference with an ink designed for traditional solar applications is the presence of glass powder. The glass is required to etch into the silicon when the silver is fired to aid in current transfer. For back side electrodes, the glass is optional. In general, the four classes of conductive inks are (in historical order):

1. Thick film inks- Contain glass as a binder. Need to be fired above the glass' melting temperature (around 450C). Used for solar, LTCC capacitors, and hybrid ceramic packages. Resistivites ~3x bulk silver are possible.

2. Polymer thick film (PTF) inks- Contains polymer as a binder. The polymer can be cured either thermally or with UV (less common). The shrinking of the polymer during the drying/cure pulls the flakes together to form a continuous conductor. Resistivites of ~10x bulk are possible.

3. Reactive inks - These inks contain a organo-metallic or metalo-organic binder that can thermally decompose into a metal in order to fuse the flake/powder together. Resistivities of ~6x bulk are possible.

4. Reactive nano inks- These inks are made with highly surface active nano particles that can sinter at temperatures as low as 80C. Resistivities of ~2x bulk are possible (limited only by the resulting porosity of the cured ink).

I used to work on type 3 inks, now I work on type 4 inks. You can make a simple low performance type 2 ink by mixing some silver flake/powder with clear nail polish or white glue.

>>>the collecting stipes are blocking light, and "need" to be that wide to carry the miniscule current. He's found a new silver-ink (you got something new?) which can be built-up instead of wide, block less area.

This has been the biggest request from ink manufacturers for silicon based solar. My nano inks are not at all suited for this application. There are other solar technologies of interest though. Check out Konarka as a starter. OPV manufacturers and researchers are making some very large promises for the future.

>>>(Why can't they extrude silver ink around a hair-fine wire and lay that on the Silicon? The ink just joins the Si surface to the copper. The copper carries the current lengthwise. Too many steps for a product which already costs too much?)

Bingo. A 20 micron stainless steel mesh costs >$10/sq ft. A 20 micron printed silver mesh (if you can do it) costs <$0.05 per square foot.

To be pedantic: we are regulating-out the variations, so it must be an "AC amplifier" in the regulator, and capacitance matters.

But would capacitive coupling between the regulator and the heatsink matter? Short of dropping a glob of silver grease on the PC board, is there any real danger here that I am missing? (not a rhetorical question btw).

>>>You say White costs more than Silver???

Just comparing two similar formulations at retail pricing for 2.5 grams (Artic Silver 5 and Artic Silver Ceramique). I'm sure you could find an industrial bucket of simpler oxide grease for a fraction of the cost of an equivalent silver product.

-Chris