EMRFD Message Archive 13665

Message Date From Subject
13665 2017-03-09 15:15:35 nothdurftm IRF510 heatsink insulator
There have been many posts about the need to provide a good heat sink for the IRF510.  A further requirement  being a good thermal path between the IRF510 and the heat sink.  What is the best (or recommended ) insulating material to use?
Mike N.

13666 2017-03-09 18:25:39 Jim Strohm Re: IRF510 heatsink insulator
Personally, I prefer mica, but the rubber insulator pads are more convenient -- and isolating the heatsink from the chassis is the easiest, at least if you don't get issues from the capacitive effect of the heatsink.

Add up your BOM -- insulator (rubber or mica), silicone goop (as required), hardware, insulators for the hardware if it's not nylon .... etc. etc.

I have never tried to calculate the thermal resistance in a system like this.  YMMV.  I sure would like to know if anybody has, and if they can give representational numbers for thermal resistance on the various ways to mount a metal-tab, metal-backed TO-220, it'd help us all.

We don't use TO-220s where I work, so I can't go bug the engineers.  But ... we have a contract programmer who's even older than I am, so I may ask him.  "Everybody knows" that the "old f@rts" have seen everything, done everything, and forgotten more than most youngsters know. 

Too bad Bob Pease isn't still with us.  He'd know.


13667 2017-03-09 18:37:25 kb1gmx Re: IRF510 heatsink insulator
Avid Thermalloy 4880(S,M) 

Those are mica and themasil material, 

There are aluminum oxide pads for to220 cases.

Heavy anodized aluminum works well too.

All must be used with a dab of heatsink grease (the white messy stuff) that is sold in a tube from GC chemicals.  There is Artic-silver but the sample I tested was good thermally but conductive and not suggested.

Don't get to worked up as the thermal resistance of the package (to220) is near 2 degrees C/watt.  That includes the mica insulator which adds little and some aluminum oxide thermal grease (white messy stuff).

The key item is a good hunk of finned aluminum.

13668 2017-03-09 18:52:44 Edwin Carter Re: IRF510 heatsink insulator
Diaper rash ointment works also.

13670 2017-03-09 19:38:20 Ronan McAllister Re: IRF510 heatsink insulator

>Too bad Bob Pease isn't still with us.  He'd know.

Thank you for mentioning Bob Pease; he is one of a small handful of true genius-craftsmen I have been intrigued to follow the past 30 years or so. 

Bob IMHO exuded a unique combination of genius, curiosity, humility, and humanity.



Too bad Bob Pease isn't still with us.  He'd know.

13672 2017-03-09 22:16:11 AncelB Re: IRF510 heatsink insulator
I laser cut mylar (superthin) 0.5mm for optimum insulation and heat
transfer...had some in my Tindie store. Tested to 200V insulation and
about 3x better than sil pads etc.
13673 2017-03-10 05:16:45 bobtbobbo Re: IRF510 heatsink insulator
In cases where an IRF fet is used I have gone to the RD series. The source is the mounting tab and therefore can be directly bolted to the heatsink w/o any insulators. As a caution, I do put a  very thin skim of heatsink compound between the two in case there are any slight surface irregularities on the heatsink. I have used RD16HHF1s successfully, plus they are designed to be RF devices. Just remember to change the pinouts. While not as cheap as the IRFs, their cost is reasonable.

Bob, K1AO
13676 2017-03-10 09:03:58 K5ESS Re: IRF510 heatsink insulator

Well there HAS to be a story behind THAT discovery ;-)   and thanks for the other replies.




13679 2017-03-10 09:51:08 anglerhamming@out... Re: IRF510 heatsink insulator
I wonder if Kapton tape could be used. I use a couple of widths of the 1 mil stuff for masking off adjacent traces while hand soldering, and taping PCP artwork to the yellow transfer "paper" from China to send it through the laser printer. The adhesive is silicone, and Kapton is good 260 degrees c/500 degrees F.  The dielectric strength is 8kV, and the insulation resistance is "1 million mega ohms" from the datasheet.

It is used in thermal insulation, for example by NASA. Maybe I'll tape some on my iron that I use for toner transfer and stick my finger on it and see if it feels hot.

Ted, KX4OM
13682 2017-03-10 10:21:20 kb1gmx Re: IRF510 heatsink insulator
Kapton is a good high temperature insulator with stable dielectric properties but only poor as a thermal conductor.  Compared to Copper, Aluminum its closer to air.

The prefered thermal conductors that are also insulators are mica, aluminum oxide, anodized aluminum,
 and  BeO (which is also toxic as dust!).  

Check here for various substances.

13685 2017-03-10 10:25:51 Thomas Noel Re: IRF510 heatsink insulator
Get that on video.

Thomas W Noel

13687 2017-03-10 11:56:18 Thomas Noel Re: IRF510 heatsink insulator
Zinc oxide in petrolatum base (Destin ointment OR zinc oxide sunblock agent) has been widely recognized as a decent thermal grease since at least 2005. I envision some geek overclocker, alone in his Mom’s basement,  looking at the ingredients list of his jock-itch cream, and thinking AHA! 

Thomas W Noel

13689 2017-03-10 12:20:53 AncelB Re: IRF510 heatsink insulator
To test the thermal transfer of the Kapton do it like this.
Setup a couple voltage regulators like LM7805 etc on identical heat sinks. Have good temp. sensors on hand and have them drive the same loads. Use the kapton on one and a regular sil-pad or other (mica) type as a comparison and then compare the steady state temps of the black part of the plastic regulator casing.

If you do the test with good instrumentation (I used an FLIR E8 temperature measuring camera with 6 digit benchtop multimeters) you can calc. the actual thermal transfer characteristics of the insulator.
The 0.5mm Mylar translates to just 0.3°C-in²/W thermal impedance with no special torquing, better than most other pads you can buy. Certainly better economically than all I have seen. Aluminized Mylar is found in the Apollo missions thermal blankets.

I have Kapton here as well, perhaps I should run a test on it for higher temp application use with SiC semiconductors.
13691 2017-03-10 18:19:38 iam74@rocketmail.... Re: IRF510 heatsink insulator
If I need to heat-sink a TO-220 transistor, I use a mica shield and lithium wheel bearing grease.

The mica is cut down from TO-3 shields (which are fairly plentiful) and the bearing grease
is made for high temperature dissipation. It is probably advisable to set a dab of new grease
out for a while to drain some excess oil.

Works like a charm, but I agree that those adapters made especially for this use are superior.

13692 2017-03-10 18:19:50 anglerhamming@out... Re: IRF510 heatsink insulator
I found some info on the thermal conductivity of Kapton. The DuPont data sheet for Kapton MT (not what I have) gives the thermal conductivity as 0.46 W/m.K and notes that Kapton MT has 3 times the thermal conductivy of Kapton HN (the stuff that I have.)  One of the applications listed for Kapton MT is "Insulation pads-heat sink"

So, the Kapton tape I have is probably not the stuff to use.

Ted, KX4OM
13694 2017-03-10 20:06:00 Jim Strohm Re: IRF510 heatsink insulator

I _cannot_ un-see what you just wrote.

Now every time I look at my specific-for-electronics heatsink compounds, I will see that geek-in-the-basement image.

FWIW, the zinc oxide is jut there to fill in the voids in the various mating surfaces.  The petrolatum is to allow the particles of ZnO to snuggle into the crevices -- and leave a telltale marker of the presence of heatsink compound.

If I were making my own brand of premium heatsink grease, I'd use Krytox (a WONDERFUL fluorine-based lubricant) as the carrier, and seek to find something better than ZnO. 

Problem -- to fill in the voids, smaller particles are better, and the machinery needed to mill any substance fine enough for super-duper heatsink compound is controlled as WMD, because it also can be used to mill anthrax spores down to where they can be dispersed as an aerosol and easily inhaled by soon-to-be dead victims.

I bet we have some semiconductor engineers on here who know about cleanrooms and particle control, who can add their USD$0.02 to the discussion.