Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Pass Labs This forum is dedicated to Pass Labs discussion.

Experiment: Which mosfet thermal interface material is best?
Experiment: Which mosfet thermal interface material is best?
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 2nd November 2018, 05:44 PM   #1
Eric is offline Eric  United States
diyAudio Member
 
Eric's Avatar
 
Join Date: Nov 2000
Location: Central PA, USA
Experiment: Which mosfet thermal interface material is best?
Default Experiment: Which mosfet thermal interface material is best?

My motivation:

Basically, I wanted to know two things:
1) what are the REAL differences among thermal interface products for mounting mosfets to heat sinks?
2) what are the REAL differences among heatsink finishs/colors/paints in thermal performance for your amp?

Some Background:
Sorting your way though which thermal interface product to use when mounting mosfets to heatsinks for your next amp can be confusing at best, especially if you are new to this arena. There are a variety of products from which to choose - some are easier to find, some are less expensive, and comparing the performance data for each can be confusing. Part of the problem is that thermal performance data is presented in different formats for different products. Some data sheets use the term "thermal impedance" and present data using units of Cin^2/w or Cmm^2/w while others use the term "thermal conductivity" using units W/mK. While related, these are different measures. This is both confusing and makes it rather difficult to conduct an easy or understandable comparison among the available options.

In addition to the differences in the data sheets and the data they present, other variables are relevant including the mounting pressure that holds the mosfet to the heatsink, relative temperatures of the heat sinks, surface area of contact, etc. In short, you need to know quite a bit in order to make any sense of things. I've been reading about this for years and still don't fully understand every single detail that is involved. (yep, go ahead, poke fun at my expense...)

Specific Materials Compared:
In an effort to help myself and others sort through these differences, I spent some time comparing some readily available thermal interface products: the "traditional" mica wafer & goop approach, various "sil-pad" types of materials including Bergquist K-10, Bergquiest 1500ST, and Keratherm, as well as Aavid 4180G - an Aluminum Oxide Ceramic wafer. Links for each are below and datasheets are attached.

1) "Traditional" Mica insulator (I used the one that was supplied in the VFET2 kit along with some arctic silver compound)

2) Bergquist K-10 (thin brown pad)

3) Bergquist 1500ST (thin blue sticky pad)

4) Keratherm (from DIYAudio store - thin pink pad)

5) Aavid 4180G Aluminum Oxide Ceramic (2mm thick wafer, used with arctic silver compound)

Methods:
In order to explore the two questions above, I used my MoFo Amp as a test platform because it is a simple amp: there is a single output mosfet that attaches to the heatsink. This makes comparing thermal interfaces while keeping everything else relatively constant just about as easy as it gets. In all cases, the test set up was the same:
a) The MoFo amp started with a cold, room-temperature sink.
b) The bias point was measured and adjusted as necessary to be as consistent as possible from run to run.
c) Rail voltage was measured (and confirmed stable) for each round.
d) Ambient room temperature was recorded before I began each measurement and final temperature measurements were made at exactly 60 minutes with a timer.
e) I made multiple temperature measurements, but the trends were best represented by two measures: 1) placing the thermocouple from my DMM into the divot in the head of the phillips mounting screw and 2) pressing it directly on the top of the sink (without touching the actual probe with my fingers) directly above the mosfet mounting location. I did measure the mosfet body itself in several locations. In each case, the mounting screw demonstrated higher thermal rise and greater consistency of measurement.
f) The same torque was used to mount the mosfet for each interface.
g) I worked in a room where breezes from opening and closing doors would be minimized.
h) The furnace and air conditioning did not run during the time that I was recording my data.
i) Mosfet mounting location was prepped by wet sanding with 300, 400, and 600 grit paper wrapped around an aluminum block to smooth the surface.

While nothing is ever perfect, I tried to make these situations as comparable as possible by only varying the thermal interface material or heat sink finish.

In the first round, I compared five different thermal interface materials with the same amp and same heatsink without changing anything but the thermal interface. The interfaces that used arctic silver compound (mica and aluminum-ceramic) were done last to avoid contaminating subsequent materials.

In the second round, I compared three different finishes for the same heatsink profile: raw aluminum (no coating), the same sink painted brown, and the same sink painted flat black. I don't have the ability or opportunity to have that same sink anodized, so I am unable to make any comparisons here.

Results:
I am sure there is room for criticism here, but the following is what I found:

There weren't any "real" surprises (products that I expected to perform better did) but the magnitude of these differences were somewhat smaller than I had expected.

Which Thermal Interface Performed Best?
Traditional mica & goop (arctic silver) performed the worst. The mounting screw exhibited a 40c rise. The sink experienced a 22c rise. This was equaled by Bergquist K-10.
Bergquist 1500ST performed notably better (5c lower temp) than the previous two. The mounting screw exhibited a 35c rise, though the final sink temperature was unchanged from the previous two runs and showed a 22c rise.
Keratherm and Aluminum Oxide Ceramic (with arctic silver compound) performed equally well - both exhibiting a 34c rise in the screw head and a 23c rise for the sink itself.

Which Heatsink Finish Performed Best?
Three configurations were compared: nude aluminum, same profile with a flat brown paint, and the same profile again with flat black paint. Again, there weren't really any surprises:
Flat black paint performed the best with the screw head exhibiting a 34c rise and the sink itself exhibiting a 23c rise
Brown was close behind with a 37c rise at the screw head, though the sink itself was only one degree cooler with a 22c rise
The worst performer was nude aluminum with a 41c rise at the screw head and a final sink temp that was the same as brown paint.

Conclusions:
Thermal interfaces DO make a difference. Duh... but now I have data. There was an overall difference of 6c across the various materials. While this is worth paying attention to, the resulting difference in final heat sink temp was smaller than I expected. I suspect if my meter has resolution to the tenth of a degree there would have been greater variations to see. My DMM only reports integer temperatures in Celsius scale.

Heatsink color also makes a difference. Predictably, darker colors radiate heat better. This made a bigger difference for the heat experienced by the mosfet than it did for the final temperature of the heatsink itself. Again, a thermal probe with greater resolution would probably show more detail.

Is it worth the effort to reduce the temperature of your mosfets by 5c or 6c? I think it is. Most data sheets indicate that a 10c reduction in junction temperature typically doubles the life expectancy of the output transistor.

Comments, feedback, and discussion are most welcome!
Attached Images
File Type: jpg TempAnalysis.JPG (369.9 KB, 477 views)
File Type: jpg Temps.jpg (117.4 KB, 509 views)
Attached Files
File Type: pdf Bergquist K10.pdf (196.7 KB, 15 views)
File Type: pdf Bergquist 1500ST.pdf (48.3 KB, 15 views)
File Type: pdf Keratherm datasheet.pdf (831.2 KB, 17 views)
File Type: pdf Aavid 4180G Datasheet.pdf (159.4 KB, 16 views)
__________________
My DIY Basement Theater, Aleph-X Amps, Avro Open Baffle Speakers, LDR based BA-3 pre, Pearl II, highly modded Technics SL-1200, F4, M2, a40, several ACAs, FHXL, Tang Band 1772 bass-reflex

Last edited by Eric; 2nd November 2018 at 05:47 PM.
  Reply With Quote
Old 2nd November 2018, 06:15 PM   #2
JonSnell Electronic is offline JonSnell Electronic  United Kingdom
diyAudio Member
 
JonSnell Electronic's Avatar
 
Join Date: Dec 2012
Location: The Jurassic Coast, England. GB
All any design engineer needs to do is read the data sheet that is supplied with or before you select the insulator.
All data is supplied by the manufacturer. Takes the guess work out of it.
__________________
Support for Fender, Marshall and all Valve Equipment; Audio Innovations, Audiorama FU29, Quad and Leak. www.jonsnell.co.uk
  Reply With Quote
Old 2nd November 2018, 06:36 PM   #3
gl is offline gl  United States
diyAudio Member
 
Join Date: Dec 2003
Location: Sierra Foothills - California
Eric, thank you for doing this and reporting the results. This is DIY data from a DIY context. I relate well to this type of experiment.

Graeme
  Reply With Quote
Old 2nd November 2018, 07:23 PM   #4
plasnu is offline plasnu  United States
diyAudio Member
 
Join Date: Jan 2011
Location: NY USA
Thank you for the report. I had a problem with Arctic Silver. I do not think it is perfectly electrically insulated. Arctic Alumina is.
  Reply With Quote
Old 2nd November 2018, 07:49 PM   #5
Eric is offline Eric  United States
diyAudio Member
 
Eric's Avatar
 
Join Date: Nov 2000
Location: Central PA, USA
Experiment: Which mosfet thermal interface material is best?
Quote:
Originally Posted by JonSnell Electronic View Post
All data is supplied by the manufacturer. Takes the guess work out of it.
I agree- to an extent... The data is always provided, just not in a form that is easily digestible to everyone. Then, there is always the difference between "theoretical" conditions and "real" conditions. Anyone that owns a car knows this about gas mileage.

How much better is Keratherm (at 35 Cmm^2/w) than 1500ST (at 0.23 Cin^2/w)? How much better is thermal conductivity of 6.5 w/mk than 1.8 w/mk? Turns out, not very much at all - about 1 degree C in a real setting. I can get a 10"x12" sheet of 1500ST for $35 - it can be cut into ~150 pads. That's 23 cents each. Keratherm pads are $2 each. Big price difference for a very small performance difference...

I just wanted to make things tangible for people in a real setting.
__________________
My DIY Basement Theater, Aleph-X Amps, Avro Open Baffle Speakers, LDR based BA-3 pre, Pearl II, highly modded Technics SL-1200, F4, M2, a40, several ACAs, FHXL, Tang Band 1772 bass-reflex
  Reply With Quote
Old 2nd November 2018, 08:34 PM   #6
Itsmee is offline Itsmee  England
diyAudio Member
 
Itsmee's Avatar
 
Join Date: Feb 2009
Its good to know that my box of ceramic insulators are as good as Keratherm pads
  Reply With Quote
Old 2nd November 2018, 08:58 PM   #7
vdi_nenna is offline vdi_nenna  United States
diyAudio Member
 
vdi_nenna's Avatar
 
Join Date: Oct 2000
Location: PA, USA
Experiment: Which mosfet thermal interface material is best? Send a message via Yahoo to vdi_nenna
I though I was the only one using ceramic insulators.
Bothers me to pay $20 for 10 Keratherm pads.
__________________
I build, therefore I am.
  Reply With Quote
Old 2nd November 2018, 09:12 PM   #8
Eric is offline Eric  United States
diyAudio Member
 
Eric's Avatar
 
Join Date: Nov 2000
Location: Central PA, USA
Experiment: Which mosfet thermal interface material is best?
I'm not sure who it was around here that pointed me in the direction of ceramic insulators - it was either Itsmee or plasnu. I was skeptical when I first saw them due to their thickness. But, it turns out that the little buggers work very well!
__________________
My DIY Basement Theater, Aleph-X Amps, Avro Open Baffle Speakers, LDR based BA-3 pre, Pearl II, highly modded Technics SL-1200, F4, M2, a40, several ACAs, FHXL, Tang Band 1772 bass-reflex
  Reply With Quote
Old 2nd November 2018, 09:14 PM   #9
Dennis Hui is offline Dennis Hui  Canada
diyAudio Member
 
Join Date: Aug 2002
Location: Toronto, Canada
Experiment: Which mosfet thermal interface material is best?
Hi Eric,

Thank you so much for doing these testing.

Cheers,
Dennis
__________________
We are fearless amplifier builders. A little heat doesn't scare us at all.
- NP 2002
  Reply With Quote
Old 2nd November 2018, 09:28 PM   #10
danny_66 is offline danny_66  Belgium
diyAudio Member
 
danny_66's Avatar
 
Join Date: Jun 2005
Location: Lierde
Quote:
Originally Posted by vdi_nenna View Post
I though I was the only one using ceramic insulators.
I also use and prefer the AlO3 ceramic insulators,
they're indestructible and much less prone to failure then softer insulators that could be cut by the sharp edges of transistors.
__________________
My RefSpeaker
  Reply With Quote

Reply


Experiment: Which mosfet thermal interface material is best?Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
LM3886 Thermal Experiment (with data) tomchr Chip Amps 406 8th November 2018 07:35 PM
thermal interface materials for transistors Toster Solid State 7 16th June 2017 08:12 AM
Phase Change Thermal Interface Material anyone? nattawa Solid State 10 12th May 2015 11:36 AM
Pyrolytic Graphite As A Thermal Interface Material Mr Evil Parts 6 7th January 2014 12:10 AM
Improving thermal interface of transistor to heatsink. Circlotron Solid State 12 23rd September 2002 03:31 AM


New To Site? Need Help?

All times are GMT. The time now is 04:02 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 14.29%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio
Wiki