Hi Guys,
Since some people are gearing up to build a V5 Wolverine I wanted to share some information which might be beneficial as you gather parts to build your amplifier.
During some testing which pushed the thermal limits a lot harder than usual, I had a batch of silicon insulator pads which were not very good and I lost a couple of output devices so I asked around and did a few searches of what might be a better option for the insulator pads on the ouput devices.
Initially I moved to Kapton (0.06mm) and ordinary (Unick) white thermal grease with an aluminium spreader bar.
This made a big improvement.
However through some discussions I found out Alumina (Aluminium Oxide) pads are much more thermally conductive Kapton tape despite being 0.6mm or 1mm thick Alumina is a far superior thermal conductor compared to Silicon or Kapton.
Over time, I have tried mica, silicon sheet, silicon pads, adhesive silicon pads of varying types, Kapton and Alumina.
More recently I have tried Aluminium Nitride and I would say it is definitely worth checking out.
Aluminium Nitride has considerably higher thermal conductivity than Alumina.
I have in the last few days changed over my EF3-5 test build to Aluminium Nitride insulator pads and Maxtor CTG8 thermal paste. I also use controlled torque for the fasteners. So far so good
Hope this is useful info
- Dan
Since some people are gearing up to build a V5 Wolverine I wanted to share some information which might be beneficial as you gather parts to build your amplifier.
During some testing which pushed the thermal limits a lot harder than usual, I had a batch of silicon insulator pads which were not very good and I lost a couple of output devices so I asked around and did a few searches of what might be a better option for the insulator pads on the ouput devices.
Initially I moved to Kapton (0.06mm) and ordinary (Unick) white thermal grease with an aluminium spreader bar.
This made a big improvement.
However through some discussions I found out Alumina (Aluminium Oxide) pads are much more thermally conductive Kapton tape despite being 0.6mm or 1mm thick Alumina is a far superior thermal conductor compared to Silicon or Kapton.
Over time, I have tried mica, silicon sheet, silicon pads, adhesive silicon pads of varying types, Kapton and Alumina.
More recently I have tried Aluminium Nitride and I would say it is definitely worth checking out.
Aluminium Nitride has considerably higher thermal conductivity than Alumina.
I have in the last few days changed over my EF3-5 test build to Aluminium Nitride insulator pads and Maxtor CTG8 thermal paste. I also use controlled torque for the fasteners. So far so good
Hope this is useful info
- Dan
I am using some which I bought to try from Aliexpress. I normally use Mouser, Digikey, RS and Element14 for parts but in this case I didn't.
Make sure you find some which are the right size for your ouput transistors and have a mounting hole.
Also thicker I.e. 1mm wil be fine for up to EF3-4 and will have less capacitance to the heatsink.
For EF3-5 I would suggest to use 0.6mm AlN for thermal performance reasons.
I hope this is not against forum rules:
I just found this on AliExpress:
AU$31.00 | 50pcs ALN Tablets Aluminum Nitride Aluminum Nitride Ceramic Sheet High Thermal Conductivity Insulating Sheet Thickness0.6mm/1mm
https://a.aliexpress.com/_ms4gmVz
Make sure you find some which are the right size for your ouput transistors and have a mounting hole.
Also thicker I.e. 1mm wil be fine for up to EF3-4 and will have less capacitance to the heatsink.
For EF3-5 I would suggest to use 0.6mm AlN for thermal performance reasons.
I hope this is not against forum rules:
I just found this on AliExpress:
AU$31.00 | 50pcs ALN Tablets Aluminum Nitride Aluminum Nitride Ceramic Sheet High Thermal Conductivity Insulating Sheet Thickness0.6mm/1mm
https://a.aliexpress.com/_ms4gmVz
Hi Guy's,
I just wanted to give a massive thank you to @danieljw for his exceptional research and testing of the thermal interface pads for the V5 Wolverine. Thank you for sharing your journey from silicon pads to Kapton, Alumina, and now Aluminium Nitride (with Maxtor CTG8 thermal paste), is a goldmine of knowledge for optimizing thermal coupling, heat transfer, electrical insulation and lowering output capacitance of the output transistors.
Guy's I know first hand that, this work by Daniel has spanned several weeks, involved the meticulous task of removing and re-soldering the output transistors to accommodate the thicker AlN pads—definitely not a 5-minute job! Dan’s thorough testing and measurement under extreme thermal conditions and his practical sourcing advice (like 0.6mm AlN for EF3-5 builds) are a huge leg up for the community. Your dedication and generosity in sharing these insights are truly appreciated. Thanks again for helping us all build better amps.
Cheers Stuartmp.
I just wanted to give a massive thank you to @danieljw for his exceptional research and testing of the thermal interface pads for the V5 Wolverine. Thank you for sharing your journey from silicon pads to Kapton, Alumina, and now Aluminium Nitride (with Maxtor CTG8 thermal paste), is a goldmine of knowledge for optimizing thermal coupling, heat transfer, electrical insulation and lowering output capacitance of the output transistors.
Guy's I know first hand that, this work by Daniel has spanned several weeks, involved the meticulous task of removing and re-soldering the output transistors to accommodate the thicker AlN pads—definitely not a 5-minute job! Dan’s thorough testing and measurement under extreme thermal conditions and his practical sourcing advice (like 0.6mm AlN for EF3-5 builds) are a huge leg up for the community. Your dedication and generosity in sharing these insights are truly appreciated. Thanks again for helping us all build better amps.
Cheers Stuartmp.
Not sure if anybody does this for heat dissipation but my plan was always to mount the transistors onto a large, thick, pure copper plate. That plate will be mounted to the aluminum heat sink. All mounted with boron nitride paste inbetween with a protective coating for the non contacting copper surface.
Copper pulls heat way faster than aluminum. The copper will be about as hot as the transistors. With the plate, you have a much larger surface area to the aluminum heat sinks.
The only downside is cost. But I have a 1/8" thick copper plate at the shop so costs me nothing.
We do this copper/aluminum pull/push thing on thermally difficult projects. It works really well. It's all about conductivity of heat and surface area.
Link to the boron nitride paste we use on customer projects. Embedded team says this is the best stuff.
https://www.mouser.com/ProductDetail/Chemtronics/CW7250?qs=IjV9znPSaN2bVYU6q2%2By8g==
Copper pulls heat way faster than aluminum. The copper will be about as hot as the transistors. With the plate, you have a much larger surface area to the aluminum heat sinks.
The only downside is cost. But I have a 1/8" thick copper plate at the shop so costs me nothing.
We do this copper/aluminum pull/push thing on thermally difficult projects. It works really well. It's all about conductivity of heat and surface area.
Link to the boron nitride paste we use on customer projects. Embedded team says this is the best stuff.
https://www.mouser.com/ProductDetail/Chemtronics/CW7250?qs=IjV9znPSaN2bVYU6q2%2By8g==
Hi Bmsluite,
That paste looks ok but at 1.85 W/mK it will not compare well to the Maxtor CTG8 at 3.85W/mK. (Manufacturer claims 12.8 but it is hype)
I did try some even higher performing paste than this but it was not really suitable for this application since it was.too putty like to spread thinly enough, even after warming it up substantially with a heat gun.
Fyi both will likely be completely fine in this application.
Also some people go this way with copper but it is unnecessary and adds another thermal interface between the heatsink and the transistors. What it will give you is a more even heat distribution, which is why in some high powered applications (i.e. 2kW linear smplifiers) people go this way because the devices contact area is realtively small compared to the heatsink amongst other reasons, "Heat spreader"
I am sure you know all of this any how but I thought it worth mentioning.
- Dan
That paste looks ok but at 1.85 W/mK it will not compare well to the Maxtor CTG8 at 3.85W/mK. (Manufacturer claims 12.8 but it is hype)
I did try some even higher performing paste than this but it was not really suitable for this application since it was.too putty like to spread thinly enough, even after warming it up substantially with a heat gun.
Fyi both will likely be completely fine in this application.
Also some people go this way with copper but it is unnecessary and adds another thermal interface between the heatsink and the transistors. What it will give you is a more even heat distribution, which is why in some high powered applications (i.e. 2kW linear smplifiers) people go this way because the devices contact area is realtively small compared to the heatsink amongst other reasons, "Heat spreader"
I am sure you know all of this any how but I thought it worth mentioning.
- Dan
Last edited:
I'm a slave to data. I'll grab some the Maxtor. I'll give some to the embedded team for them to try out as well.
There is better out there but usability and availability were factors. Nice to have an embedded team at your disposal !
I rely heavily on those guys for these guys of projects. I've learned a lot from them but not enough to go it alone. Nicest guys in the whole company I think
Igors Lab for thermal compound testing - note the target use case is CPU coolers however a lot of useful data there. We don't want to use thick thermal CPU pastes on mica sheets as it could end up being too thick and cracking the mica when compressed, but for alumina or AlN ceramica pads just target the more viscous pastes. I see a lot of similarities to CPU cooling thinking of every transistor as a little CPU, the difference being we aren't tightening down with as much force as a CPU cooler does. Another interesting phenomenon to avoid is the "pump out" you get from the paste thermal cycling.
MistyBlue's thermal pad testing - a fellow forum member recently did a comparison exercise with interesting results and has also kindly allowed us to include the document with the Wolverine documentation.
MistyBlue's thermal pad testing - a fellow forum member recently did a comparison exercise with interesting results and has also kindly allowed us to include the document with the Wolverine documentation.
Thanks for the links Mainframe both of these sources have some excellent data.
Basically just reinforces the boron nitride paste we already use and the viability of the copper pull/push setup though his copper heat pipe setup was pretty flawed and not optimal. You need a plate that has a much large surface area of contact between the copper and the aluminum.
I'm glad to see that aluminum oxide insulators with paste are being discussed here. I started to use them for my class A builds, which subject the OS transistors to much higher abuse levels than Wolverine ever would, several years ago and never turned back to mica or Keratherm.
All my builds use them now, 0.6mm thickness provides enough durability and fantastic heat transfer.
The only issue with using rigid insulators is they will crack if the heatsink is wonky.
Looking forward to breaking ground on my V5 Wolverine!!
Awesome research, development, verification and documentation from all of the Wolverine Team fellas, mucho appreciated!!
All my builds use them now, 0.6mm thickness provides enough durability and fantastic heat transfer.
The only issue with using rigid insulators is they will crack if the heatsink is wonky.
Looking forward to breaking ground on my V5 Wolverine!!
Awesome research, development, verification and documentation from all of the Wolverine Team fellas, mucho appreciated!!