F5 power amplifier

Moderator
Joined 2002
Paid Member
Just use regular white compound. Arctic Silver will get you a slightly lower FET temperature but is highly capacitive even if it is not conductive. If you feel like spending money, get either Ceramique or MX-3 (Arctic Cooling). These are high-performance compounds but only really useful for high heat loads (100 watts/device). In the case of this amp, you may not see more than 0.5-1 degree difference, if at all.

Also rub down the heatsink with a bit of 100 grit paper on a wood block so the mounting surface is uniform. You should see about 10 degrees difference between the FET tab and the heatsink.
 
Last edited:
Just use regular white compound. Arctic Silver will get you a slightly lower FET temperature but is highly capacitive even if it not conductive. If you feel like spending money, get either Ceramique or MX-3 (Arctic Cooling). These are high-performance compunds but only really useful for high heat loads (100 watts/device). In the case of this amp, you may see more than 0.5-1 degree difference, if at all.

Also rub down the heatsink with a bit of 100 grit paper on a wood block so the mounting surface is uniform. You should see about 10 degrees difference between the FET tab and the heatsink.

Thanks I am going to try it!
I own some MX-2 (Arctic Cooling)!
 
I saw a couple of posts up a balanced f5 - unless I'm mistaken, i believe the one posted by Patrick (EUVL) which the OP ref with the tosh fets is this one ... haven't really seen much written up on this (well not in one place), so perhaps someone who knows much more abt the amp can give some comments re supplies, power o/p, grounding and what degradation (?) occurs if you use same resistors in both top and bottom halves... best that I can recall from reading the archives, they're unequal due to diff transconductance between the P n C channel devices, but wonder how much of this is, well, academic...
 

Attachments

  • balanced_f5.jpg
    balanced_f5.jpg
    128.1 KB · Views: 627
balanced or bridged

as one speaker terminal voltage goes up the other goes down.
The voltage between the two speaker terminals is double what would appear between one speaker terminal and ground.

If the load impedance is the same the current will try to double.

Double the power into double the impedance.

NOT
quadruple the power into the same impedance.
 
as one speaker terminal voltage goes up the other goes down.
The voltage between the two speaker terminals is double what would appear between one speaker terminal and ground.

If the load impedance is the same the current will try to double.

Double the power into double the impedance.

NOT
quadruple the power into the same impedance.

Is there a disadvantage of balanced or bridged?
 
Also rub down the heatsink with a bit of 100 grit paper on a wood block so the mounting surface is uniform. You should see about 10 degrees difference between the FET tab and the heatsink.



Imo, regular everyday heatsink compound works well enough. No need to mess about with exotics.

Kapton is ok, mica is ok, the Bergquist silicone pads are supposed to suck up differences in surface, but I have never had good results with them. Kapton tends to "punch through" so you have to be careful with them, they are thin and fragile in my experience. I have never had good results with kapton either. <waaa waa>

100 grit is rather coarse.
I'd start with 360 or 400 and move down to 1600 for a mirror finish.
You need a dead flat backing for the abrasive, and the abrasive paper should be wet or dry type. I'd work with a bit of water to keep the paper from clogging... clean carefully when done.

First just try it with a mica pad and standard heatsink compound and see what the temps are, before doing anything else.

_-_-bear
 
Proop, I had very similar results when I first built my f5 using silpad insulators. Heatsinks were never that warm and as I turned up the bais my heatsinks remainded cool @ 0.7v I had 41c heatsink and 81c mosfet. Solution I removed the silpads and used mica and standard compound aka goo. Actually I split my mica using a razor blade to get it as thin as possible, following a suggest I read from Mr. Pass. My bias for the last 6mos been set a 0.7v (1.7a @ 23.5v) I typically measure 65-67c on the mosfet tab and 50-55 on the heatsink based on room temp and using a contact therm. Good luck

I want to share some more impressions for the f5.

As I said I use large skins, so after many hours the temperature stays to about 40oC, so after reading I started playing with the bias...

0.60v >>> 40oC >>> 65oC (mosfet)
0.71v >>> 43oC >>> 80oC (mosfet after 30min)
 
100 grit is rather coarse.
I'd start with 360 or 400 and move down to 1600 for a mirror finish.
You need a dead flat backing for the abrasive, and the abrasive paper should be wet or dry type. I'd work with a bit of water to keep the paper from clogging... clean carefully when done.

_-_-bear

Would simply lapping the back of the FET to the matching portion of heatsink work here? Neither needs to be perfect, but only to match each other.
 
as one speaker terminal voltage goes up the other goes down.
The voltage between the two speaker terminals is double what would appear between one speaker terminal and ground.

If the load impedance is the same the current will try to double.

Double the power into double the impedance.

NOT
quadruple the power into the same impedance.

Hmm, i had always thought bridging would give you 4 times the power ...


Question ! How many here are running snubbers in the power supply ? recommended ?
 
Last edited:
Just use regular white compound. Arctic Silver will get you a slightly lower FET temperature but is highly capacitive even if it is not conductive.
Also rub down the heatsink with a bit of 100 grit paper on a wood block so the mounting surface is uniform. You should see about 10 degrees difference between the FET tab and the heatsink.

As has been corrected, 100 grit is TOO rough, start with 400 or higher and progress to 1600.
Could you please explain what you meant by "capacitive" when referring to the Arctic Silver compound? (IT is NON conductive!) Surely your not suggesting that it is electronically capacitive??? Which would surely fry a CPU with Billions of transistors.

Andrew T,
I believe he was referring to Arctic Silver, a thermal conductive compound designed for the mating surfaces of CPUs and it's respective heatsink.

Npapp,
Thank you, I too am using Silpads and I am not impressed by their performance. Tinitus generously sent me some of the Kapton thermal pads but like BEAR, I found them to be easily punctured (and I was careful!) which was one of the causes of my rebuilding my F5 3 times. (my fault for not checking continuity first) (grounded out at full voltage and Bzzzzzzttt.)

I had purchased some Mica pads but they turned out to be 1mm too small on the sides. What is the proper size? TO-247 ?

Ron
 
Sil-pads and snubbers

I've always had good luck with sil-pads, good heat transfer, easy to use and inexpensive. No complaints here.

I replaced the rectifier bridge in my Acurus A200 with an RFB02 board with discrete components. I used MUR6030s with 2 ohm resistors and 100nf caps.
The difference in sound was noticable and immediate.
Was that due to the snubbers? I don't know.

I have used the same recifier boards with my F5 build. I couldn't be happier with the subjective performance.