F5 power amplifier

Hi,
I don't like the idea of separating the heat flow of the output transistors by using separated heatsinks.

The bias current is sensitive to device temperature.
The output offset is extremely sensitive to device temperature differences/changes.

If you do try to use this method of cooling then look at the manufacturers data and find what Tc and Tj the output transistors can be expected to operate at.
 
Hi,
I don't like the idea of separating the heat flow of the output transistors by using separated heatsinks.

The bias current is sensitive to device temperature.
The output offset is extremely sensitive to device temperature differences/changes.

If you do try to use this method of cooling then look at the manufacturers data and find what Tc and Tj the output transistors can be expected to operate at.

Thanks Andrew. It's just that they're so darned cheap compared to all of the other options I've seen.

What if I were to have two heat sinks per channel, linked with a 1mm sheet of copper, thermal pasted and bolted to the two heatsinks. The mosfets could join to the copper sheet, then to the two heatsinks. Or, as I suspect you will now tell me, will the joint between copper and heatsink not be efficient enough to dissipate the heat?

Alternatively, do you or anybody else have any cheap and cheerful UK Farnell/ RS/Maplins/UK based suggestions?

Many thanks
Lucas
 
If you are sourcing your ClassA amp components at New Retail prices then you can never do it cheap.

If you are prepared to wait while you accumulate suitable secondhand and/or special offers or someone else's scrap you can do it very much cheaper, but it always seems that completion will be after tomorrow.
You have to choose: cheap or quick.
 
Hi,
I don't like the idea of separating the heat flow of the output transistors by using separated heatsinks.
Hi AndrewT,
(with all respect to you) do you mean generally?
What's wrong with following separated solution.
:cheers:
 

Attachments

  • HEATSINK.jpg
    HEATSINK.jpg
    113.8 KB · Views: 528
Lucas,
those little heatsinks have a rating of 1.4*C/watt so for the output fets consuming about 30 odd watts each, the heat rise will be about 30 X 1.4 = 42 degrees above room temp (usually about 25 or thereabouts) so those heatsinks will be getting to be about 65*C and much hotter on a summers day, naturally.

So, really, put these aside for another project and find some heatsinks that are rated at about 0.3 - 0.4 degree/watt for a heatsink temp rise about the 25*C.

Curiously, I run my fets about the mid 50*C mark as I think they do actually sound better when hotter, but many people don't agree with this notion - the main problem is that the heatsinks makes the case really hot and all the components inside will also be really hot and this is bad news for those expensive power caps. Something to keep in mind ....

For your parts bin, get the best transformer you can afford, top quality power caps and find really good shottky diodes and then keep them cool (ie seperate heatsink/bar and don't connect them to the hot chassis) - the power supply is most important for the final sound of this amp.

You may never need another (amp).

... good luck.






9
 
Moderator
Joined 2002
Paid Member
@ma_coule the offset of the F5 has some correlation to the bias level and temperature of each half of the output pair, so separating them thermally may have some implications for output offset. I had asked earlier - and Peter had a very nice solution where he thermally coupled the halves, he used a Conrad MF (IIRC) heatsink for each device, and aluminum bars to couple them.
 
@ma_coule the offset of the F5 has some correlation to the bias level and temperature of each half of the output pair, so separating them thermally may have some implications for output offset. I had asked earlier - and Peter had a very nice solution where he thermally coupled the halves, he used a Conrad MF (IIRC) heatsink for each device, and aluminum bars to couple them.
Hi Sangram,
thank you!
:cheers:
 
Thanks Andrew. It's just that they're so darned cheap compared to all of the other options I've seen.

What if I were to have two heat sinks per channel, linked with a 1mm sheet of copper, thermal pasted and bolted to the two heatsinks. The mosfets could join to the copper sheet, then to the two heatsinks. Or, as I suspect you will now tell me, will the joint between copper and heatsink not be efficient enough to dissipate the heat?

Alternatively, do you or anybody else have any cheap and cheerful UK Farnell/ RS/Maplins/UK based suggestions?

Many thanks
Lucas

I don't like copper<--> aluminum joints, but they seem to be doing just that in computer CPU heatsinks!

If there is a really good reason not to have separate heatsinks for the N and P devices, I don't see it... they're both running very hot, and both are going to be in close proximity to each other, so the thermal environment will be very close...

Otoh, there is no reason not to bridge the two heatsinks with a backing plate. As previously discussed, it is important to have the two surfaces as flat as possible (the backing plate to the heatsink) so that the thermal circuit is low resistance.

I'd use a thick aluminum plate myself... if you look back in the thread there was a discussion of the specifics of this method.

You do need sufficient heatsink surface area for proper cooling... the various manufacturers sites have the math for getting the right extrusion and figuring out the length of a given extrusion that is required...

_-_-bear
 
Thanks Bear.

Has anybody ventured to make a water cooled F5 yet? I mean, it's not such a crazy idea is it? Unless, of course you can overdo the cooling. A tiny and silent submerged pump could pump water around, over a tiny heatsink coupled to the chip, perhaps creating a nice sculpture as the copper pipe runs in spirals through a huge glass bottle of water, outside of the amp (as I have seen with DIY computers). I've got plenty of copper pipe out there in the shed, and a tap just bursting full of water. Hmmm...

Your thoughts, please gentlemen....
 
Thanks Bear.

Has anybody ventured to make a water cooled F5 yet? I mean, it's not such a crazy idea is it? Unless, of course you can overdo the cooling. A tiny and silent submerged pump could pump water around, over a tiny heatsink coupled to the chip, perhaps creating a nice sculpture as the copper pipe runs in spirals through a huge glass bottle of water, outside of the amp (as I have seen with DIY computers). I've got plenty of copper pipe out there in the shed, and a tap just bursting full of water. Hmmm...

Your thoughts, please gentlemen....

the amount of space that would be taken up by a sufficiently silent water cooling setup would be about equal to the space taken up by a sufficient heatsinking setup. Consider the heat output, and the required radiators (and fans), and you're really not saving anything. Plus, the radiators would still be a heat dump into the room.

even if instead of a radiator/fan combo you went with a cooler and ice combo, you're still saving effectively no space, and not getting it any quieter than the passive solution.

or, as it sounds like you're talking, an open-loop setup, would just waste water.

a work of art, however, it could be. If you find yourself heading in this direction, there's a few computer cooling forums that you may want to peruse first.
 
the amount of space that would be taken up by a sufficiently silent water cooling setup would be about equal to the space taken up by a sufficient heatsinking setup. Consider the heat output, and the required radiators (and fans), and you're really not saving anything. Plus, the radiators would still be a heat dump into the room.

even if instead of a radiator/fan combo you went with a cooler and ice combo, you're still saving effectively no space, and not getting it any quieter than the passive solution.

or, as it sounds like you're talking, an open-loop setup, would just waste water.

a work of art, however, it could be. If you find yourself heading in this direction, there's a few computer cooling forums that you may want to peruse first.

You are absolutely right, of course. It isn't a sound idea.

My thoughts came from the shock of having to spend £70 on aluminium extrusion. My idea was that I could make a water cooled heatsink much cheaper from microbore (6mm) coiled copper pipe and a tiny fish tank pump.

I'm going to buy some massive ally instead...thanks for setting me straight.

Lucas