Water cooled audio circuitry

This is an interrim report on the water-cooled Aleph 2 I'm building. The emphasis for the moment being on the water-cooled aspect, not the Aleph circuit.
First, two disclaimers:
1) This isn't going to be for everyone. It's cumbersome, to say the least. Frankly, I'm not sure I will stay with it. It's just an idea that occurred to me since I can't get big heat sinks for a reasonable sum of money.
2) Sorry, but for those who might wish to recreate what I've got here, I'm going to be at a loss for particulars in some cases. In particular, the heat exchanger is from a dead heat pump. I don't know the make or model number. I don't know any reasonable way to measure its efficiency as a separate unit. It's part of the system here and will have to stand or fall as part of the overall system. I will provide physical measurements below.
Here are the particulars of the Version 1.0:
I have one channel of an Aleph 2 built and functional. The quiescent heat dissipation is on the order of 300W for the one channel.
The heat sink system consists of the following (for the moment--this is only a test setup to determine whether I want to proceed). Each bank of 6 output devices is mounted on an 8" long piece of aluminum L 3/16" of an inch thick, with 1" flanges. The devices are mounted on 1" centers spaced from the center line of the L. The L is bolted to an 8" long piece of 2x4" aluminum strut (not anodized) such as you see as framing around doors and windows. The aluminum is 1/8" thick. The 2x4" is capped at the ends with 1/4" thick Plexiglas. One end is tapped for two NPT brass barbs for water in/out. The Plexiglas end caps are glued in place with clear Dow Corning silicone sealant. This is *not* the best arrangement you might wish for. The sealant adheres nicely to the Plexiglas, but only so-so to the aluminum. It's good enough for the time being, but is not something I'd want to use in a 'for real' version.
Coming off of the brass barbs, I'm using braided vinyl hose. Not because the lines are under pressure, but because vinyl tends to sag and crimp when faced with heat. The braid maintains the shape so that the hose remains open at all times.
Now comes the main variable--the heat exchanger. The one I'm using came out of a heat pump. It consists of two separate heat exchangers banked together in an A shape, one on each side. Each end of the 'tent' is closed with thin sheet metal; the only opening is at the bottom. There are three individual 1/4" copper lines running up each side from a plenum (I had to build the input plenums). The three lines from each side meet at the top in a single plenum where the water comes out. Each side of the A is 2" thick, 16 3/4" wide, and 18" tall; filled with the usual thin aluminum vanes that you see on heat pumps and air conditioners. At the moment, I'm running one bank of output devices into each side of the A.
On the output side of the heat exchanger, I'm using a Little Giant 3-MDX. I don't remember how many gallons per minute the pump is specified for, but that shouldn't be hard to find out. I'll try to track that information down, as that will be the critical number. Note that the Little Giant 3-MDX is a *food grade* pump. I removed it from my brewing system. There is no reason to run such a pump in a water-cooled heat sink system; it's just what I had on hand. The water coming from the pump goes back to the aluminum struts via a T fitting to split the water into two flows, one for each bank.
Performance: Excellent. At the moment, I'm only using convection through the heat exchanger. I could easily use a fan to force air through. Moreover, there is little air moving through the heat exchanger due to the fact that it's sitting on a shelf with only the front lip over the edge. The resulting slot is about 1" by 13" compared to about a 15" by 13" opening across the bottom of the A. Obviously, I could increase efficiency considerably by allowing more air to flow though the heat exchanger, but I'm running out of space on my work bench. Add to that the fact that I wanted to 'torture test' the system to see how it would operate under less than optimal circumstances. In spite of being hamstrung in this manner the temperature at the aluminum L right behind the output devices is on the order of 108 degrees F after 2 hours of operation. Ambient temperature is about 72 degrees F.
I've spent about $30 out-of-pocket for the aluminum, the braided line, and the brass fittings. The pump was, I think, about $70 or so, although a non-food grade pump would be much much cheaper. The heat exchanger was free. Call it about $50 if you were to build such a system from scratch.
Now the interesting thing is that you could conceiveably run more than one amp off the same system. For those into solid state, it would not be a major difficulty to "T" out more lines and water cool more channels. That's a *lot* of money you could save on heatsinks.
Noise from the pump: Yes, the pump makes noise. Other pumps may be quieter. I would not want to have this pump in the same room while I was listening. My intention, both for the noise and for heat, is to put the heat exchanger and the pump in another room. I had considered the possibility of a purely convection system, but decided to go with the pump, since I had one on hand.
What I would like to do for a Version 2 is to use copper bars, perhaps 1" wide, 8" long, and 1/4" thick soldered directly to copper pipe. This should provide for far better heat transfer, as copper is superior to aluminum in that regard. It would also be more compact and sturdier than the current setup. The problem will be availability of copper bar. I'll see what I can do.
Oh, speaking of compact...it is. Compared to the size of the passive air-cooled heat sinks you'd need in order to build a circuit like this, it's positively tiny. The entire circuit, complete with aluminum 2x4's, would easily fit in a two-space (3 1/2") rack mount enclosure. The heat exchanger, obviously, is rather large, but can be put somewhere else.


P.S.: Do not fuse the individual rails going into an Aleph circuit...
Version 2.0
I have built copper heat exchangers for the system. They replace the aluminum 2x4" exchangers described above.
The new version consists of a copper bar 1/4" thick x 1" wide x 8" long, soldered (silver solder, in case it matters) lengthways to a length of 1/2" copper pipe with fittings on each end. The resulting cross section thusly: O-
The output devices are mounted on 1" centers, evenly spaced from the center line. (Same as above.)
All else is the same as above.
Performance: On the order of .06 degrees C/Watt. The disclaimer here is that I'm still running the heat exchanger from the heat pump with the same 1" air intake. I'd be surprised indeed if the performance didn't improve with more air flow.
The cost isn't bad at all, really. The copper bars were about $8 ea. The copper pipe is nothing special, the usual stuff you'd use for plumbing around the house. Same 1/2" NPT brass barbs for fittings on the ends.
Mechanically, these are far superior to the aluminum 2x4" exchangers, as all connections are soldered or threaded instead of glued with silicone.
They are also even more compact. The limiting factor (on their sides) is the 1" diameter of the copper fitting to receive the brass barb. I confess to being tempted to place the circuitry in a single-space (i.e. 1 3/4") rack mount chassis. I probably won't--the passive air-cooled heat sinks for the IRF9610s in the front end stick up higher than the output heat sinks...and when's the last time you saw a front end with bigger heat sinks than the output section?
Of course, I could water-cool the front end...


P.S.: The copper is nicely polished and looks pretty sexy...
Worse yet, the 300W represents only one channel. *Sigh* And with summer coming on it's hard to justify, heat-wise, although I shall be glad of it next winter, no doubt.
Pictures...I'd be glad to post pictures, but I'm probably the only author in the known universe without a web site. Sooner or later I've got to do one. My enthusiasm level is somewhat less than low, unfortunately. I'll do something, promise. (Just not sure when. If nothing else, it will give me a way to put up a few schematics, as well.) I know it's hard to visualize things from a bare bones description, because I've run into the same problem myself.
I'll take pictures as I'm building the other channel. That way I'll sidestep all the what-ifs and why-don't-I-trys that I went through to get where I am now.
Oh, yeah...one other thing. I anticipate that the copper version will be easier to mount, mechanically, than the aluminum one.
What ideas did you have? I'm not too proud to take a good idea from elsewhere, and it's not too late to revamp the system if it were to work better and/or cheaper than what I have here.

Take a look at this!

Hi all,

take a look at this:


This guy has DIY water coolers and even heat pipes (a very interesting subject IMHO). He's cooling PCs, but well, we can cool amps with the same technology.

One particularly cool element in his heatsink projects is that he used copper from a roofing supply and this was quite cheap. Hmm....

grataku, would you mind telling us about your idea?
GRollins, thanks for reporting about your experiments. I would also like to see a picture...

Great site. Yes, the guy seems to be working along parallel lines to what I've been doing. However, I'm not set up to mill things--my shop is well equipped only in the woodworking category. Machining things is cool, but beyond my capability as things stand now. (Nelson got by this by buying his own milling machines...but he can justify the expense by doing his commercial product in-house. I have no intention of doing this stuff commercially.)
My goals are low expense, and KISS (Keep It Simple Stupid...). The less complicated this is, the less can go wrong. I'm already *way* out on a limb as far as complexity goes (compared to air-cooled), but I'm making a semi-successful effort to keep things under control in that department. As far as cost, I'm coming in far less than I had guessed that I'd spend. Setting aside the water pump, which I already had on hand, the most expensive thing I'll be dealing with is the braided vinyl line at about a dollar & somethin' per foot...total price determined by the distance you want between your amp and the pump noise (and, arguably, the heat, but that will depend on the season and where you live).
The only tools needed for my way are a tubing cutter (hacksaw in a pinch, but I like tidy ends on my pipes), and a propane torch; neither of which are expensive. Oh, and a wrench or two to tighten everything. Not much to it.
Speaking of computers--somewhere in another thread (X100 back engineered thread?) someone mentioned buying a mess of CPU heatsinks. I don't remember whether they specified the cost, and I know they didn't specify the degrees C/Watt. Since the ouput devices in an Aleph circuit are dropping (at idle) about 20-25W each, I don't know how practical that would be for a class A circuit. (Somebody who's current on PC hardware drop in here and tell me how many watts an average CPU has to dissipate these days. I'm sure it probably said somewhere in that site, but I didn't read the whole thing...) Now, the X runs AB--might be practical there.
I etched and populated the boards for the second channel this weekend. My intentions were to pick up parts for the second heatsink today, but Murphy's Law bit me. May not even get it done tomorrow the way things are looking now. (Grumble, grumble...won't be much longer before I start muttering about Communist conspiracies to keep me from finishing this amp.)
Yeah, I want to get pictures up to help folks visualize what I've done so they can come up with a better way to do it, but I don't have a scanner or digital camera, and I don't have a website. One way or another, I'll get something done, as this is the most economical way to dissipate lotsa watts that I've come up with so far.


modern CPUs dissipate up to 70W. Older ones, for example the Pentium II 500 Mhz, about 30W.
But the heatsinks are used with fans. Without fans i would only dissipate one tenth of the heat in these sinks. Overall i think they are overpriced. An array of them would have unpredictable properties IMHO. A single CPU heatsink has about 0.25°C/W.

I'm with you on the low expense and KISS front.
Water cooling and heatpipes are not exactly KISS. ;-)
A big, slow fan is simpler and cheaper. Of course, not as silent. (OTOH maybe comparable to a pump)
I have such an air cooling system in mind for a bigger class D amp (approx 1kW) with temperature controlled fan speed.

I considered fans, of course, but I'm trying to get rid of a fan (the one on my poor old Hafler 500, which needs to go back into my bass rig), not replace it with another. Passive air-cooled is out of the question simply because big heat sinks are made out of unobtainium.
Another unconventional option I considered was to do air-cooled heat sinks, but place the fan in another room. Huh? How you do that? Easy--use that plastic flexible ductwork that they use behind dryers and such. It should be pretty cheap. Use it to duct air in from a fan mounted elsewhere. However, I couldn't convince myself that I could completely isolate the noise of the fan. Seems to me that a fair amount of the noise would come down the duct. (Stand outside your house and listen to the sound of the dryer...) Still, I offer the idea for anyone who thinks they can make it work.
I got the idea from the big mainframe computers I work with. Although the current generation of IBMs are air-cooled, the industry has gone through alternating cycles of air-cooled/water-cooled. We used to have 3080s and 3090s which were water-cooled. They had TCMs (Thermal Conduction Modules) with water fittings going in and out. I figured that something of that nature should do the trick.
And as a buddy of mine is fond of saying: One you get it in your head, you're not happy 'till you get it in the rump. (Translating roughly as I'll pay dearly for my hubris.)
Sure enough, once I remembered having that heat pump heat exchanger on hand, I was on the slippery slope and gathering speed. The idea is to spend as little as possible so that I don't feel guilty (my Scottish blood won't let me toss money about easily) if I choose to back away and try something else.
Where are you, buddy? What was your idea? Somebody come up with a better/cheaper way to do this, so I can slap my forehead and say,"Why didn't I think of that?"
I'm jealous of the guy on that website--if I had that kind of machining ability here, I'd be doing the fancy stuff regardless of cost.



2000-12-31 9:31 am
Sorry Grey and Tim0,
I disable the email notifications because they fill up my mailbox so I have to check the site from time to time. Plus, I spent the weekend working on my biamp project for a change and I didn't surf the net at all. Grey, if you care to know I just finished the external transformer box (remember that post about transformer noise?) it looks great. I am getting better and better at machining stuff so I can let my imagination loose with enclosure designs.
So more to the point, I am afraid that I have little to contribute to the discussion here, my water-cooling ideas involve big hunks of aluminum and a milling machine...so nothing that would save you guys time or money.
Also, I am not sure that building costs are significant if you look at the big picture. I mean, we are talking about dissipating 600w plus the pump, double that for air conditioning as you need to get the heat out of the house in the summer (and that is assuming a 100% efficiency air conditioning system!).
If one could carry a pair of thermally insulated lines outside the house and run them through a an old air conditioning heat exchanger without a fan that would save some money on the electric bill in the summer.
So, the more I think about it, the less I know where I would put the heat exchanger, and the less I want to built it.

[Edited by grataku on 04-17-2001 at 04:29 PM]
Hi Grey, grataku,

i mentioned the fan cooling in conjunction with a class D amp, which has typically 90 % efficiency (dominated by Ron of the MOSFETs). So, the fan wouldn't need to run unless the amp was asked to deliver lots of power continuously, in which case it dissipates about 80W/ch. The amp isn't actually planned yet, these are just figures emerging from the number of output devices i would have to use for 1kW in 2 ohm.
For a class A amp fan cooling is out of the question if the amp is to be situated in the listening room. (...)

I live in northern Germany.. we don't have air conditioning here and most of the year the central heating system is on anyway. So, apart from the cost (electricity is more expensive than gas) the wasted heat in the livingroom is of little consequence. Cost penalty is lower than in areas with air conditioning. OTOH electricity is quite expensive here (DM 0.30 per kWh, about 15 euro-cent).

The cheapest and yet effective way to cool the water in this area would be to use a pipe system buried in the ground. Never gets hotter than 20 degrees if about 50cm below ground level.
Of course this is lots of work...
Or a pool outside... a barrel with water in it... so that it can be recycled. But i'll be damned if i put water lines through the livingroom, in the backyard and back. :)

A friend of mine once used a total loss system to cool his class A amp which dissipated 600W. The water was used in his garden. He lives in Australia.. unfortunately the amp died when a family member turned on the amp and forgot the water. Note: make the amp switch turn on the cooling, too!
He didn't use alot of water. Imagine it like this:
A rectangular, hollow aluminium profile.
The transistors are attached on the long sides of this.
Inside of the aluminium profile, at the top, runs the fresh water tube with holes in it. The water flows through the small holes directly on the walls on which the transistors are mounted. At the bottom of the alu profile the waste water is collected. According to him, this minimizes water usage and had very good efficiency.

I like the fan/duct idea. The noise problem could be minimized by running a duct inside to the amp and another one outside again. A big duct diameter would help reduce the noise. Big Downside: holes in the house are required.

Grey, don't the mainframe cooling systems use fans to cool the water radiator?

I know what you mean about the E-mail notifications. I had them on at first, but got buried. I had to turn them off.
Glad to hear that your external transformer box is going well. I wish I had room in my shop (and money) for a milling machine. (I'm envious, fella...)
Something I mentioned elsewhere is that water-cooled gives you the luxury of putting the heat anywhere you want. Outside is a great option. In my case, my listening room is in the basement. It stays comfortable in the summer without AC. I plan to pipe the heat into another room (also in the basement), a total distance of about 15' or so. It seems at this point as though I won't be needing to use a fan on the heat exchanger, so the only extra electricity I'll need will be the pump. Come winter, the heat will be quite welcome indeed.
The cooling water went next door to the Energy Plant. Although I've never been over there, they've got what appear to be cooling towers outside which steam quite nicely in the winter, so I'm assuming that they're using evaporative cooling to reduce the temperature. One of these days I may go over there and get a tour.
I was over at my friendly local junkyard (I call 'em that, actually they're metal recyclers; don't do cars) looking for heat sinks in some industrial power controllers. While there I saw them scooping up a big pile of *brand new* coolers of the sort you're describing. Crushing them, of course. I don't know if they were for oil, AC, or transmissions, but they were pretty cute. These were about 12"x18" or so. Given that they sell metal by the pound over there, and that these were aluminum, they would probably have been about a dollar or two each. If I were to use just one I'd probably use a fan, but there's no reason you couldn't series a few and get by without.
Incidentally, to get your heat exchanger for free, call HVAC folks in your area. Ask if you can have the heat exchanger out of a heat pump that they've pulled out of a house. Be sure to tell them you'll do the labor to get it out of the air handler yourself. Most will say no--that they recycle them--but eventually you'll get one who'll say yes. You can't beat free.
Now if I can only figure out a way to get pumps for free...


I know relatively little about this sort of thing, so I'm just floating another idea.

Instead of water cooling, why not use a Freon-like system? It would be self-contained (not requiring fresh water) and the liquid-gas system would not require a pump (just gravity feed: the liquid level in the heat sinks needs to be lower than the heat exchanger).

Possibly one could adapt something from an automotive shop that might not cost too much.

I had considered a convection based system for the straight water-cooled setup I'm using now, but wasn't sure I could get a good flow rate. The interesting thing about convection is that it would be self-regulating (and quieter). When you first turn on the amp, the water would not flow, because there would be no temperature differential. As the amplifier warmed up, the flow rate would increase, running faster as the temperature differential increased. At the time, I was still using the aluminum heatsinks, and the configuration was not going to work well, as there wasn't a clearly defined 'up' and 'down' so that the water would 'know' which way to flow. My current copper pipe sinks would work much better, but this would still require that I elevate the heat exchanger up towards the ceiling. I may yet try this, as the simplicity (no moving parts), reliability (if nothing moves, then nothing will break), and lower cost (still haven't figured out a way to get free pumps) are appealing. For the moment, I'm still using a pump.
As far as Freon goes, I'm trying to stay away from Bad Chemicals. Not to mention that Freon (and its variants) are hard to get. I believe that you have to have a license to buy the stuff now; not like the old days when you could just walk into a car parts store and pick up a can of Freon to recharge your car's AC unit. Granted, you could step back in time and use ammonia or something, but there's still the mechanical system to build, which would require a relatively high-pressure pump, an orifice from which to spray to allow expansion, etc. etc. etc. I think in this case that, although the refrigeration route might dissipate more heat, the basic water-cooled system that I'm trying is simpler and probably cheaper.
But, by all means, toss in any other ideas you may have.
Progress report: As of yesterday, I got the second channel up and running. I was in the process of moving the amps from my workbench to the system when the phone rang. I had an opportunity to go hear a system that includes a nice, fresh pair of Genesis Model 1.1s. Sorry, guys, the invitation won. The amps are still sitting in the middle of the floor...

Water-cooled heat sinks:
As of yesterday, both channels are running in the system. After three hours yesterday (another two hours today), the heat sinks get comfortably warm to the touch--about 110 degrees F. The devices themselves are hot, but touchable. I repeat: the *devices* are hot but touchable. Ever try that with a class A solid state amp? Good way to blister fingers.
Three things to note here:
1) These are TO-220 devices, hence not as good at dropping heat as, say, TO-247s...and they're still pretty comfortable.
2) The above is in relation to an output section roughly 10% over-biased. The output stage of an Aleph 2 runs at 3 amps--these are running at about 3.3A, hence running even hotter than usual (arguably even better sounding as a result).
3) I'm still not using forced air cooling on the heat exchanger. It's doing just fine with convection.
Circuit note: The blasted heat sink system is working so well that the amp doesn't bias properly. The output section doesn't heat up enough to reach Nelson's intended bias current. I had to adjust the bias to make it come up to 3 amps (and overshot, hence the 3.3A, although I'll probably leave it since the output devices are so comfy and warm). Jeez, all life's problems should be so bad...right?
The Aleph 2 circuit:
Fair warning: Those of you who listen with meters instead of your ears may not be happy with this amp. It doesn't have that nasty artificial brightness in the upper midrange and lower treble that many refer to as 'accuracy.' (Only those who never go to hear unamplified music are fooled by this soi disant 'accuracy.' To them I say, get thee to a concert hall. Music is the proper reference--not numbers.)
Tonal balance--This one I'll have to Pass on (okay, I admit it...pun intended, just be glad that I decided not to praise this amp's 'liquid' midrange...). But seriously, I'm using it to run the midrange units in a quad-amp system (see elsewhere on this site for a rundown on the back half of my system. I won't take up the space here. Suffice it to say that the Aleph circuit is driving a pair of Bohlender-Graebener RD-75s...5.3 ohm resistive load, from 250 Hz to 5 kHz, as things stand today.) so it's difficult to comment on bass extension or extreme high end. (I've moved the Threshold S-500 to the subs, replacing the Hafler, and put the tubes on the tweeters.) Overall, this is a warm sounding amplifier, more towards the tube end of the spectrum than the stereotypical solid state unit. To me, that's a plus. To others...you have been warned.
Dynamics--So far, I've listened to rock (virtually zip for dynamics) and jazz (small ensemble, decent but not huge dynamic swings). No classical yet. I have not played it at high volumes yet. So far, at moderate volumes the amp shows no sign of running out of steam. No sense of compression. No collapse of the stage width or depth on dynamic peaks. Disclaimer: The power supply is a bit larger than the stock power supply.
Imaging--On a par with my tube amps, and I mean that as a complement, as solid state doesn't generally image as well as tubes, for whatever reason. However, there's something 'different' about the image that I haven't been able to put my finger on yet. This will take further study. But still...the imaging capability is exemplary. I'm going to have to get back to my workbench, as Nelson is nipping at my heels on the imaging front.
Detail retrieval--Very, very good. I have arguably heard better, but this circuit is among the best. Again, there's something here that I will need to check at length in order to describe properly, but I can tell you this: After listening to Dire Straights "Private Investigations" (Love Over Gold--vinyl, not CD) it is possible to have a completely serious discussion as to exactly what *sort* of glass they broke during the recording of the song. A bottle? A glass ashtray? A glass? A sheet from a window?
Overall rating...well, I'll leave it to my wife (*not* an audiophile)--direct quotes:
1) "Whoa!"
2) "We're not worthy!" (Genuflecting in front of the system.)
3) And last, but not least: "People who say you can't hear image and stuff are *idiots!* They ought to hear this."
I'm not sure I'd have gone so far as to genuflect, but I must say I agree with her, point for point...
Kudos to Nelson Pass for his design, and a separate, but no less heartfelt, thanks for his willingness to let us have *access* to his designs.


P.S.: I have an idea or two in mind to improve the practicality of the heat sinks. Will report soon on the Version 2.1
just a thought

This may not be an ideal solution (for 300W PD) but perhaps worth thought for lower powered amps such as the zen. Sorry if this is a little bit technical and long winded but it is the only way I can explain it.

- The specific heat capacity of water is Cw = 4.187 KJ/Kg.K
- The work (W) required to raise the temperature of a given mass (m) of water by dT is given by:

W = m x Cw x dT (1)

- The time taken to raise the temperature of the water by dT is given by:

t = W/PD (2)

where PD is the power dissipation of the amp.

Just say you wanted to be able to listen to a 10W zen amp with PD = 70W and dT = 25K for 5hrs continuously:

from (2) W = PD x t

W = 70 x 5 x 3600 (3600 = secs/Hr)

W = 1260 KJ

from (1) m = W/(Cw x dT)

m = 1260/(4.187 x 25)

m = 12Kg

Therefore a 12 litre tank is required per channel (1 litre = 1Kg of water at room temperature).
The length of one side of a 12 litre cube (shape is not important, only volume) is approximately 230mm.

Please note the following:

1) This is a very rough approximation of a quite complex heat transfer problem.

2) Inefficiencies between mosfet case and heatsink, heatsink and water and water and tank are neglected for simplicity.

3) If the tank is made from a good heat conductor (-ie- aluminium or copper) heat would also be dissipated from the water to the tank and then to the air which may quite significantly lessen the required mass of water.

4) The heatsink does not have to be very large (about a quarter of what is required for natural convection in air) but must be placed at low level in the tank to aid in convection.

5) Natural convection set up by the varying temperature gradients in the water distributes the heat evenly throughout the water and ensures the heatsink has a constant flow of fluid over it.

So there it is, obviously won't be for everyone but it is just a thought. If anyone is interested I can derive a model of a full heat transfer system incorporating all of the systems listed above to give a more accurate required mass/size of the tank.

Anyway thats all for now

Water Cooling

I got a quote today for 6 heatsinks suitable for for my latest amplifier project and nearly died of shock - they wanted £750 ($1000) for them.

I have looked around at various oil coolers and heater matrices and saw one of 18.5 inches by 6 inches by 3.5 inches - dont know how many of these i would need to dissipate about 300 watts of heat, but i definately don't want to use a fan - i tried it in the past and found it so irritating i binned the amp.

I have experimented with water cooling CPU's on PC's before using freon to cool the water - the compressor was too noisy so that's a non-starter as well. I used a circulation pump from a domestic heating system to move the water around - that was silent and cheap and capable of pumping good volumes of water.

Grey - What is the size of the cooler matrix you use?

Otherwise does anybody know of a supplier of surplus heatsinks in the UK?