Water cooling, practical matters.

Hi All,

I was just looking at Tom's Hardware page http://www.tomshardware.com/ where they describe a DIY CPU water cooler. With the power levels involved in modern Class A amps, esp Single ended designs, and the problems associated with getting rid of that heat, what are peoples thoughts on water cooling.

With the design described, the pump and radiator could be located a long distance from the amp, thus elminiating any problem with pump and fan noise (Radiator inside for winter, outside for summer). The size of the enclosures could be dramatically reduced with the heatsinks eliminated, and with regards to portability, self sealing, quick release connectors may be used so the tubes containing the coolant may be diconnected quickly and without spillage.

The other option for those who are not billed by the amount of water they use, the water supply could go straight to the amp, and then to the drain, eliminating the need for the pump and the radiator.

Has anybody experiemented with water cooling. I'm especially interested in making it as socially accpetable as possible (i.e. as invisible and unnoticed as possible)

Thanks, Adrian
This topic has come up in the past. One of the best examples is 'http://www.diyaudio.com/forums/showthread.php?threadid=190'. I'm sure that other threads can be found with the 'search' feature of this web site (Thanks Jason).

Grey did a good job documenting what he had done, and there are several ideas from other people in the thread. Personally, I'd like to use a different liquid (Flourinert), but water is the only functional, inexpensive, alternative.

I have personal experience with large (over 100KW) liquid-cooled computers (including liquid immersion), and I can tell you that it's a hassle. You have to consider such things as corrosion, bacteria growth, evaporation and expansion/contraction of coolant. This doesn't include such things as filters and flow control valves. Also, fault detection should be included, or your amp will SMOKE when the pump fails, or a hose breaks.

On the scale that's required to cool an amp for your house, most of the above issues become smaller, but they never go away. There are simple solutions for most of them, but you have to recognize that the problem exists, or you'll never look for the solution.

Good luck.
I am just starting up building a pair of Pass Aleph 1.2's and have thought about the exact same thing. Water cooling is really the most eficient way to get rid of heat. I have also thought of Peletier cooling devices as well but after weighing all aspects of things a large heatsink is still really the practical way to go plus it looks better if the right(designer type sinks are used. With water there is always the chance of water leaking and one HAS to use de-ionized/distilled water to cool with other wise a scum buildup will occur. Other things such as keeping water level topped off and cleaning the cooling system are a reality.
I work with high wattage Xenon lamps and projectors that are water cooled and for these it HAS to be de-ionized/distilled water...nothing else and it has to be extremely pure.
With Peletiers you always will have moisture buildup and some method has to be sought up to deal with the moisture problem other wise you'll have a corrosion problem. Overall, Peletiers were a pretty big failure for CPU cooling.
Mark Gulbrandsen
Salt Lake City, UT
The problem with Peltiers is the same as any active cooling system. All it really does is move the heat around, adding to it in the process. Water cooling is the active cooling method that adds the second least amount of heat (the power into the pump; usually 10-20W). Forced air cooling (low speed) adds the least (power for the fan motor; 2-5W), but it's harder to get a quiet fan than it is to get a quiet water pump.

Peltiers are very limited in the way they can function. They are limited in the number of watts they can transfer, and the temperature difference between the two sides. If properly controlled, they can be used to cool CPUs, but they dump still more heat into the case. This is, in my opinion, a net loss.

Using a Peltier to cool an amp makes no sense. You'd need at least 2 large units in parallel for a ZEN (8-10 for a big Aleph). The power drawn by this is VERY high (1x-2x power moved), and would have to be disipated by the heat sink, along with the initial power (from the amp). Also, you'd have to closely monitor the 'cold' side of the Peltier. If it gets too cold, you'd have to shut down the Peltier (condensation is BAD for electronics). If it gets too hot (Peltier failure), you'd have to shut down the amp. NOTE: Peltiers seem to be less reliable than amplifiers.

If you have room, and can stand the heat, plain old heatsinks are still the best way to cool an amp. If you can move the amp into a remote location (like a closet with ventilation), forced air cooling will work. Otherwise, liquid cooled is the only acceptable way left.

There are chemicals that can be used to keep growth down, reduce corrosion and increase the acceptable heat range. The most common is plain old anti-freeze (50/50 mix with tap water). You'd be suprised at how much heat a truck radiator can disipate, even without a fan.
My Alephs are still water-cooled and running like champs.
No, you need not use de-ionized water. No, you need not top the system off if it's a closed system. I've been going for six months or so, and loving evey minute of it.
So far, I'm still running without forced air on the heat exchanger, but as I'm assembling a second pair of Alephs and plan on using the same water-cooled system, I anticipate that I'll need to install a fan once the second pair is operational.

In the motion picture industry where we use water cooling on large xenon lamps the Xenon lamp manufacturer specifies that one has to use de-ionized/distilled water to keep the lamp warranty valid and to eliminate the possibility of a mineral scum form forming in the cooling circuit. This is a very common procedure in Imax and several 8 perf 70mm projection systems. It is also done in high power UHF TV transmitters as well where any buildup of scum would only lessen the cooling efficiency. Once a scum forms it is really hard to get rid of it. I have seen countless 35 and 70mm projectors film gates that had plugged cooling circuits from not using at least distilled water. In normal projector operations I reccommend using 50% distilled water and 50% antifreze. Not only does this make the cooling more effficient,it eliminates stuff growing in the water. It however only delays the formation of scum but it does not eliminate, or stop it completely. So perhaps you understand that if I did a cooling system it would use de-ionized/distilled water only...same as an Imax pojection system uses. Over the long haul there would be no problems at all. I'd love to see pics of your water cooled amp.
Mark Gulbrandsen
Salt Lake City, UT
re: Water cooling

Another option might be to use forced-air cooling with the fan sitting outside the room where the amp is located and 'pressurized' air is supplied thru some hosing to the bottom of the heatsinks [eg make a small box under the shelf the amp is sitting on and make a slit in the shelf in the position of the heaksinks] and so support the normal convection. I have no figures available about the reduction in R<sub>th</sub>.

It is 'less' critical than water cooling in a sense that no things have to be 'water tight'. In case of failure of the fan, failure of the amp will occur depending on the 'normal' R<sub>th</sub>.

Since other peoples' approaches to water-cooled systems may differ, I can't speak for everyone, but my system--with a single pair of Aleph 2s on it--only runs about 105 degrees or so, maybe less. I don't remember exactly, but it'll be in the Water Cooled thread. The point being, that's less than the temperature of hot water coming out of a tap. Tepid water just isn't subject to a lot of precipitation of calcium salts, etc. Antifreeze isn't necessary because I don't need to keep the water from boiling, and the metal parts of the system are all copper, which is quite stable in the presence of water, else houses wouldn't have 50 year-old copper water pipes. Someone with a system that runs hotter or needs corrosion protection might need to consider the use of antifreeze, but I don't see any benefits--only negatives.
C. Simpson,
I haven't looked into any additives because the system runs very cool as it is. I suppose you could argue that it might be nice to get it down a degree or two more, but it's already running at or below the temperature of my old Thresholds. If, after hooking up a second pair of Alephs (or possibly a SOZ variation), I discover that I've run past the cooling capability of the system as it is, the next thing I'll do is add a fan to the heat exchanger. So far, I'm still running passive convective cooling there.
The idea of running air in from another room (or even a cool basement) has been tossed about before, possibly in the Water Cooled thread, possibly somewhere else. I think it's a good idea, myself. My problem was that I couldn't find big heatsinks for any reasonable sum of money, and I was determined to run the outputs as cool as possible. Unfortunately, not having heatsinks, there wasn't anything to blow air past. Perhaps someone will discover a cheap, reliable source of big heatsinks so I can go air-cooled. Until then, I'll keep fiddling with the water-cooled thing. I just did it as an experiment. It worked even better than I expected, so I'll stay with it for the time being.

water cooled amplifier

i know its a while since this was posted
but have you thought of oil cooling i did this about 25 years ago wth a class a design in fact i cut open the outer case of the output transistors so that the oil actually contacted the junctions
this was a 30 watt class a design that i used to drive a pair of b+w dm6 result fantastic as long as you use the good clean oil it seemed to be trouble free as i remember i made a tank that held about a gallon of oil and totally imersed the output devices in the oil bath it only ever ran warm to the touch never really hot while developing it i noticed how the oil seamed to convect from the junctions
the amplifier was also a self biasing design of my own that i hope to put on the forum when i sort myself out re puting on schematics as i have a few ideas that i have not seen yet
More than 15 years ago I water cooled a high power Crescendo amp. Interesting exercise, wouldn't do it again.

1. Heat exchanger was welded Al channel section, capped at each end with a ledge for the semis.
2. 10mm clear plastic tubing for the water.
3. Purge pump from a discarded dishwasher, scaled back for flow with a loopback around the pump and adjustable 'feedback' valve.
4. Small one litre reservoir mounted outside the room just above the amp to eliminate air pockets.
5. Level sensor in the reservoir (simple IC comparator triac connected to a solenoid from mains water to refill should the level drop).
6. Water lines buried in the ground for a distance of 12 feet to absorb the heat.

The system worked well for about five years. I had no problems with corrosion or bacteria. The heatsink ran below room temperature, but did not seem to damage the amp with condensation. It was cumbersome, not at all portable, and the pump made the all usual cranking noises of a primitive, shaded pole motor. It used a fair amount of additional power, and any fault had the potential to stop it dead, although there were very few problems. Occasionally you could hear a gentle susuration as the water coursed through the pipes and 'sinks. It was oddly distracting when listening to quiet music, something I never got used to.

I would not recommend it as too much infrastructure is needed and reliability problems could lead to catastrophic failure.

In the last ten years a huge variety of quiet fans have become available. These are outstanding, particularly the larger 120 x 120mm DC versions with low speed (typically 1800 rpm). Papst make temperature controlled fans which are around 21dBa, and with careful mounting, and use of a plenum chamber, it's possible to bring fan noise to below room ambient.

Since fans increase heatsink dissipation around three times, it's a very effective way to remove heat. I once tested five mosfets at a combined 175W on a single 150 x 200 sink with a half speed Sanko 120 x 120mm fan. Air flow was very small, just discernible six inches out from the sink. Temperature rise over a long period of several hours stabilised to 27C over ambient; this is better than 0.16C/watt, some three times better than the still air rating. Forced air cooling also distributes the heat more evenly across the sink, which allows greater device packing density.

While PC fans/heatsinks are now very efficient with 50W dissipations commonplace, generally these fans are high speed and fairly noisy. Nothing beats a low speed fan, and a 120 x 120mm fan can exactly match the end profile of two MF18 Conrad heatsinks I use, which are specifically designed for forced air cooling. For projects using SE mosfets - all you Pass guys out there - this is a very effective, reliable, and inexpensive solution.


CPU liquid cooling is absolutely a different story.They use a peltier to cool down the chip below 0 celsius , and a liquid cooler system to remove large amounts of heat.It is only useful for overclocking.
My opinion is that liquid cooling is not practical for amplifiers , unless you have to remove kilowatts of power.
Once upon a time......

in 1979 I build a 2channel poweramp with 80 workinghorses in two half-bridges.......


  • tube902-1.gif
    49.2 KB · Views: 378
The pump and the blower used old gap pol engines, both of them are temperature regulated, but you can hear it working gentle.
The pump motor also stand in oil, together with his regulation powertransitor, this is the small box beside the cooler..
This poweramp sounds nice with my 100W Canton boxes...
I gave them really 1000W in peaks !
So I got the feeling I hear a JBL-Paragon with 50W:D


  • tube902-4.gif
    62.1 KB · Views: 316