Obtaining Best Transconductance Using Water Cooled Heatsinks.

[Mark A. Gulbrandsen]

I'm about to build an Aleph X using water cooled heatsinks for the first time. The heatsink will be machined from a 10" X 5" X 1" piece of copper block. 1/2" holes will be bored through the 5" long side and U shaped copper pipes will be silver soldered to these holes to complete the circuit back and forth through the copper block. I end up with one input and output for the coolant. Actually going the copper block route is much less expensive than buying new alumnium heatsinks large enough for this X. Copper cost is just under 50.00 U.S. for this size. Heatsinks at least a hundred dollars per channel for new, plus a long wait to get em. I can make each heatsink in a weekends worth of spare time.

All said and done I remember that Nelson mentioned in one of his posts that the output devices need to reach normal operating temperature for best transconductance and lowest distortion. My concern is that the water cooling may be so efficient that the O.P. devices may not get to optimum operating temp and best transconductance. Has anyone that has used water cooling out there experienced this problem at all? I suppose I could vary the water flow to achieve optimum device temperature, however the whole idea behind doing the water thing is to keep the room and the amp cool. It will also allow me to keep the amp much smaller as well. The X will be P to P wired right on the copper sink using those nifty turrett terminals that I used in my 2's. I can also mount the input devices to the sink and perhaps achieve better thermal equiliberium in a shorter amount of time as an added benefit.

Any comments or suggestions would be appreciated.

Mark

[Variac]

What are you using for a radiator?-how far away? pump?

[GRollins]

About the only bright spot in my recent heat pump crisis was that I came out of it with two more heat exchangers, scavenged from the dead gear. Don't need them at present, but who knows what the future might bring?
Still need to buy a dedicated pump, though.
I haven't tried to modulate the temperature of my rig, but it's easy enough to do. Suggestions:
--Use a valve to reduce water flow. Disadvantage: you're loading the pump unnecessarily.
--Use a light dimmer to reduce the flow rate of the pump. The downside to this one being that the dimmer would throw hash into the AC line.
--Probably the easiest thing to do, at least for me, would be to place a piece of cardboard, wood, or fabric over the heat exchanger to reduce the air flow. Cheap, too. In my case, I'm still using passive air flow, i.e. convective. If you're using forced air, you could always vary the fan speed.
Given the thermal mass of my system, I'd have to wait at least twenty or thirty minutes for things to stabilize, possibly longer. Given the way that folks have to go at this, no two people are likely to have the same system. Because of that, anything I say has to be adapted carefully to any other system. Too much cooling won't hurt anything, but too little could be disasterous. So could water leaks, for that matter. So far, no problems here, though.
One thing I can say is that my system as it stands now easily dissipates the heat from four Aleph 2 channels (about 1200W of waste heat). I've considered the possibility of building an Aleph-X Verson 2.0 with a decent amount of power, but the project will have to wait for other projects to complete and for finances to recover.
Incidentally, I suppose that it goes without saying that the guys who installed the heat pumps were very, very interested in that funny looking heat exchanger with the braided water hoses coming out of it...

Grey

[SY]

You may want to look at junction-to-case and estimated case-to-heatsink thermal resistances. It may well be that you'll get sufficient temperature rise even with a "perfect" heat-sink.

[Mark A. Gulbrandsen]

Yea, actually that gave me an idea regarding the pump flow thing...... adjusting the intake of the pump rather than loading it down by limiting the output flow would be the better way to control via the pump. I believe that the pump should be allowed to run at its normal speed for best motor life anyway. The pump could also be controled with a small variac anyway and be practically noisless. I want to include a hall effect drive motor on my pump though for highest reliability. There will also be overtemp and water flow sensors as well.

The other idea that Grey pointed out and which I will most likely follow though would be to vary the speed of the fan with a temperature servo system fed back from the overtemp sensor. This would be by far the better route IMHO. It not only allows unimpeeded pump operation but it would allow me to actually have an operating temperature adjustment feature. I could set the operating temp of the X with a distortion analyzer and know when the 2nd harmonic is lowest that the best tranconductance point has been reached. I also do believe that the actual heat sink will end up running very cool to just warm to the touch while the chip/device temp is still actually quite hot. Fortunately I have quite a bit of experience with water cooling large film projectors and believe me all the water jackets are copper. Copper conducts much faster than alumnium does. Silver plating the copper substantially increases the heat conduction as well. A silver flash plating is also cheap to do. Coolant will be 50/50 antifreeze and distilled water. Antifreeze also helps drastically at increasing heat conduction.

Any thoughts on advantages/disadvantages of fastening the input devices to the sink would allow for faster DC offset and thermal stabilization? In fact perhaps making a plexi cover that is like a small shallow box to cover the entire parts side of the sink would also be to advantage....or simply potting the entire P to P circuit less O.P. devices....... Either of these would in effect allow every part to operate at the same temperature very quickly.

Mark

[Bakmeel]

Originally posted by Mark A. Gulbrandsen
adjusting the intake of the pump rather than loading it down by limiting the output flow would be the better way to control via the pump. I believe that the pump should be allowed to run at its normal speed for best motor life anyway. The pump could also be controled with a small variac anyway and be practically noisless.

Industrial applications that need to power down their pump, are always encouraged to use speed regulators instead of valves. Valves wil cavitate in the water flow, wear down and eventually need service. It produces noise in your water flow. Next to that, (but that isn't really an argument in Class A amps) if you run your pump with a speed regulator, it saves about 40 - 60% energy.

If you plan to use a hall effect motor (brushless DC machine) you can't use a variac to power down, because the motor is field-vector operated (hence the hall effect devices). You will need a field-vector speed controller. (e.g. complex switchmode stuff) Only normal asynchronous AC motors or DC motors can be used for voltage (current) control.

A simple aquarium pump with a variac or dimmer will work just fine

Bouke

[Circlotron]

Originally posted by Bakmeel
If you plan to use a hall effect motor (brushless DC machine) you can't use a variac to power down, because the motor is field-vector operated (hence the hall effect devices).

Brushless DC machines e.g. PC power supply fans usually respond well to reduced dc supply voltage. So, instead of a variac have an adjustable dc supply. Much cheaper too.

[Coulomb]

You could use a DC Servo Stepper motor and use feedback circuit to control the speed. I thought about adapting one of the new high tech liquid cooling systems designed for PC processors.

Anthony

[kevin gilmore]

Overcool it, then use an appropriate inline cartridge heater
to add back just the amount of heat necessary to maintain
the temperature you want. Pump runs all the time. No valves,
no cavitation, no water noise if you can get all the bubbles
out...

[Mark A. Gulbrandsen]

Yes, I am aware that a Hall effect cannot be controled with a variac. The Hall motors require their own specialized control circuitry. Fortunately I have a bit of experience with them.

Gary, Installing a heater in line is probably not the route that I'd go but I do know what you're getting at. Although the temp feedback would be a bit more complicated it would serve to do two things....sense over temp and shut the amp down in that condition, and it would be the feedbck element in the temp. control circuit...so no clixon sensor would be needed. Water flow loss is another option but in a closed circuit very little water is lost.

Mark