[Amps] Solid state amplifier cooling

David Lisney g0fvt at hotmail.com
Thu Jun 6 13:34:28 EDT 2013


Hi Bill, you raise some significant points, in the overclocked pc arena a few people got obsessed with making optically perfect contact patches between the copper cooling "jackets" and the relevant patches on the cpu, they then used silver loaded heatsink compound. Thermal resistances were very low, as you can appreciate many CPUs were dissipating over 100w through a tiny contact patch, much smaller than most driver transistors in the average 100w HF rig.

I have no experience of any really advanced cooling systems in homebrew gear. But I personally believe attaching the devices to a good billet of copper is a reasonable start. 

The little pump I used was an immersible mains driven pump intended for water features, it drove the impeller by induction so I imagine it would be fine in antifreeze which will keep algae growth down and improve wetting. I am not sure how close to the ragged edge these PA devices are junction temperature wise. Often with power supply regulators or audio the answer is to parallel up more devices to get sensible junction temperatures but RF devices are so expensive that you don't want to derate them, or at least not by that much. My late father (G3MNO) used a Morris 1000 radiator and deionised water to shed heat from an SHF load a long time ago. 

Sent from my iPhone

On 6 Jun 2013, at 17:50, "Fuqua, Bill L" <wlfuqu00 at uky.edu> wrote:

>  Yes, that is the common way to do it, however, it does not address the problem with the need for very, very good reliable metal to metal contact between the 
> device and the block.  I have dealt with water cooled solid state device for the past 40 years, however, the power flow density of these new high power devices
> makes it necessary have very smooth mating surfaces, assure that they don't warp and so on. Direct contact water contact with the device and the 
> cooling liquid would be much better. It would require sufficient turbulent flow against the surface. One problem with air cooled engines is that they are limited
> by not only the heat transfer from the air but by the ability of the, in the best cases, aluminum to conduct the heat to sufficient fin area. So at best you are 
> limited to about 50 HP per cylinder as in a 917 compared to 200  HP or more  per cylinder as in a formula 1 (water cooled) car. A standard cast iron air cooled cylinder is limited
> to about 25 HP per cylinder as in a type 1 VW. Now you can get much more for a few seconds but continuous power you can't.
>  An extra note. The pistons in these engine run at much higher temperatures due to the fact that they are cooled mostly by the hot cylinder walls.
> 
> 73
> Bill wa4lav
> 
> ________________________________________
> From: David Lisney [g0fvt at hotmail.com]
> Sent: Thursday, June 06, 2013 11:13 AM
> To: Fuqua, Bill L
> Subject: Re: [Amps] Solid state amplifier cooling
> 
> With watercooling I would probably use a copper block with water ways drilled through, actually it may be a good thing to tap threads through even if you solder the pipes in, the threading would increase surface area in contact with the water and also increase turbulence... just a thought.
> 
> Sent from my iPhone
> 
> On 6 Jun 2013, at 16:09, "Fuqua, Bill L" <wlfuqu00 at uky.edu> wrote:
> 
>> I noticed just now someone suggested water cooling.
>> The advantage of doing it this way is that the electronic parts remain dry and
>> no problems with the dielectric properties of water or corrosion of the electronic
>> parts. You could use oil but it has less heat capacity than water, from about 1/4 to
>> 1/2. Also, makes a real mess if spilled.
>> 
>> 73
>> Bill wa4lav
>> 
>> ________________________________________
>> From: Amps [amps-bounces at contesting.com] on behalf of Fuqua, Bill L [wlfuqu00 at uky.edu]
>> Sent: Thursday, June 06, 2013 11:04 AM
>> To: David Lisney; amps at contesting.com
>> Subject: Re: [Amps] Solid state amplifier cooling
>> 
>>    You have given me an idea.
>> May not be original but here it is.
>> One problem with the  heat transfer from the newer high power devices is making
>> a good reliable contact with the heat sink. You could use liquid cooling.
>> The thought is to mount the device on a flat  hollow channel carrying water with a rectangular hole and
>> seal or gasket such that the water comes into direct contact with the devices heat
>> dissipator. With adequate liquid fluid flow the device dissipate its heat directly to the
>> fluid, perhaps water, much better than you would have with a metal to metal to air
>> heat transfer scheme. Warping of the metal is not such a problem either.
>> 73
>> Bill wa4lav
>> 
>> 
>> ________________________________________
>> From: Amps [amps-bounces at contesting.com] on behalf of David Lisney [g0fvt at hotmail.com]
>> Sent: Thursday, June 06, 2013 10:35 AM
>> To: amps at contesting.com
>> Subject: Re: [Amps] Solid state amplifier cooling
>> 
>> I used to be in the PC overclocking scene many years ago, I used peltier heatpumps and a homebrew water cooling system to keep the CPUs cool, and it did let them operate at extremely high clock speeds... However, the CPUs rated power interpolated to about 40w a piece and the peltier devices were using over 70w a piece. There are published curves of heatpumping capability versus temperature differential etc, mine was quite a high differential trying to run CPUs at close to freezing with a water cooling circuit at probably 30c.
>> 
>> I found the biggest challenge was dealing with condensation, ambient air needs to be excluded from areas that are operating below the dew point. The second issue is algae growth!
>> 
>> One thing I did learn and will use again is the enormous heat capacity of water, on doing the maths it was surprising just how small a pump is required and just how little flow is required.
>> 
>> At the KW level I am sure a vehicle radiator with a very low power fan would suffice, in the case of my computer experiment I only had a few hundred watts to shed and a vehicle heater matrix easily coped (blown by a pair of 120mm 12v fans operated at 7v).
>> 
>> I think I would still use a computer fan in the vicinity of the PA section, obviously it is not just the output transistors that generate heat.
>> 
>> Watercooling also enables the fans to be remote and keeps the noise away from the operator.
>> Mind you a transistor PA with no tuning is easily remote controlled....
>> 
>> Just some thoughts, I have not posted for years but I do read this thread every day.
>> 
>> 73 de David G0FVT
>> _______________________________________________
>> Amps mailing list
>> Amps at contesting.com
>> http://lists.contesting.com/mailman/listinfo/amps
>> _______________________________________________
>> Amps mailing list
>> Amps at contesting.com
>> http://lists.contesting.com/mailman/listinfo/amps


More information about the Amps mailing list