[Amps] Liquid cooling

[TexasRF at aol.com]

Manfred, are there computer repair shops there? Many of the thermal grease  
products are targeted for CPU cooling applications.
 
73,
Gerald K5GW
 
 
 
 
In a message dated 8/24/2011 2:19:19 P.M. Central Daylight Time,  
manfred at ludens.cl writes:

Roger,

> Agreed.  I use Artic Silver when I can not  solder.

That's good. I would like to buy some Arctic Silver, but need  to find a
store that will sell me a small amount, and export it! Here in  Chile, of
course, that stuff is unobtainable from the local distributors. I  can
easily get the usual white ceramic thermal compound, packaged in  little
jars with no label, so I don't know what's really inside! And  with
ceramic thermal compound ranging from 0.7 to 3.8 W/mK in  thermal
conductivity, I would like to know what I have... So I guess that I  will
include a small jar of labeled, specified thermal compound in my  next
order from Digikey, but that's still only 3.something  thermal
conductivity, while Arctic Silver is rated at 8.something. Digikey  has a
silver loaded thermal compound too, but with no specs. Any advice,  anyone?

> The problem with solder is the temperature compared to the  critical
> temp for the device

I used to believe the same, but  since starting to build SMD equipment by 
reflow soldering in the kitchen  oven, and never killing a component, my 
beliefs in this regard have  changed! Nowadays I believe that it's pretty 
hard to fry a component by  soldering.

Normal 60/40 solder melts at 189 degrees Celsius. Silicon  devices can 
operate continuously at 150 degrees junction temperature, and  some 
manufacturers claim 175, 200 or even 225 degrees for some devices.  And 
what they can survive for a few minutes is even higher. I have  
oven-soldered many devices so far, both to circuit boards and to heat  
sinks. I heat the oven to 200 degrees, wait for the solder to melt, wait  
a little more to make sure it really has molten well at every place, and  
then switch off the oven. That has worked great so far.

> and  the need to make sure the entire area between
> the device and heat sink  is wetted.

As long as everything is clean, that's no problem. And even  if you get 
some air bubbles trapped in, it's still better than 100%  filling with 
thermal grease!

> OTOH there
> are "solder  pastes" available that can be applied to the mating
>  surfaces

That's what I use. I bought a 10 gram syringe of 63/37  no-clean solder 
paste, and it still lasts. A little of that stuff goes a  long way, 
because the layer formed between the part and the PCB or heat  sink is 
very thin. When you apply the proper amount, almost none of the  solder 
flows out from under the part.

> and then melted by a  brief induction pulse

I don't know about that technology. I would be  scared to fry sensitive 
parts by inducing such a high current in teh metal  as to melt the 
solder! I prefer to use the kitchen oven. That takes about  5 minutes for 
a PCB, and half an hour to solder a large transistor to a  thick copper 
plate, but is electrically safe!

> A bucket under  the desk would be a disaster for me the fist time I
> put my feet under  the desk. <:-))

Either bolt down the bucket and put a lid on it, or  tie down your feet! ;-)

> What you need are two 1CM square plates,  separated by 1 CM which
> should give a close approximation of the  resistivity in ohms per
> CM^3.  

I just did that test, and  got 280 kiloohms. But it seems that the 
resistance from the electrodes  into the water is more important than the 
bulk resistivity of the water,  because changing the distance between the 
electrodes only produced a small  difference in the resistance, while 
inmersing more or less area of them  changed the resistance in inverse 
proportion.

In any case it looks  like the resistance from side to side of my 
amplifier will be on the order  of tens of kiloohms, at least, and that 
should be fine. So I can use  spring water!

> I think the limit we set was on the order of 3000  ohms/CM^3.

Well, what I got is nearly 100 times better!

>  Of
> course that was with 6 to 8KV, but we used about 12 to 15 feet  of
> hose for isolation. 

I will only have 140V. In principle I  would have only about 20cm of hose 
between the sides, but if necessary I  can use a longer hose and coil it 
up, to increase the  resistance.

> IF you only have water running through, or in the  device when it is
> in operation (IOW it drains back into the bucket  when not running)
> eating metal ions will take a very long time to  become a problem as
> in many years..

That's a good  info!

> Here's a thought...put the coolant exhaust 6" to a foot  above the
> device or higher (what ever is handy) with a quart container  (or what
> ever is handy) in the line. Take the return from the top of  the
> container. When power fails to the pump, you will have a quart  (or
> what ever size container you use) for a safety reserve that  will
> drain back through the device, giving plenty of time to shut down  
> before lack of cooling becomes a problem.

I have a much better  plan: I intend to build a flow sensor into the 
system. No flow, no power  to the amp. Digikey has a flow sensor for 18 
dollars that seems to fill  the bill for the flow level I expect to use. 
The output is simply a reed  switch. I don't like the very wide 
hysteresis that sensor has, and maybe I  can find (or build) something 
better. But even this simple flow sensor  should provide pretty good 
protection. Digikey also has better flow  sensors, but at several hundred 
dollars apiece, they are out of all  consideration.

I plan to switch off the pump during prolonged RX, but  keep the circuit 
primed. That doesn't play well with the safety reservoir  system.

> I'd really like to build a SS 160 through 6 meter amp  capable of
> 1.5KW CCS with enough overhead to comfortably run that  legal limit.

What for? I can see the need to run 1.5kW key down for a  few minutes, 
such as for RTTY and SSTV. But CCS? Hardly. At least I don't  see why a 
ham could ever need 1.5kW CCS! Maybe you can give me some  example!

> Many commercial tube amps advertised as a legal limit amp  just don't
> have the power supply or components for sustained operation  at the
> legal limit in SSB or CW let alone digital.

Yes. They  are aimed at hams who WANT CCS specs, but don't NEED them, and 
will never  actually use the amp that way! ;-)

> Some of them even push 
>  the tubes way beyond their ratings to get the power they do and
> that's  at a greatly reduced tube life.

That's usual practice in ICAS, and has  a long tradition.

> IOW I'd want an amp capable
> of at least  2KW continuous and preferably 2.5KW that should loaf
> along at 1500  watts without breaking into a sweat.

Not a good idea! A 2.5kW amp  running at 1.5kW will necessarily be less 
efficient, as it will be  operating at part load. If you really want to 
broadcast in RTTY around the  clock at 1.5kW, then you should build a 
1.5kW amp with generous thermal  capability, and not a 2.5kW amp! IOW, 
the electrical design should be  optimized for 1.5kW, but using 
components, cooling, etc, designed to run  significantly below the 
temperature limit of each part, at that  power.

> The current project is a 4CX3000A7 that I want to run class  A. 

Don't allow promoters of energy efficiency to hear that. In this  world 
where energy is getting ever more expensive, and energy wasting  devices 
draw the scorn of many people, we should be looking at class D,  class E, 
and class F, with envelope elimination and restoration. Not  really at 
class C, much less class AB, let alone class A!

>  The
> problem with the higher power tubes is their size. Plus they use  more
> filament power than the average ham station uses in total  power.

That filament power, even for a class AB amp, is one of the  reasons why 
I want to replace my tube amp. I run my house on my own  microhydro 
turbine, and in winter I have all the power I want, but in  summer it can 
get really scarce. I use battery backup in summer, and  drawing 2kW 
pulsed for high power SSB transmission is OK, for short  transmissions, 
but running filaments and bleeder resistors and a blower  even while in 
RX is unacceptable.

> (Bout ~375 watts) One amp on  HF and one for 6-meters. It' d be nice
> to have one for 2-mters as  well.  I have the tubes and sockets but
> not enough components to  build complete amps.

I got two 4CX1500B tubes some time ago, for free.  Broadcast pulls. The 
guy said they were still in perfect shape, but I have  no means to test 
their emission. Maybe some day I make a cavity amp for  two meters 
moonbounce with one of them. But more likely not...
I have  no sockets for them, and would probably have to make them. But 
really a  MOSFET amp for HF is a more attractive project to  me!

Manfred.

========================
Visit my hobby  homepage!
http://ludens.cl
========================

_______________________________________________
Amps  mailing  list
Amps at contesting.com
http://lists.contesting.com/mailman/listinfo/amps