Amps
[Top] [All Lists]

Re: [Amps] QST, and class AB/F MOSFET amps

To: Manfred Mornhinweg <mmornhin@gmx.net>, amps@contesting.com
Subject: Re: [Amps] QST, and class AB/F MOSFET amps
From: mike kendall <ke6cvh@yahoo.com>
Date: Sun, 24 Dec 2006 21:36:16 -0800 (PST)
List-post: <mailto:amps@contesting.com>
Manfred,
    I have already worked out details for my liquid cooling scheme.  I found a 
good source for a cold plate and recieved it.  I had a guy help me in USA (I am 
overseas in Japan) with adaptors.  The reason why I need adaptors is because 
the water cooling pump that I ordered has different connections.  If you need 
details, I can provide some but it sounds as if your water cooling scheme is 
figured out.  I used a radiator that is close to what you would find on a car 
but is for industrial applications and was found on Ebay.  The fan is 12vdc but 
is so powerful that the stainless steel blades are flatted out on it when power 
is connected.  Possibly running the thing outside would be a solution or just 
to use a different electric fan that is cooler.  It think this fan in it's 
present configuration would cool a 5kw linear or more.
  73,
  Mike
  

Manfred Mornhinweg <mmornhin@gmx.net> wrote:
  Hi Marv,

> You might consider a simple envelope tracking scheme to reduce the
> dissipation in AB. Just keep the VDD high enough to output the
> applied signal and perhaps idle at 1/4 the normal VDD/dissipation.

That's a possibility, but the advantage in efficiency is not nearly as 
great as running an envelope-restored class E or even class C circuit.

> Rather than liquid cooling, have you seen the little heatpipe
> heatsink assemblies which are used for CPU's in the latest computers?

I have not seen them. So far I have seen heat pipes only in some high 
end audio amplifiers, and in satellites! But I cannot imagine a heat 
pipe cooling scheme to be convenient in this case. Either it would have 
to be connected to a really giant heat sink, or it would still need a 
noisy fan, which is exactly what I want to avoid! With water cooling 
instead, I can simply put the heat generated by the amplifier during an 
operating session into a bucket of water, and let it cool off naturally. 
And if I need to run the amplifer for an entire weekend non-stop, it's 
easy to connect two water hoses to my bucket, and change the water every 
few hours!

> I considered a scheme to take the audio from the input of a standard
> transceiver and use that to run the modulator for an amplifier in
> Class AB/D. The RF delay is long enough but, all the filters in a
> typical transceiver play havoc with the group delay. The DSP procesing
> necessary to "auto-equalize" an existing transceiver is in excess of
> that required to make a transmitter from scratch.

I can imagine! No, it's a fixed decision that I want to build a linear 
amplifier to be driven by a standard 100 Watt SSB transmitter. I don't 
want to do anything weirder than that. Saulo has demonstrated that the 
approach works good enough for practical use, without any compensation 
for the different group delay. Before his publication, I didn't dare try 
this, but now, I think there is evidence enough that it works well 
enough, even if it is clearly not theoretically perfect!

> Have you seen the Directed Energy (DEI) parts from IXYS? They are
> their best though pricey!

I checked their data now. I had not looked at them the first time, 
because they are not for linear use. For a pure class F amplifier they 
would be great, but for that I need a different drive source, not a 
standard SSB signal! I need to run class AB at low power, which means I 
need DC bias, which in turn means I need devices that can be made stable 
with that DC bias applied. It seems the DEI parts are not good for that 
kind of use. But maybe I should just try, considering that the AB 
operation will be at very low voltage only. I could add variable gate 
bias, reducing it at higher instant envelope levels, to prevent problems.

So many things to consider! :-)

At least I have found information about the water cooled heat sink 
design: The thermal transfer between copper and water is 350 + 2100 
SQR(speed), expressed in Watt per square meter and per Kelvin, with 
"speed" being the water flow speed in meters per second. This will allow 
me to design the heat exchanger on which I will mount the MOSFETs and 
also all the other components that generate lots of heat. It will be a 
hefty piece of copper! But much smaller for EER class AB/F than for a 
straight AB!

Manfred.

---------------------------
Visit my hobby website!
http://ludens.cl
----------------------------
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps


 __________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps

<Prev in Thread] Current Thread [Next in Thread>