On Thu, Jun 6, 2013 at 2:02 PM, Manfred Mornhinweg <email@example.com>wrote:
> it's great joy to see how much discussion your post has stirred up.
> ++++++++Agreed. Excellent thread.
> I'm thinking of water cooling.
> It's probably the only way to go. I suggest that you spend some time
> learning how to calculate heat flow through copper, through copper/water
> boundaries, and so on. Doing that, you will find out facts like these:
> - The only practical way to remove heat fast enough from those transistors
> is to have a very short, broad path through copper, reaching a sufficiently
> large copper surface in contact with fast flowing, highly turbulent water.
> That means a tiny copper heatsink with moderately thick fins, with a water
> jacked around, fed by a pump that delivers enough pressure to cause lots of
> - You may need to solder the transistors to the water-cooled copper
> blocks, to avoid the additional thermal resistance of a grease layer, but
> maybe you can get enough thermal conductivity in the rest of the system to
> manage without soldering the transistors down.
> - Simply immersing the transistors in water will not afford enough
> cooling, because their surface is too small. The water would boil in
> contact with the transistors, and the steam bubbles would keep most of the
> surface away from the water!
> - Do not feed the cooling from the tap. If you do, all it takes is pretty
> cold tap water and a moderately high humidity inside the room, to get
> condensation, and water dripping all over your circuit! Instead use a big
> enough water tank inside the room, that's never much colder than the air in
> the room. Exercise caution if the temperature in the room isn't reasonably
> Notice the unusually high power supply voltage. I'm thinking of running a
>> full wave bridge rectifier directly off the 240 VAC line, thus eliminating
>> the power transformer and giving about 340 VDC no-load. The amp would have
>> to be isolated from ground of course,
> That's what I proposed too, time ago, and what is used in industry in some
> A set of four 10 amp 600 PIV diodes is available on eBay for less than $30
> A ready-made bridge rectifier is usually cheaper than four separate diodes!
> > and a 6800 uF 450 VDC capacitor is available for about $50 including
>> shipping. There's most of your power supply. Pretty cheap, huh? :-)
> You may need to add at least an inductor, to keep the input power factor
> from becoming too terrible. A 3kW load with a power factor of 0.3 needs a
> circuit capable of supplying 10kVA, and I dout you have that! Maybe you
> need to compromise, use a smaller capacitor and accept some more ripple, to
> get a manageable power factor. It depends on how bad the power factor can
> be, before it gives trouble with your electrical installation!
> My main use would be RTTY and CW so linearity is not an issue, but do you
>> have any thoughts on what the IMD might be for SSB?
> I have no experience with those transistors, but I have heard rumors about
> the high voltage transistors not being good enough for linear use. Some
> rumours said it was because of poor linearity, others said it was because
> of hotspotting and device failure. I frankly don't know what's true about
> them, if anything.
> Maybe you can linearize them enough by proper use of negative feedback.
> Actually ANY solid state amplifier should use negative feedback! MOSFETs
> typically have far greater power gain than a ham needs. Instead of
> attenuating the drive input and then keeping the full gain, it's MUCH
> better to reduce the gain by using negative feedback. That way the excess
> gain is put to good use by enhancing linearity.
> Since the output impedance is about 50 ohms, I'm wondering if a 1:1
>> broadband antenna balun might serve for the output toroid? Just a thought.
>> Might not be a good idea.
> Antenna baluns usually don't have galvanic insulation between input and
> output, so you can't use them in an amplifier powered directly off-line.
> But using the core of such a balun, and rewinding it properly, should be
> Like I said, I'm a complete newbie at solid state amps so all comments are
>> welcome. This almost seems to be too easy to be true. Go ahead, burst my
>> bubble. :-)
> OK. You asked for it! ;-)
> I would suggest that you first get some very cheap, small RF MOSFETs,
> could even be switching type MOSFETs that cost one buck each, and go ahead
> building a push pull amplifier with them, that delivers maybe a few tens of
> watts. As driver you can use any 100W HF radio with power control and an
> additional attenuator on the output, so you get down to roughly 1 watt of
> drive. On that small, cheap amplifier you can grind your teeth, learn how
> they work, what problems they give, you can learn how to measure IMD, and
> so on. If you burn up something, the damage will be limited to a few bucks,
> at most.
> Once you feel fully confident that you thoroughly understand how it works,
> make a version #2, that basically milks those cheap little MOSFETs for the
> absolute maximum power you can get from them. This will allow you to learn
> about the issues involved with cooling transistors that work hard.
> Once you got to the point where you can take two MOSFETs rated at, say 40
> watts absolute maximum dissipation each (think of the IRF510, for example),
> and make an amp with them that delivers 40 to 50 watts output, with decent
> IMD, efficiency, and that survives continued transmission at full power,
> you are ready to buy a pair of ARF1500's and build a 1500W amplifier with
> them. At this point you will no longer be a newby, but you will know all
> the basics, so that you will be ready to tackle the additional problems
> that come with high power.
> Of course you might want to skip all the previous stages, and jump
> directly into building the high power amp. But that will make you suffer
> all your unavoidable beginner's mishaps on the big amp, each time taking
> out maybe 500 dollars in parts! And it will have you deal with basic
> problems and with high power related problems at the same time.
> It can be frustrating to see expensive parts go up in smoke, and not even
> knowing why they did. Instead when it happens with cheap parts, you can
> smile about it and just try again, to try finding out why they blow up. It
> builds a lot of confidence to be able to make a good 50 watt amp out of two
> one-buck transistors, and that's the kind of confidence you need to build a
> big amp. And everything you learn with the small amp can be applied to the
> big one.
> Visit my hobby homepage!
> Amps mailing list
Amps mailing list