Hi Mike!
> Manfred, The IXYSRF website has changed. They did have their module
> for 300 watts listed several months ago specifically for HF 2-30MHZ.
> The schematic they have that is downloadable is certainly workable
> for HF.
Now they have no schematic, or at least I couldn't find it.
> Possibly, we could collaborate a little but like I said no time for
> me for now.
I hope I can free some time for this project, but for now I'm only
trying to pull together the facts, to analyze whether it makes sense
building this!
> I was told that some of the high power devices have such a small
footprint for their cooling that air cooling is not suitable.
Indeed the very small footprint is a very big problem! But this is not
solved with water cooling per se. What we need is to spread out the heat
to transfer it to the flowing medium, be that water or air. For this, we
need the very best material (pure copper, because silver is too
expensive and only marginally better), and we need to design the cool
plate so that the little contact area of the transistor is connected to
the dissipating surface by an in-copper path as short and wide as
possible. Water cooling makes things easier, because less transfer area
is needed for water than for air, and this allows a shorter path from
the transistor to the surfaces. But the biggest thermal resistance is
still in the volume right below the transistor, and there is no way to
improve this! So, the dissipation possible with an air-cooled heat sink
is only marginally worse than that achievable with a water-cooled one.
The bigger difference is made by the material. It MUST be copper! At
least for a heat spreader.
> For the 100 watts per device, air cooling should be OK.
That depends on the power you want to dissipate, and the area. With 100
Watt dissipated power per device, assuming class AB with 45% efficiency
and 1500 Watt output, you would need 16 transistors and still be a tad
short! Not very practical.
> The cold plate is made by Lytron and they can be googled.
Thanks for that info! Not surprisingly, they can be found at lytron.com.
I checked their data. The best performing cold plate they have is the
CP25. It is rated for a thermal resistance of 0.33 Kelvin per Watt per
square centimeter. The IXZ2210N50L is rated for 470 Watts dissipation if
you hold the heat sink at 25 degrees Celsius (impossible) and use the
best imaginable mounting technique (hard). It has a mounting surface
area of 3.5 square centimeters. Since the mounting plate is very thin,
the really active thermal area must be much smaller, just the die size
plus a little around it. I don't have the number for this, but I will
assume that half of the mounting area is thermally active. That would
make 1.75 cm2. So, the CP25 cold plate would present a thermal
resistance of 0.19 Kelvin per Watt. As a result, the mentioned
transistor could dissipate only 245 Watts on this cold plate, at 150
degrees Celsius junction temperature, which is the highest I would
design for. That's assuming a water supply at 25 degrees Celsius, and
excellent mounting! In practice we would land even a bit lower, probably
around 200 Watts. That's VERY poor cooling performance!
I would say that due to the point-source nature of these transistors, a
cold plate from Lytron would have to be combined with a rather thick
copper heat spreader. Such a sandwich would allow much better thermal
performance. And in this case there are several suitable cold plates
from this company. But I would prefer a cold plate designed from the
start for point sources. Such a plate would have a thick copper
baseplate, and then the internal fin structure. It would save the
(relatively minor) thermal resistance between spreader and plate, but
more importantly it would avoid the need of having to obtain a very flat
spreader that matches the very flat cold plate! Maintaining thermal
contact over a large area is much more difficult than over the small
area of a transistor.
> The cold plate I ordered/recieved is probably twice the size I need
> to get the job done so I have room for all the modules, no problem.
That's great! Just put a copper spreader on top of it! And then you can
use it too for the power supply transistors and diodes, attenuator
resistors and whatever else will produce lots of heat.
> I want to be able to use this linear 100 percent duty cycle in a room
> without air conditioning if necessary.
I would also like to design for this goal, just for peace of mind!
> The IXZ2210N50L is suitable for HF linear service and is the device
> that I would like to use due to the low cost and availability.
Yes, indeed it's the best I have seen, for linear service, together with
the single version of it. But if I go ahead with my plan of using
nonlinear amplification and envelope restoration, I would prefer to use
bigger devices, ideally just one pair of them, for simplicity. The much
lower dissipation would allow that, and for switching service the bigger
MOSFETs are available! I have to decide, and still need some more
information before I do, specially on the issue of phase modulation in
such an amplifier.
> I would like to collaborate with you on this project and have
> purchased most items already except the actual RF devices.
I have nothing yet. Just planning!
> I also have spent time on the bandpass filters as well.
Bandpass? Wouldn't lowpass filters be more convenient? I think for class
AB this is a pretty straightforward task. With class E or class F it is
much more complicate.
> Due to the high power expected at 1KW plus, I have purchased enough
> Russian vacuum relays to cover all the filters.
If they are cheap enough, great. But I wouldn't worry much about these
relays. 220VAC, 6A relays are widely available and inexpensive. Any of
these that exhibits sufficiently low inductance should do the job. And
if not, 400VAC, 10 or 15A relays are still inexpensive. I don't think a
vacuum relay is an advantage here, since these relays will never be
hot-switched.
> I suspect the IXZ2210N50L is suitable for 6 meters
Yes, sure it is. Only the gain falls and the impedance gets even harder
to match.
> as well and was considering using the 54mhz low pass filter out of
> the QST article that uses copper plate inside the project box.
Sounds reasonable.
> I have purchased/received some TELCOM switchers that are pretty high
> quality and plan to use 100v power supply.
Yes, 100V seems reasonable to run at 300 Watt output from one
IXZ2210N50L in push pull.
> Their website has changed since I started to get all the parts but
> origionally had a schematic for an HF module with the IXZ210N50L. It
> is no longer on the website and I have it saved to a different
> computer that has some problems on it right now. When I remove the
> hard drive and slave it I should be able to pull the info off it and
> send it.
I would love to see that information!
> I have less than a month here and will be gone again for another 6
> months at sea again
I will also be at sea a month from now, aboard the sailboat Capricornio!
The first two trips on it are described on my web site (nautical
section). If we don't sink, I will be back in february, and if we do, I
will use ham gear on board to call mayday! :-)
Dear Hsu,
> Manfred, I'm very interesting in it. Maybe you can make some small
> power amplifier model and combine them together as a big power
> amplifier.
My ideal situation is having just one RF power amplifier. If that proves
impossible, I'm willing to accept up to four. So, each would have to be
rated at close to 400 Watts output. Instead for experimentation I would
prefer something much smaller!
Mainly, I would like to experiment with the phase modulation introduced
by an EER amplifier. It might be faster than waiting for my math guru
friend to simulate it! Depending on the results, I might decide to go
for the big guy, and make an amplifier with a fast reacting 30-125V
switching power supply and a single high power RF stage using RF
switching class transistors, running class AB/F if I find ways to do it
and AB/C if not. If the phase modulation problem seems big, I might
instead opt for the brute force mode, using pure class AB, with a fixed
switching power supply, and in that case I would probably use 4 modules
with two IXZ210N50L each. But that would fall a tad short of 100% duty
cycle rating. It would be good only for SSB at 1.5kW, with more
demanding modes requiring a power reduction. I don't like that so much!
Manfred.
----------------------------
Visit my hobby website!
http://ludens.cl
----------------------------
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|