Hi all!
Thanks for all the answers. I will reply one by one:
Bill:
> I do mean Full-Wave Voltage Doubler. There are two kinds of Voltage
> doublers. One where one side of the transformer is grounded
> (half-wave) and the other where one side of the transformer goes
> between two series capacitors. That is the full-wave.
Now I see what you mean. Sorry for the mixup! Yes, indeed I have not
seen any amplifier using such a half have voltage doubler. It would make
pretty little sense!
Angel:
> If this project is the same quality and benefit to the home-builder
> as your Fodtrack (and I am sure it will be), please go ahead!
Well, I never know how popular a new project will be, when I start it
out! FodTrack was a completely unexpected success. I developed it for my
very own use, since I didn't like the high cost of available solutions,
and then other hams followed. My 40 Ampere switched power supply is
another such successful project. Instead, some other projects have found
no takers. As far as I know, nobody has ever duplicated my automatically
tuned HF mobile antenna! Its combination of electronics and mechanics
seems to be too demanding for almost anyone.
In any case, these days I'm trying to do good, well designed projects,
not just quick and dirty minimal ones, so there is some chance that this
amplifier might turn out good! But I don't know yet what exactly I will
build, and much less any time schedule.
Mike:
> Why not consider using some devices from IXYSRF?
Frankly, because I had no idea about their existence! That's why I
joined this forum... I need hints like yours!
I checked the IxysRF web site. First, I was disappointed by noticing
that their biggest single RF MOSFET is for 300 Watts output, and with
significant restrictions on the power dissipation, but the other data is
good. Then I noticed the price. Wow! At that price, I'm willing to
combine several modules! The hybrid combiners and other additional parts
will be paid for by the savings in MOSFETs, and the risk is lower
because those other parts usually don't burn out...
It looks now like I will go the IxysRF route.
> There are already people building linears using the ARF devices.
I know of the Japanese hams who published their design in QEX, but no
one else. Any more hints?
> I have been working on building a MOSFET linear. I have already
> planned and purchased the devices to use liquid cooling.
I also plan on liquid cooling. By the way, do you have simple and easy
equations to calculate how much contact area between water and copper is
needed for a given thermal resistance? I know very well how to
calculate thermal resistance in the copper block, how to calculate the
temperature rise in the water, but not this metal-to-water transfer,
which I suppose depends on surface turbulence and such stuff.
> I have already purchased the power supplies.
I will make mine. In that field I have enough experience...
> I would like to use the devices from IXYSRF but am very busy as I am
> at sea most of the time as I am in the US Navy.
I'm also very busy now, working at 2600 meters altitude in the middle of
the world's driest desert. Definitely your place is wetter, but we are
probably just as busy! Anyway, I hope to retire very early, quite soon.
Then I should have time for doing the really important things, like
developing my pet projects!
> I would like to build the linear using the devices from IXYSRF as I
> found the ARF devices rediculously hard to obtain with extremely long
> lead times.
Indeed I had been looking for a source of ARF devices for over a year,
until I came across Richardson, which have at least some of them. The
others still seem to be unobtainium.
IxysRF has web site ordering, but when I tried it, the page didn't work
correctly. Will try later.
Peter:
> It seems to me that one parameter which could cause major problems
> would be AM to PM conversion in the MOSFETs as you varied the
> voltage.
I would have to look into that as the project advances. I really don't
know how serious this would be.
> You could get around that by using either Cartesian or Polar
> Loop feedback techniques,
Sounds a bit like Chinese to me... I will have to google up those terms! :-)
> although you would have to take into
> account the poles in the loop response determined by the modulated
> power supply characteristics.
There will be basically just one pole of any importance, around 8kHz.
> I don't know if the MOSFETs are
> characterised for all 4 S parameters over a wide supply voltage
> range:
The ARF ones apparently are not. The IxysRF ones are.
that's the info you'd need to determine (by simulation or
> otherwise) the amount of AM to PM. The rest of the loop could be
> simulated, although simulating the PSU response might take a long
> time unless (or maybe even if!) you have a Sun or similar workstation
> - depending what software you have available.
I'm mostly a practical guy, so my approach is to build the basic power
supply, making the response as fast as practical, then measure its open
loop response, then calculate compensation and other things. It's safer
than simulating the whole thing, and could actually be quicker too!
About simulation software, the only I have is a VERY old release of
MicroCap! It runs in DOS... so you can imagine! :-) But that thing
actually can simulate the startup of oscillators. Another much more
modern software I used at work some time was NOT able to do that!
Marv:
> There is a small problem, often overlooked, in the envelope
> restored Class D/E/F "linear" scheme. The group delay through the
> modulating path must be equal to the RF path delay for the system to
> exhibit low distortion. This can be realized over a narrowband but,
> when a wideband system is considered it may prove to be easier to
> build an entire envelope restored transmitter than to construct a
> standalone linear.
Well, my self-inflicted task now is to make a noiseless legal limit
linear amplifier driven from existing 100 Watt radios, so building a
whole transmitter is not something I'm considering now. If the AB/F
scheme turns out to be excessively troublesome, I will probably revert
to a classic class AB design and beef up the thermal design. Less
elegant, but almost guaranteed to work well. And just in case if MOSFETs
turn out troublesome, I have two trusty 4CX1500B tubes lying around! But
they are not suitable for silent (blower-less) use.
In any case, in his QEX article about his AB/F amplifier, Saulo included
screen shots of the dual tone waveform and the spectral analysis. There
is indeed a plainly visible distortion caused by the delay in the power
supply response, and this reflects in the spectrogram as slightly raised
IMD. But as Saulo points out, the IMD is still very acceptable, actually
it's better than some commercial AB amplifiers! So it seems that this
AB/F approach without controlled delay is "good enough", even if it is
clearly not perfect.
My intention is to make the AB to F transition at the lowest possible
power level. Saulo made it at about 25% of the amplitude. I intend to go
lower, as low as the driving of the MOSFETs will allow. That should
reduce distortion. Also I was toying with the idea of adding a delay
line to the RF drive, but that might be simply not worth the effort.
When you see how many hams are using their amps without any ALC, happily
flattopping away with the IMD less than 12 dB below the output, it
should become clear that an imperfect AB/F amplifier with the IMD at
-28dB should be quite acceptable!
But then, the "recently discovered" IxysRF devices are rated for stable
linear operation at 150 Volt, so one of my reasons to use class F and
envelope restoration is weakening... Maybe I go the brute force route of
pure class AB! It depends on the outcome of the experiments. After all,
when anyway using a SMPS, the physical difference between a class AB and
an amplitude-restored AB/F is just in the small components and in the
output network! It's easy to change along the road.
I have gotten lots of info to think about. Thanks to all of you, and
keep it coming!
Manfred, XQ2FOD.
----------------------------
Visit my hobby website!
http://ludens.cl
----------------------------
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|