[TowerTalk] amplifiers for phased arrays

[Jim Lux]

At 07:49 AM 3/4/2005, Tom Rauch wrote:
> > Feedback and real time adjustment is a given...  Even in
>an all passive
> > array, one needs this.
> >
> > I'm just looking for "gain blocks" that are available
>essentially off the
> > shelf and won't require extensive
>redesign/modification/extra parts. Sure..
> > one could gut a MFJ and move the filters outboard with
>relays, for instance,
> > or, even, hook a stepper motor up to the band select knob.
>There's more
> > than one tetrode based amp that basically does just this,
>with motorized
> > tank components.
>
>You'd better look close at IMD. Solid state amps are
>generally ratty. The ALS500 is suitable only for use in
>mobile installations where the signal level is modest. Most
>of the other 12V amps are worse yet!

Of course, meaningful IMD specs are somewhat hard to come by (at least in 
the mfr literature).


>The ALS600, having real 50V FET's designed for linear
>applications is OK, but not exceptionally clean. It compares
>very favorably against tetrode amateur amps (better than
>many). Of course tetrodes are not nearly as clean as GG
>triode amps. Then you have the problem of mutual coupling
>between elements back-feeding each amplifier. How will you
>handle that?

It turns out that one can (usually) find a phasing configuration that 
optimizes power squirted in a particular direction, while constraining all 
amplifiers to operate with maximum output power. It's a different problem 
than the classical power divider phased array (where you can have 
circulating currents between elements) or in a simple pattern synthesis 
scheme relying on choosing element currents.

On transmit, you're not so concerned about odd side or back lobes that 
might be formed, providing there's not much power in them (a sidelobe that 
is 10 dB down, and very narrow, doesn't have much power in it).

I'm doing some more modeling now to find out if there are certain 
(hopefully rare pathological) configurations for which a solution cannot be 
found.


>I can't imagine why anyone would want to use amplifiers to
>drive each element in an HF amateur array. It is almost
>certainly an inefficient wasteful system,

Why almost certainly? Complex, yes, but not necessarily inefficient. Using 
transformer power combiners to combine the outputs of 300W modules to make 
a 1kW amplifier is also inefficient. Not only is there the few tenths of a 
dB in the combiner itself, there's also the losses from mismatches in the 
modules into the combiner. Sure, one can get this pretty darn good, but 
still, it requires design effort, some amount of complexity, perhaps 
matching of devices, etc. It's extremely hard to do it in a broadband 
device covering a decade.

Independent amplifiers per element at least gives you the potential for 
zero loss spatial combining. Amateur radio is also somewhat unique in 
having a wide band of frequencies over which one might operate, but a small 
instantaneous bandwidth. Since you already have to control phase and 
amplitude in a closed loop way, you can compensate for amplifier variations 
on the fly, and the narrow instantaneous bandwidth means that you don't 
need a lot of degrees of freedom in the compensation process.

Besides... amateur radio doesn't care about efficiency, whether in terms of 
electrical, radiation, or dollars.

The limit is on the power fed to the antenna, not on the DC or AC power fed 
to the amplifier, so hideously poor efficiency isn't really a problem. 
Perhaps in practical terms one can't tolerate too poor efficiency, where 
there's a limit to how big a branch circuit you have in the shack. Mind 
you, I think that limiting "wall power" to the transmitter for the amateur 
service would encourage a lot of creative designs.

As far as radiating efficiency goes, it's no better or worse than any other 
scheme, since it's really the elements and their immediate environment that 
determines radiation efficiency.

In terms of dollar efficiency (bang for the buck).. I won't go there... 
it's amateur radio, after all.


>  and for all the
>aggravation it wouldn't work on receive (where most of the
>advantage of having directivity occurs) without even
>substantially more hardware.

Indeed... and, in fact, the optimum phasing/combining for receive is 
usually different than for transmit.  But on receive, you're working with 
low level signals, and fairly straightforward digital signal processing 
techniques work there, quite nicely.

There's no particular requirement that one has to use the same "box" for 
receive and transmit, or, for that matter, the same antenna elements.. what 
makes a good transmit element might not make a good receive element.



>My curiosity is piqued. Why would anyone want to do
>something like this?

Separate T/R modules for each radiating element is a pretty standard way to 
build a phased array.

Doing most of the processing and array calibration in the digital domain 
means that the complexity is in software.  Likewise, moving complexity from 
the analog hardware into a digital controller makes the hardware simpler 
(fewer overall parts) so it's cheaper and more reliable, and the digital 
controller allows more sophisticated and higher performance systems to be 
built (because you can implement a higher order or more complex control law).

Compare to modern internal combustion engines.  There's not a whole lot of 
analog fuel injection controllers being made any more (if any), and even 
fewer carburetors, and, in general, engines run better, over a wider range 
of loads and environments, and are more efficient, cost less to produce and 
maintain, etc.


In any case, as antenna restrictions proliferate, and as the price of 
computing goes down, phased arrays will become more and more useful as a 
solution.   Someone with a 100x50 foot lot potentially has an enormous 
aperture to work with, if you can deal with the non-uniform dielectric 
environment, the RF noise, and the overall complexity.  Someone putting 
physically small radiators on their roof still will face the "height above 
ground" problem: there's no getting around that (short of changing the 
rules to regulate far field field strength instead of PEP power to the 
antenna).


As for why "I" am doing this.. It's something I've been pursuing for the 
last 10 years, and is tangentially related to my professional work (the 
following links might work)
http://rtd.jpl.nasa.gov/reports/rtd04/report.html
http://rtd.jpl.nasa.gov/reports/rtd04/pdfs/R04026011.pdf  <<my work





>73 Tom