At 01:44 PM 3/4/2005, Dave Bowker wrote:
>The performance of any phased array is critically dependent upon proper
>[phasing] and [amplitude] of the power delivered to each element in the array.
Sure
> If we consider only the transmitting case, this is almost impossible to
> accomplish with what your are proposing.
Nope
>The phase from the central feed point to the input of each element
>amplifier can readily be controlled, but how do you plan to control the
>phase shift through each amplifier at the element?
One calibrates the system, at full power (because phase shift through the
amplifier varies with power). In practice, one calibrates the components
infrequently to get close (i.e. you build up two port models (over
frequency) for the amplifier, and for the tuning network). You also measure
the mutual impedances between elements of the antenna. Then, on the fly,
you calculate a first estimate for how to configure it (using standard
network analysis), apply power, and tune for maximum smoke with an optimizer.
Really, it's not much different from how antenna autotuners work. Either
they start from scratch and go through an iterative process to get a match,
or, they look at the current frequency, and use a previously memorized set
of settings.
>No two power amps exhibit the same phase shift when matched to the load
>impedance ... PERIOD
Which is why one calibrates.
>... there are literally an infinite number of L/C combinations which will
>match the amp to the load, but each one exhibits a different phase shift.
Not quite an infinite number of L/C combinations.. turns out there's
actually relatively few, particularly if your tuner adjusts L and C in
discrete steps. The real complication is that your load impedance (the
element driving point impedance) is a function of all the other element
source impedances.
As far as the phase shift of the adjustable networks go, they are both
measureable, calibrateable, and calculatable. They're relatively stable (in
the minute to minute sense) passive devices. One can fairly quickly step
through a sequence of Ls and Cs and remember what the two port parameters
of the device are. Then, when in use, you can get fairly close by
calculation, then bump up and down to optimize.
Considering 4 elements, you'd have, typically, 12 variables to play with:
the phase of the signal going to each amplifier, the L and the C of the
matching network. You can calculate the initial guess, then, in the space
of a few seconds, try bumping each one up and down one step and see which
one produces the desired element currents to match the calculations.
> Amplitude control is possible (at great expense of complicated circuitry),
Amplitude control is actually fairly trivial, particularly at low
powers. A diode attenuator will work. An even easier way is just to
generate appropriately phased and amplitude signals using something like a
DDS (or, for example, a SDR-1000, which uses a DDS and I/Q
upconverters). On the low power side, you don't care much about
mismatches; it just changes the phase (maybe) and the power, which you want
to do anyway. The key is the ability to calibrate what you get. It doesn't
have to be particularly linear (in the sense of constant dB per volt, for
instance)
> but each element of a phased array must have precise [phase ]AND
> [amplitude] of power applied to it.
>
>And how do you propose to handle the reverse situation (on receive) ...
>the amps you are contemplating ARE NOT BI-DIRECTIONAL !!!
One doesn't really want to use the same technique on receive anyway. The
optimization strategy is actually quite different on transmit and receive.
On transmit, the objective is to squirt the maximum amount of power in the
desired direction, but if there's a sidelobe, it doesn't matter. On
receive, the objective is to have the minimum response in directions you
don't want. What you really want is adjustable nulls.
It also turns out that on receive, there's a significant advantage in
having more elements, preferably with different polarization
characteristics, so that you can use adaptive beamforming. On transmit,
because you don't know what the path looks like to the "other guy", there's
no advantage in trying to do polarization diversity.
>I don't think what you are contemplating is realistic from an engineering
>point of view, let alone economics!
It's extremely realistic from an engineering point of view (as in, it's
been done before).
As far as economics goes... sure, it's not as cheap as a rockmite and a
wire, but then, it's not all that bad.
Here you go with a typical equipment stackup:
4 SDR-1000 digital radios (w/100 Watt output) - 4@ $1300
4 SGC500 500W amplifiers - 4@$1200
4 high power computer controlled tuners 4 @ $500
4 radiating elements (one nifty approach would be a pair of switch
selectable horizontal dipoles at the top of each of 4 middling high posts)
PC with necessary multichannel sound card interfaces $1000
Probably $15K by the time you get it all put together. Compare that to a
big tower and a SteppIR or two and a 2kW linear. It's in the same general
ballpark.
10-15 years down the road, the potential is for small modules, costing
perhaps $500-600 each incorporating the digital stuff and a suitable 100W
or so PA, integrated with some reasonably compact radiating element. Put
10 or 15 of those modules on your roof, along a fenceline, on a pair or
three towers, etc.
>73, Dave, K1FK
>Fort Kent, ME
_______________________________________________
See: http://www.mscomputer.com for "Self Supporting Towers", "Wireless Weather
Stations", and lot's more. Call Toll Free, 1-800-333-9041 with any questions
and ask for Sherman, W2FLA.
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|