The totally cool way to do phased arrays...
Placeholder page until I find the original one...
With an active phased array, all of the phasing is done at low levels, and there is a separate amplifier at each radiating element (PA for Tx and LNA for Rx) The phasing can be done at IF or baseband, with the final frequency translation at the element.
With the division of functions at the element, there is really no reason why you need to have receive and transmit elements be the same, other than for mechanical convenience.
Here's a block diagram of the transmit side.
In this sort of array, one would want to design the array phasing such that the full output power of each power amplifier can be used, and a simple equal power divider used to distribute the power through phase shifters. The matching network transforms the apparent feedpoint impedance to the 50 ohms the amplifier wants to see. With mutual coupling between elements, the apparent feedpoint impedance is going to depend on the drives to the other elements, making figuring out the phases of the drive a bit tricky.
Power amplifiers - You want a broadband amplifier module that puts out, say, 100W.
The HL-50B looks interesting.. It's a $250 (approx) 50-60 W linear designed to hook up to things like a FT-817.
What sort of load is that amplifier working into?
In a phased array, the terminal impedance of the element is usually nowhere near 50 ohms resistive, nor is the power distribution between elements necessarily equal (in general, it's nowhere near equal). What's the impact of this on, say, a 4 square type array where the front element has negative power? Perhaps one just wants to make power>0 a constraint? Peak power capability costs money, so in the generalized sense, you're always going to buy more capacity than you need for any given phasing.
What if you make the phasing constraint that you put constant power into each element. An (automatic) antenna tuner would adjust the impedance being seen so that the amplifier is working into what it likes (i.e. 50 resistive) by minimizing reflected power. You'd adjust the phase of what gets fed to the amplifier (at low level). This is a bit (!) tricky to calculate (and probably trickier to adjust), because the apparent impedance at the feed point depends on what's being fed to the other feedpoints. Clearly, a lot tougher than just figuring out what element currents you want and driving to get that.
On transmit, one can develop transmitting radiators that have low mutual coupling, which means physically small. This leads towards a sort of "isotropic radiator" model, because physically small radiators have low directivity. Unfortunately, they also have low radiation resistance, which generally implies high Q (narrow bandwidth) because the Rr is a small fraction of the reactance. That low radiation resistance also means that they may be inefficient (electrically) because the loss resistance is large compared to the radiation resistance.
Finally, ground losses are something that can't be neglected on transmit, although, for two radiating elements at the same height, one might expect the ground losses to be similar, regardless of size. However, this is really a near field issue, and needs some analysis.
Consider a compact loop as the Tx element: One can tune it with a variable capacitor. How fast can one tune it? A screwdriver antenna is kind of slow to tune, but a standard variable capacitor only requires half a turn to go from min to max C, and 1 second is a very reasonable speed. The typical compact loop has a 3:1 or 4:1 frequency range. The capacitor voltage is quite high on loops: corresponding to the high current in the loop, which has very low impedance; CV^2 = LI^2
On tuning speed.. Small "bumps" in frequency, or even scanning through the band, shouldn't be a real problem. Typically, it's only a few pF out of tens or hundreds that you need to move (because LC is inversely proportional to frequency squared, C varies as the square root of f). Band changes, though, require a more substantial change in C (or switching to a new loop, perhaps with traps? A trap loop?) which can take some time. (many, interminable, seconds on a screwdriver, for instance).
However, in real life, you don't typically change transmit frequency from band to band in seconds (contesting aside). For receive, one is conceivably using a different non resonant antenna (i.e. a voltage probe, etc.) and that doesn't even need tuning. So, the long time between bands might not be an issue.
In fact, why not build single band transmit modules, including a radiating loop? The loop can act as the resonant tank/harmonic filter for a very simple RF output stage. One can size the loop for a 1-2 ohm radiation resistance on the band of interest, which provides a nice match to the typical FET output impedance, or for that matter, size it according to the Q (if it's too narrow, it's hard to transmit voice!). Even better, the typical transistor output stage is really a current source, and with a high enough Vcc (or, more properly Vdd) you'll actually create some of the reactive power right in the amplifier. Given the fairly high Q, you're still going to need an adjustable capacitor, but it doesn't need nearly the adjustment range.
radio/antenna/phased/active1.htm - 3 Feb 2003 - Jim Lux
(phased home) (antenna home) (radio math) (radio home) (Jim's home)