A year ago I was drowning in zoning and city council meetings over a proposed dish ordinance that would have prevented me from grandfathering my farm. Along the way I had an evil vision that if the ordinance would pass, I could theoretically replace my BUDs with a bunch of 1m dishes configured as a phased-array. I didn't actually set this up, but I performed an engineering proof-of-concept to demonstrate it could be done without exotic technology (mostly just a bunch of coax switches, measured coax segments and one or more splitters). At the following meeting I gleefully presented a photograph of my 3.2m dish, which they thought they could force me to take down, and rendition of 16 OTARD-protected dishes that would equivalently replace it.
Following the gasps a board member asked the city attorney whether they would have any jurisdiction, to which the attorney said they would not (I had embarrassed him in several previous meetings on OTARD and amateur radio antenna provisions). This was a key turning point, and since I had the city attorney on the run, I then took him to task over a Colorado Supreme Court ruling in which an adult entertainment establishment prevailed over a similar attempt to obviate grandfathering. Most of the council's enthusiasm evaporated and an emaciated, token ordinance was passed instead. So while there is now an ordinance on the books, it has little or no effect on me.
In case anyone cares to take this a bit further, the LNB's reference signal can easily be injected through a TV/satellite diplexer. In terms of accurate phase combining, you need a method of compensating for the delay at the carrier frequency and the symbol rate. If one does this at a nominal 1.2 GHz LNB output, the carrier wavelength is 25 cm. If you can match this to say 1/10 or even 1/20 of the wavelength, you will get most of the gain available from aperture combining. For the IF phase this means you will need a series of switchable coax segments that allow you to get within 1-2 cm (also remember signals travel slower through coax). Thus one could employ four switches to select segments of 12, 6, 3, 1.5 cm for example. Accuracy of the individual segments is not that critical because you will switch them in/out to hunt for the best CNR anyway.
Unfortunately for high SR signals, this isn't good enough as the overall path delay will cause symbol phase distortion. So you need to calculate the bounds of the dish and signal paths and be able to match this part analytically. A SR of 45000 has a wavelength of about 6.7m, so you need a similar array of switches to get this down to 10-20 cm. Overall you might need as many as 10 coax switches per slave dish, most of which are always set to the same settings for a given satellite. The last four probably need to be tuned each time, and as the bird drifts in its box. I didn't analyze how quickly this would have to be done, because I would not have chosen this path in the end.
Instead I would have taken the signals down to baseband and digitized the data. The aperture combining then becomes fairly trivial, but I would either have to wire this into an off-the-shelf demod or implement my own DVB-S(2) demod in software. This type of project is more for bragging rights than practical use, but I'm only showing it might be possible to do this with low-end technology.
