Tom Clark, K3IO wrote:
Robert McGwier wrote:
But the discussion here is not about the Microwave phased arrays. Given the rate at which we are signaling and the demodulator requirements that will entail, the spin modulation is a minor factor on the ACP. For the very low bit rate SMS text messaging, and the 70cm uplink, where we were planning absolutely no phasing whatsoever, it becomes a major factor and might even be a deal killer without greatly increased complexity in the 70cm and 2m system. Essentially, you would have to compute and predict a phase offset from the (say) 70 cm patch and apply the correct again before we HELAPS or while we HELAPS on 2m, given information from the phased array for other bands and multiply each sample by the time varying phasor correction in the SDX to compensate.
I think you are over estimating the complexity of the computation. For a non-phased antenna, the amplitude of the phase correction needed is (he pk-to-pk value is twice this):
2pi * sin(angle between boresite and earth) * (distance phase center is offset from spin axis/lambda)
The path that the element makes w.r.t. fixed observer is an ellipse. I seem to recall that any ellipse can be expressed as the sum of two circles of radius R and r, where R*(1+e) = the semi-major axis and r is the semi-minor axis r=R*(1-e) (where e is eccentricity). Let the center of the smaller circle "ride" the larger circle, let the larger R circle spin one turn per orbit prograde (the direction the satellite moves) and the t circle spin retrograde (in opposite direction). Pick the spot on the small circle (x,y) so that R+r is the apogee and R-r is perigee. Then the equation of motion of the spot will be
x = R*cos(Wt ) + r*cos(-Wt) = (R+r) cos (Wt) and y = R*sin (Wt) + r*sin (-Wt) = (R-r) sin (Wt) , where W (really omega) = 2*pi*(spin rate)
I am indeed thinking about the cost of computing the phasor and doing a complex multiply on every sample coming in and going out in addition to all of the other tasks we know we must do. The angle with respect to the center of the hemispheric mass changes continuously throughout the orbit so the eccentricity is a function of the position in the orbit. The antenna paints a circle at "apogee" if we are perfectly nadir pointing then and not a circle elsewhere in the orbit segments we care about. At the nadir pointing place in the orbit, the phase distance to the center of the visible earth hemisphere is constant so no correction is needed and the correction needed changes before and after this does it not? I hope we can parameterize a model and not have to solve for the state vector but this will still not be even as simple as this calculation. I think as you say, we can compute what we need from the interferometer but please do not overestimate the computational power of the DSP chip. At 100 kHz doing STELLA and HELAPS predistortion, we are going to have to be careful with this computation. I would like to do this de-spin before we apply any processing and then we must figure out the appropriate place in the HELAPS construction going out the door.
Off to teach class, later!
This is sometimes called the equation of a central ellipse and the method dates back to Eudoxus of Cnidus (c. 400-347 B.C.) and then to Hipparchus of Rhodes (c. 190-120 B.C.) (http://astro.isi.edu/games/kepler.html -- see Fig 1) although it is sometimes attributed to Ptolemy (c. 85-165).
So what the ancient Greeks tell you is that you can tweak one LO in the system (either TX or RX) to have a phase offset that is the sum of two contra-rotating phasors running at the spin rate, with the phase, amplitude and eccentricity determined by geometry determined from the S2 interferometer. No math higher than trig is needed.
73, Tom