...The antenna switching frequency would be AF (500-2000Hz) which would produce that frequency tone. ...May want to try helical polarization on the antennas so the antennas can be wrapped around the airfame.
I think your right, and circularly polarized antennae would be a good choice, helical or otherwize. not only because they are compact, but to help insure good reception in all attitides.
Another variant on this theme would be to use 2 RF front-ends instead of an antenna switch, and feed both directly into a phase discriminator. That will directly output a DC voltage in proportion to the phase difference.
This might be more practical if the RF beacon you are using is also part of a telemetry radio transceiver system, carrying FSK or otherwize modulated data.
I think this will also exibit a sensitivity curve similar to the accelerometers - meaning that when the line between the 2 antenna is directly parallel to the RF propagation path, you will have the lowest sensitivity to rotation. What you want is the other way 'round i think, so i would place the antenna side-by side, perhaps at the fin tips or another suitable place, depending on the separation needed. that way, when the rocket is straight up, you have no phase difference (a null) but the slightest deviation from 'vertical' will produce a very large change in phase.
Of course there is a downside to this: now you need at least 3 antenna to cover both X and Y rotation of the airframe.
This would favor a higher frequency in smaller rockets:
at 433 Mhz, 1/4 wave = 17.3cm , just under 7"
at 915 Mhz, its 8.2cm, or about 3-1/4" which could fit inside a 4" airframe.
at 2.4 GHz , it's 3.12cm, side-by side antenna could fit in a 38mm body tube.
at 5.8 GHz, youre at 1.56cm, or about 5/8" - you can guess where that fits!
I would be temped to try it a 2.4Ghz, which is worldwide ISM band that allows relatively high power transmitters. its pretty trivial to put together a high power narrow CW beacon at this frequency. something in the 1W-10W ballpark should be enough to cover just about any HPR flight.
You don't actually need an exact 1/4 wave difference, but the sensitivity will be reduced as you move away from that.
Also remember the actual feedlines to the receiver fromthe antenna have to be equal length, otherwise there will be a fixed phase bias that won't help and may cause problems.
I'm starting to really like this concept; because you're not interested in any data on the RF signal, the receiver can be extremely simple and cheap, and your transmitter beacon too. Now if you combine that with the IR horison sensors and a little intelligence (MCU), we may have a real winning and reliable solution!