Many thanks to each of you who have replied to my inquiry.
VE7LDH: I am curious what got you to thinking about Doppler frequency tracking on the BPSK signals. Did you consider a specific approach on how? Had it occurred to you that it might have application beyond BPSK or AO-16? If you would rather not post the reply, you may contact me directly.
KD6OZH: Your reference to pilot carrier technique allowed me to find some interesting web reading, mostly IEEE and patent stuff. QEX had nothing of interest and the AMSAT journal does not seem to be available prior to 1995. Pilot carriers seem to have been proven to be workable, but not advantageous in the long run. I do not have the expertise to demonstrate, but I suspect that a pilot carrier required a minimum power to acquire and maintain lock on the signal and that the more modulation power being sent, the more pilot carrier power needed to maintain the lock. I will comment more below, but with a Costas Loop, the more power in the modulation information, the easier it is to acquire and maintain lock. As a result, a pilot carrier consumes power without benefit, while at the same time, as you suggest, accurate satellite data allows the shift to be predicted. Tuning the radio becomes something that can be learned by the "handy" or built into the radio. From my limited experience it seems intimidating and more than a touch expensive. Nothing suggests that DSB will work, or that it might become popular if it did work.
If DSB had ever been considered, it seems to have been cast aside pretty quickly. That may have been due to the double bandwidth of DSB. I was somewhat puzzled that AO-16 had used DSB downlink. I can only assume that the weight penalty for an SSB transmitter was traded for what was intended to be single channel BPSK. If the downlink design had been multiple BPSK, it would have required a more powerful transmitter and DSB bandwidth would have been a bigger problem.
KD6BD and ZL1TYF: I am glad that I sparked an exchange of ideas.
My first thought concerning CTCSS is that it shouldn't be necessary. The time for the Costas Loop to lock should be no worse than the time for a VOX circuit to key a transmitter. That is an opinion based on faith in the concept and the building skills of someone much better than I. But I suggest that it would do no harm. At least no more harm to the equipment than someone using very long AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHs! between words. The Costas loop would be able to maintain lock. It fact it seems to me that the level of CTCSS should be specifically set low, to barely maintain lock. The engergy to acquire the signal should come from speech, particularly intelligent speech rather than someone whistling into the mike.
As for a standardized tone, and again this comes from my personal bias that a tone should not be required to make the system work, I fail to see the purpose of standardizing the tone. The Costas Loop will not care what tone is used. It could even be a BPSK8 signal at 50 Hz with your callsign ID and latitude / longitude co-ordinates. I would not suggest using a 1KHz tone unless you expect everyone listening to have a sharp filter, or just listen to it. Adjust the tone for "minimum lock maintenance signal strength with no audio" to reduce the annoyance.
You asked about multiple overlaping uplinks. I am not sure that I understand. If we are talking about AO-16 with an FM uplink, is this really a problem? If this was a translating repeater ( sorry if I used the wrong term, I go back before OSCAR, but I am pretty new to this stuff. It took me several hours to find a clear reference to TCA. Once I did it was obvious that is the point that has zero doppler shift and maximum doppler rate... ) then two signals in the uplink passband would result in two signals in the downlink. With an FM receiver providing audio to a DSB transmitter, you are going to hear what the FM receiver locks onto. If that is one strong signal, or the interference between two weak signals that cannot maintain lock, you are going to get one signal to downlink. Here is a slightly ficticious example: I live in Kansas and tune my FM radio to a frequency unused in the local area, but that is used in both Denver and Dallas. Most of the time, I don't hear any music from the radio. If a meteor streaks over Cheyenne, I hear Denver until the meteor trail fades. If a meteor comes in over Austin, I hear Dallas. If the meteor comes overhead, I hear the overlap of Denver and Dallas which is probably unintelligible.
Using the measured downlink Doppler shift to derive the proper uplink shift. By theory, the shift is proportional to frequency. So multiply the downlink shift by the ratio of the uplink frequency to the downlink frequency. For AO-16, it is a waste of effort since the uplink is FM. If on the other hand you had a DSB uplink that you were monitoring to correct your shift through a translating repeater, you would a much more complex situation that would depend not only on the uplink and downlink frequencies, but the inverting or non-inverting type of repeater... The control function math is getting me confused so I think you would need to break the loop somewhere to prevent instability or oscillation. What my years of flight control development tell me is to design the system from the ground up with each transmitter fixed and just compensate at each receiver.
If AO-16 is manifesting a residual "pilot carrier" it may be an indication of an imbalance in what I can only presume is a double balanced mixer. On the other hand if the uplink FM receiver does not have the functionality of a squelch, there may be a "real" DSB downlink of the audio received by the uplink. The difference would be obvious to a synchronous detector.
Thank you again for your input.
James
n5gui