[amsat-bb] FM birds

Phil Karn karn at ka9q.net
Wed Jul 30 04:05:08 UTC 2014

On 07/27/2014 09:57 AM, Jim Sanford wrote:
> Tracking doppler on 2m and 70cm was hard enough at field day.  I like
> 2GHz and up, but do not want to even think about the doppler.

At least not with analog modes.

High speed digital is very well suited to microwave in LEO as the link
budgets generally support high data rates even with small antennas and
low powers. And it's the ratio of the Doppler to the symbol rate
(channel bit rate) that matters.

Doppler correction of analog signals is inherently difficult even with
computer tuning for the following reasons:

1. Inexact knowledge of satellite position, especially for high-drag
situations (small cubesats in low orbits).

2. FM can be closed-loop DC tracked if there no low frequency modulating
components, but SSB has no pilot or carrier for closed-loop tracking.

3. Hard-to-model ionospheric delay variations.

Computer tuning of standard analog radios is problematic for the
following additional reasons:

1. Little standardization (none across vendors) in radio control
interfaces. Most seem to still use slow RS-232 type links.

2. No standardization in how the radios respond to a tuning command. How
long does it take to settle on the new frequency? What trajectory does
the instantaneous frequency take to get there?

3. If the signal is an appreciable fraction of the (usually narrow)
receiver bandwidth, frequent retuning is necessary, aggravating the
problems of #2.

A wideband software-defined-radio (SDR) front end operating on a
well-designed digital signal is much better, especially when the A/D
bandwidth is significantly greater than the signal bandwidth (as it
usually is). Then only an occasional and relatively large tuning step
need be made as necessary to keep the signal within the passband.

Most SDRs have fast USB or Ethernet interfaces so the demodulator can
time each retune to occur when it will do the least damage, such as
between frames after frame sync has already been acquired. Even if some
damage is done to the signal, it can usually be corrected by the FEC.

Most SDRs produce a "complex" (two channel, quadrature, I&Q) digital
stream that allows the signal to be converted directly to a
near-zero-frequency IF without zero-frequency folding or aliasing. (This
was a common problem with old analog direct-conversion receivers.) I.e.,
quadrature sampling distinguishes negative and positive frequencies.

The SDR doesn't need to actually keep the incoming frequency exactly
centered at 0 Hz; it only need be somewhere within +/-0.5 times the
sampling rate. For example, the Ettus USRPs being used at Arecibo and
Bochum to receive the ISEE-3 telemetry signal use an effective complex
sampling rate of 250 KHz. As long as the signal is somewhere between
-125 kHz and +125 kHz, nothing is lost.

Like the vast majority of deep space probes ISEE-3 transmits residual
carrier PM, so my first step is to locate the carrier with a FFT. It is
typically around +16 kHz in the Arecibo signals, varying slowly due to
Doppler as the earth turns. I then "spin down" the signal frequency so
that the carrier is at exactly 0 Hz and on the I (inphase) signal axis.
The BPSK data lands on the Q (quadrature) axis, with the sidebands
symmetrical around 0 frequency. I do this by multiplying the input
samples by a complex sinusoid, all in software, set from the results of
the FFT. This gives me extremely precise control of the instantaneous
carrier phase and frequency, with no uncontrolled jumps.

I included a hook, as yet unused, to "chirp" the digitally-generated
carrier precisely in frequency to track slowly varying Doppler. Although
I'm not actually using it yet, I could generate this chirp from
calculations of what the Doppler should be (subject to the above
limitations on knowledge of the orbit, the effects of the ionosphere, etc).

Not only is this a much better way to demodulate satellite signals, but
it can be considerably cheaper too when you eliminate all the
unnecessary components of a typical "conventional" receiver.

The ironic thing is that many modern ham rigs work much like this
internally, but they were designed specifically for the traditional
analog (and slow digital) modes and most don't give the user any easy
way to gain access to the internal DSP mechanisms. They try hard to make
their radios look purely conventional, and they succeed.


More information about the AMSAT-BB mailing list