[Date Prev][Date Next][Thread Prev][Thread Next] - [Date Index][Thread Index][Author Index]

Re: PSK Modem

Allen Platz wrote:

> Hi all, I see a lot of people are using the PSK-1 for AO40 telemetry
> decoding. As I was going through the junk boxes I found an unbuilt TAPR PSK
> Modem that I bought years ago at a hamfest. Is anyone using this/can it be
> used for AO-40 ? Does it have any known problems (I see tapr doesn't seel it
> any more) ? It looks like it should be possible, I see it wants a clock
> signal from a TNC to work. Any thoughts ?
> thanks
> Allen

Hello Allen -- the TAPR PSK modem was designed by Eric Gustafson (N7CL) and
me for the 1200 baud PSK signals from the Microsats (AO-16 et al) and the
early Japanese FO-series digital satellites; it was released at about the
same time (1988 as I recall) that James Miller, G3RUH released his PSK

Both designs assumed that the 2 kHz audio bandwidth of an SSB radio was
wide enough to handle the 1200 BPS PSK signals when the signal was tuned to
a center frequency around 1600 Hz. Both designs had tracking loops that
permit a few hundred Hz of frequency error (due to mistuning or Doppler
shift). Both designs looked at the tuning error and when it got too big, an
AFC-like correction was done to the radio's tuning by clicking the up/down
mike button connections.

Our design used an analog signal chain with 1200 BPS matched filters inside
a Costas loop. James's design hard-limited the signal (i.e. with a clipper)
and implemented its signal path with digital mixers. It can be shown (and
it proved to be the case when we compared the performance of the two
designs) that the linear analog approach has a sensitivity advantage of
pi/2 (the Van Vleck correction, about 2 dB). 

Both of these designs use Costas loops designed for 1200 BPS packet
operation. As a result, the lo-pass filters in the two sides of the Costas
loop are "tuned" to about 600 Hz bandwidth. The Manchester encoded PSK
signals from P3D (and all the earlier P3 series) require only 400 Hz
filters, so when these demods are used on P3D you take a S/N hit of 600/400
or about 2 dB from optimum.

Also in ~1978, Karl (DJ4ZC) released his very simple AFDEM demod which
worked fine but only had a ~100 Hz range tracking loop. A few folks still
have the original AFDEM.

The P3D data framing consists of a 4 byte sync vector (hex #39 #15 #ED #30)
followed by 512 data bytes and a 2 byte check-sum. The TAPR modem has no
hardware recognition of this sequence (it was designed for packet use), so
frame sync has to be provided in the decoding software. Also, your software
will have to provide the de-Manchesterizing clock sync operation. I'm not
sure if Stacey's P3T software does this or not.

The modem portion of TAPR design was taken over by Pacomm in their PSK-1.
The PSK-1 included a small microprocessor that performed the #39 #15 #ED
#30 sync recognition and Manchester clock recovery. I believe that Bob,
N4HY was involved in adding this to the PSK-1. Thus the PSK-1 is a superset
of the original TAPR design.

All these designs (and the Phase-3 PSK telemetry definition) are old enough
that they had to implement the modulator function (in the satellite) and
the demod function (on the ground) in hardware with real chips, resistors
and capacitors soldered onto circuit boards. On the spacecraft end, a  more
modern would have used convolutional encoding and FEC. On the ground, the
modern approach is to implement the demod function in software. Whether an
IF of a few hundred kHz (some might cal that RF) or a few kHz (which some
might call audio from an SSB radio) makes little difference because we
digitize it at a well-defined sampling rate (like using the A/D converter
on a SoundBlaster with the "audio" approach). We then implement all the
modems in DSP software without needing to sacrifice performance and without
ever needing to solder a wire.

73, Tom
Via the amsat-bb mailing list at AMSAT.ORG courtesy of AMSAT-NA.
To unsubscribe, send "unsubscribe amsat-bb" to Majordomo@amsat.org