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

Re: [aprssig] 1200 AFSK PACSAT Uplink [was: AO16/LO19 Traknet Operational!]

KD6PAG asked:
> However, with the LINUX 'soundmodem' software, not only is the modem in
> software, but the TNC is as well.  So, no, the hardware solution doesn't
> help.  Knowing the characteristics of the low pass filter, on the other
> hand, might.  Please see my previous posting to John Magliacane KD2BD
> and the AMSAT-BB mailing list for further discussion.  (By the way, i've
> been involved with the synthesis and analysis of musical sounds since
> the early 70's, so this problem is quite trivial by comparison.)
> I am also looking for information about exactly what BPSK encoding does,
> so i can receive as well as transmit.

Since I built the spacecraft receivers on 16/19, let me TRY to clarify.

When we built the original Microsats 12 years ago, we regarded
watts as our most precious commodity. Remember that these satellites are
cubes about 10" on a side, so there is not a lot of solar panel area.

Since the biggest number of spacecraft watts goes into making
power, we opted for the most efficient downlink modulation -- BPSK. 
If you go thru all the arguments, you pay a 6-10 dB penalty (i.e. you
need 4-10 times as many watts) to use straight FSK instead of BPSK.
And the penalty is even larger (like 20 dB) to use AFSK-FM (i.e. 1200 
baud Bell 202 tones FM modulating a voice transmitter -- like is done 
by most AX.25 users). So the transmitter (designed and built by Matjaz,
S53MV) employed PSK for the downlinks, and the users on the ground had
to come up with the DEM side of a MODEM to handle BPSK to receive the
signal. (As an aside -- the UoSATs had bigger solar panels, and hence
more watts, so they opted to use higher powered xmtrs running 9600 BPS

On the uplink, I was driven by the desire to use simple receivers and
opted for straight NBFM. I did not want to have to pay the extra ~10 dB
S/N penalty associated with Bell 202 tones and AFSK-FM. I did not have
room to put in an AFC for each of the uplinks, so to compensate for
Doppler and to ease the burden of having the ground users have to
pre-compensate for their uplink frequency, the output of the FM 
receiver is AC coupled into the bit slicer. 

A stream of data bits at a rate B bits/second has a maximum frequency
of the fundamental square wave of B/2 Hz; this case results when 
the bits alternate between 0 & 1. And, unless you do something
special, the lowest frequency is essentially zero. Therefore 1200
bit/sec packet data needs 600 Hz of bandwidth, going from DC to
600 Hz. Unfortunately most radios don't do a good job passing 
the DC end of this signal! So it doesn't work to just put the 
raw bit stream into the mike jack!

The HDLC part of AX.25 takes care of a part of this problem by
bit stuffing. If the raw data has a run of more than 8 zeroes, 
HDLC stuffs in a 9th bit, and the low-level protocol handlers
(like 8530 ICs) know enough to undo this added bit. This then means
that the lowest frequency is now ~B/16, i.e. ~75 Hz for 1200 b/s

All this leads up to the logic behind the Manchester encoding.
Lets take the zero-to-600 Hz spectrum data and double sideband
modulate a 1200 Hz carrier. The required spectrum is now 600
to 1800 Hz, which is a good match for a standard voice-grade

The modulator is pretty simple -- it turns out that for digital
data, all you need is an exclusive-OR gate, fed with B bits/sec
data and a B Hz square-wave clock. Since the square-waves
have a number of high frequency harmonics, it is conventional
to use an RC low-pass filter to cut off garbage above 2-3 kHz.
So it is the combination of an XOR gate (like 1/4 of a 74HC86)
plus one resistor and a couple of capacitors that WB4APR has
been calling his $0.89 uplink widget.

In the time domain world, the 1200 Hz clock should be phased so
that its transitions line up with the data. This means that there
is ALWAYS a phase reversal in the middle of every data bit.

I don't know about the particular soundblaster DSP implementation
to which you refer. If it is provided under GPL, then the source
code modification to implement the XOR gate should only be a
couple of lines of code. 

There is a lot of similarity between BPSK and Manchester. In
Manchester, it is assumed that the 1200 Hz signal is both a
carrier and a clock signal. The standard amateur BPSK implementations
use an audio carrier which does not convey any clock information.
Hence it is often centered in the audio passband of the
receiver around 1400-1500 Hz. If it was precisely locked to
the data (i.e. at 1200 Hz), then it could be used as a data clock.
If the Manchester output for the $0.89 wonder were put into
a linear frequency translator (i.e. an SSB xmtr), the the signal
would be BPSK.

Did this help?

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