[aprssig] frequency control on HF

[Jim Lux]

>
>Date: Sun, 2 Oct 2005 09:21:43 -0700
>From: "Jon Adams" <jon at jonadams.com>
>Subject: [aprssig] ARPS-HF Frequency Stability Question
>
>Good morning all -
>
>I've been running HF APRS for over 10 years and as an HF Gate for nearly
>that long. When I listen to the traffic, I hear stations high and low of any
>one frequency, and if they're more than a few dozen Hz off, they don't
>decode very well. (running Kam+).
>
>I don't know how every station sets its operating frequency, I'd heard some
>time ago that some transceivers wouldn't let the station transmit if the
>dial (no matter where the modulated signal really was) above 10,150.0 kHz,
>so for a while it seemed that a block of stations were at that frequency.
>Today, I'm hearing stations at 10,150.4. I go over and check WWV and it's
>where it's always been.
>
>I know it gets easy to get into a loop with tuning - station B tunes to Sta
>A, C tunes to B, etc., until A discovers Z off-channel, and nudges the dial,
>and the whole thing starts anew.
>
>If there isn't a procedure already in place, then I'd suggest that a 24x7
>station in the mid-country be the reference - K7PW, K5RAV, N5NGN, N0LRF,
>KM5HG. I don't know which of those is more stable than the other, but maybe
>they can decide among themselves and one be the "reference". Once they've
>established that, then we can all tune to that one and lock the dial (unless
>your own station drifts). No station would retune themselves because of some
>other off-frequency station.


I suspect that lots of these radios are just inherently not stable enough 
to "lock the dial".
Realistically, you need something like better than 1ppm stability over time 
and temperature (10 Hz at 10MHz)

What you're talking about is a sort of distributed frequency reference 
distribution, which might not be possible without some extra hardware at 
the stations, i.e. a run of the mill surplus Bell 202 modem (or equivalent) 
ain't gonna do it.  Station A would stabilize their frequency with a high 
quality reference (ovenized oscillator & discipline to check it 
periodically and retune for aging, GPS disciplined reference,etc ). Station 
B would tune to match, then station C could use station B as a reference.

One problem is that Station B needs to have reasonably good short term 
stability (what's B's retuning interval?).  You'd also pick up a few Hz of 
frequency uncertainty with every "hop".

It also requires that the stations be able to measure the frequency of the 
received signal, which isn't necessarily provided by the hardware that the 
hams now have.

If you're using a computer as the modem, then it gets easier... assuming 
you could get the modem software writers to provide a hook for "observed 
frequency offset"..

Given we're talking about APRS, and many GPS receivers provide a 1pps 
that's reasonably accurate, maybe you just need a little outboard widget 
that samples your transmitted RF (or maybe an LO) and counts it against the 
1pps, and that is fed back to your radio to retune.  It could easily be 
done in a single PIC.

But maybe the real solution is to use something other than a 1950s legacy 
modulation format that was designed for phone lines that don't have 
frequency shifts.  I doubt many hams are modifying old half duplex 
BellSystem rackmount modems for their HFAPRS work.

Use a format that is tolerant of a receive frequency mismatch of 50-100 
Hz.  Then, everyone can just transmit on their best guess as to the 
frequency, and as long as they are within 10ppm or so, they're good to 
go.  Then, the software demodulator could also look at the apparent 
received signal frequencies and using some sort of algorithm: either using 
"known good stations" or some sort of ensemble average, weighting to keep 
it in band.

Hmm... why not something wild like Mil-Std-110A/FS1045?

Jim, W6RMK.