[Lf] Freq stability

[Andre' Kesteloot]

Talbot Andrew wrote:

>      -----Hi.  I am nearing the end of writing my
>      column in the local amateur radio journal and I
>      am a bit thin on
>      frequency stability guidelines for working QRSS
>      and BPSK on LF.  Nobody in ZL has yet got going
>      on these modes but I want to publish info that
>      does not give any locals a bum steer as to what
>      to do to get into digital modes on LF.
>       Could you fire me some summary comments on what
>      frequency stability should be achieved by
>      transmitters and receivers to join the game, for
>      respective modes?Cheers,Bob ZL2CA
>
> Hi Bob - I've copied this reply to the LF Group as well
> for general information...........QRSS ( 'SLOWCW', why
> invent meaningless Q codes for the sake of it) as usually
> practiced now, calls for a resolution bandwidth of 0.3Hz
> or wider in order to see the 3s dot period.  But as this
> mode is always viewed on a spectrogram display generated
> from an FFT, we are actually seeing a large chunk of
> spectrum, typically many 10s to 100Hz simultaneously and a
> drift of a 1Hz or so probably won't be noticed.  Thus the
> frequency stability needed for this mode can be a very
> poor 10 parts per million - which even the cheapest and
> nastiest crystal can manage - hence one of the
> popularities of the mode.  A 10s dot peroiod is
> occasionally used calling for a 0.1Hz bin size or less but
> still a few ppm accuracy will suffice.For BPSK, a
> frequency stability of the order of a tenth the bit rate
> is needed,a lthough a figure of better than this speeds
> initial lock up.  Once locked, the frequency can drift
> outside this range, and manual assistance ina chieving
> initial lock by 'nudging' the tuning process can allow
> worse initial frequency setting accuracy.    So for the
> 100ms (10B/s) rate we normally use, 1Hz stability is
> ideally needed, better to aim for 0.2Hz and not have to
> play around with initial acquisition - still well within
> the capabilities of any crystal oscillator.So to
> summarise,  by recommending crystal oscillator stabilities
> you won't be giving any wrong steers - 1 part per million
> should be the stability people should really aim for, but
> a factor or so worse than this can still be useable with
> Spectrogram type modes.I myself use a Direct Digital
> Synthesizer source driven routinely by a TCXOo scillator
> adjusted up for a repeatable 0.1ppm.   The DDS tunes in
> 0.001 Hz steps and without accurate calibration achieves
> 0.03Hz accuracy, with calibration against off air signals
> (eg MSF in the UK) 0 .001Hz is possible.  For times when
> some distant stations actually get set up for some serious
> experimentation with very narrow band modes, I have
> Droitwich and TV sync locked frequency sources - both
> hangovers from microwave operating, (all home station
> operators and several /P have constructed one or the other
> of these). All these give a frequency setting accuracy of
> a few parts in 10^-9 - about 1mHz at 137k.For calibration
> purposes, a GPS receiver giving 1s pulses drives the
> external clock of a frequency counter allowing 1000s /
> 10ks (2.78 hour) or 100ks (just over 1 day) gating
> periods.    Also have a Caesium standard but prefer to
> only turn it on once in a while when needed as the tube is
> nearing the end of its life and its usefulness for LF is
> debateable.  For calibration the GPS pulses are traceable
> to international standards and can be used immediately,
> whereas the Cs standard would need to be set up each time
> and run for several days to guarantee its few parts in
> 10^-11 accuracy.Sorry to sound a bit cynical about LF
> frequency accuracies and stabilities, but having spent
> over 10 years with narrowband operation on the microwaves
> bands,o ften from exposed portable locations, where the
> 1ppm accuracy that very few LF operators bother to aim for
> is several SSB bandwidths, I am rather surprised at the
> difficulties many seem to have and the poor stabilities
> still being seen on LF now.  At least the disgusting LC
> oscillator has faded out, but the norm now seems to be two
> VCXOs mixed, which I suppose is comparable to an LC VFO on
> the HF bands, not something seen too often now,
> fortunately.   Up on 10GHz a low cost crystal heater is
> the minimum anyone ever uses and 1ppm is usually the
> result.  Operators routinely calibrate with beacons and
> other stations throughout an operating period and everyone
> knows at least when they are within a 3kHz SSB bandwidth
> tuning range.Ironically, HF has the lowest requirement for
> frequency stability.  With ionosphericp ropagation causing
> several Hz shifts over minutes, as well as multipath with
> several millisecond delays, who needs better than 0.2 ppm
> frequency accuracy?Andy  G4JNT
>
> --
> The Information contained in this E-Mail and any
> subsequent correspondence
> is private and is intended solely for the intended
> recipient(s).
> For those other than the recipient any disclosure,
> copying, distribution,
> or any action taken or omitted to be taken in reliance on
> such information is
> prohibited and may be unlawful.