[Lf] m-FSK: SNR vs bandwidth]

[Andre Kesteloot]

James Moritz wrote:

> Dear LF Group,
>
> I seem to be the only station so far that has had some success with both
> WOLF BPSK and 7FSK modes over transatlantic paths, so here is my comparison:
>
> W!TAG was able to get perfect copy on my WOLF signal, so BPSK definitely
> does work for DX communication; however, he remains the only transatlantic
> station to do so, in spite of weeks of trying. In part this can be put down
> to the varagies of propagation, but also many people struggled to get WOLF
> to work well. The reason for this seems to be mainly the requirement for
> accurate frequency and timing calibration (of the order of 1ppm), not a
> trivial problem to overcome since the equipment most people are using was
> not really designed with this in mind. Operation of the software and
> hardware is not intuitive like QRSS/spectrogram programs.Transmitting using
> this mode is relatively difficult too; although very simple QRP
> transmitters are possible, envelope shaping is required for higher powers
> to avoid causing unacceptable QRM, making high power transmitters
> significantly more complex. In the context of the European 136k band, the
> bandwidth of some 10s of Hz required by Wolf would be a problem if the mode
> was being used by several stations simultaneously. None of these are
> fundamental limitations - ingenious frequency calibration schemes have been
> devised, some success has been had with modified "variable phase" keying to
> control sidebands without amplitude modulation, it would be possible to use
> narrower bandwidths for BPSK too. However, BPSK does require significant
> time and effort on the technical side to get working.
>
> I would not claim any great technical superiority of 7FSK over other modes,
> but from a pragmatic viewpoint it has several good points. It can be
> generated by a simple extension of existing DFCW techniques, and received
> using the spectrogram software that people are familiar with. The frequency
> accuracy required is much more relaxed (of the order of 10ppm on receive),
> and timing accuracy isn't really an issue. The bandwidth, although greater
> than that of QRSS, is small enough. In its short history, readable 7FSK has
> already been received by a number of transatlantic stations, so it
> definitely works too. The big advantage is of course in the increased speed
> compared to QRSS.
>
> As for the SNR vs. bandwidth considerations, other subscribers to this
> reflector have more expertise than me - but I would observe the following:
> Both QRSS and 7FSK are made up of dots, so the probability of any one dot
> being corrupted by noise is the same. However, 7FSK transmits on 1 of 7
> possible frequencies, while QRSS uses only one, so it is 7 times more
> likely that a corrupted dot will appear in a 7FSK signal in a given period
> of time. However, fewer dots are required by 7FSK to send the same number
> of characters by a factor of about 5.5, so the relative probability of a
> character being corrupted in 7FSK compared to QRSS would be 7/5.5, ie. 1.3
> times as likely; not very different - about 1dB. I realise this is only
> approximately true, not considering redundancy and so on, and would apply
> only when the ratio of corrupted dots to uncorrupted is small, but you
> would need this anyway for a QSO, and the tests over Christmas and before
> show that these conditions do exist fairly often.
>
> I think long-distance LF communications can be done with either of these
> modes, and probably many others too. Theoretically,  I think the more
> sophisticated modes like BPSK will win out, but will always have limited
> popularity due to the more elaborate technical requirements. A mode like
> 7FSK may not ultimately be as good, but it is quick and easy to get going
> and operate. It is interesting that in the US Lowfer community this winter
> has seen an explosion in the number of QRSS beacons and reception reports,
> while last winter  WOLF BPSK was a la mode. The distances being covered
> don't seem to be much different, but more people are getting results. Where
> there is an unpredictable element involved, like propagation, the chances
> of the success of any particular mode are going  to be tied up with how
> many people have the capability and the inclination to use it - a
> technically mediocre mode in use by 100 stations will probably get more
> positive results than an excellent one being used by only two stations.
> This is no excuse for complacency of  course, but factors like ease of
> implementation, compatibility, inertia and "does it match the curtains?"
> all comes in to how successful something is - just look at the history of
> the personal computer!
>
> Cheers, Jim Moritz
> 73 de M0BMU