On this matter...
For the recent talk I gave on weak signals at a Microwave Roundtable (
http://www.g4jnt.com/MartleSham.htm ) I made some simulated CW in
Noise using accurately calibrated S/N levels.
An interesting finding came to light, if you normalise the signal rate
to the S/N, so making bandwidth irrelevent, most of the 'fuzzy' modes
end up with a similar capability. In other words, Aural CW, QRSS,
SMT Hell all need a similar S/N at their respective bandwidths to
work. The actual normalised S/N for readability is subject to the
operator's experience togther with Temperature/Time of
day/mood/Age/Gender/Alcohol intake/Hunger/Weather or any other
similar parameter, but there's no massive differences between any of
them.
So its all down to bandwidth. Even machine modes without error
correction manage a not-too-dissimilar performance once their data
rates have been normalised.
However, if you spread the signal, intentionally, by adding FEC, the
improvements can be enormous. As I think we all know only too well
when comparing QRSS etc with WSPR, Wolf and other modes with heavy
FEC. At least on microwaves we have the luxury of virtually
unlimited bandwidth, so can operate in a true Shannon power limited
channel to make the most of band spreading.
Incidently, the usually quoted signal efficiency used on one axis of
the Shannon curve "Bits/second/Hz" sounds uninspiring. But if,
instead of Hz, you use the old term 'cycles per second', it becomes
Bits/cycle. Which puts a whole new meaning and explanation to the
axis on the Shannon curve, and elicited an "Oh Wow, yes, that IS an
interesting way of putting it" when told to an experienced comms
engineer. The term was actually used in Shannon's original paper of
1948, but seems to have got lost
Andy
www.g4jnt.com
This email has been scanned for damaging side-effects by the health
and safety police
2009/12/11 Johan H. Bodin <[email protected]>:
> Hi Stefan,
>
>> Or isn't it possible to give such a relation?
>
> Yes, it is not only possible, it is in fact quite simple:
> When the speed is reduced by a factor K, the information bandwidth is
> also reduced by the same factor. This allows you to use a receiver
> bandwidth which is K times narrower without missing any information. The
> nice thing is that the noise power passing through this bandwidth is
> also K times smaller - The S/N ratio has improved K times (or 10*log(K)
> dB if you prefer). In other words, you can reduce the TX power by the
> same factor K and still enjoy the same SNR (if RX BW is is also made K
> timer narrower).
>
> In visually received QRSS, the receiver bandwidth is equal to the RBW,
> the "resolution bandwidth", which is approximately equal to the FFT bin
> width (one pixel on Argo).
>
> QRSS30 is 10dB more efficient than QRSS3, in theory at least.
>
> 73
> Johan SM6LKM
>
> ----
>
> Stefan Schäfer wrote:
>> Dear LF,
>> Does anybody know about the "gain" between QRSS3 and QRSS10 or QRSS30? I
>> mean, if the noise in both cases is equal, how much can I reduce my tx pwr
>> when switching from qrss3 to qrss10? Or isn't it possible to give such a
>> relation?
>> And: Was there ever a TA QSO in QRSS3?
>> I am new on the reflector, sri ;-)
>>
>> Stefan / DK7FC
>>
>>
>
>
|
