[Lowfer] 8 KHz bandwidth for 6 lowFER stations 11-24-20

[Clint Turner]

Hi,

A quick look at the source code (for the WSPRD program) will reveal that 
the SNR is actually calibrated at a very narrow bandwidth and the value 
is normalized to 2500 Hz (the SNR reading - which is a power ratio - 
would be meaningless without specifying a "bandwidth").  For example, a 
26.3 dB offset is applied in "mode 2" signals (120 second) which, if you 
do the math backwards, implies a bandwidth of a bit less than 6 Hz.

What this means is that for most cases, the receiver bandwidth is 
totally irrelevant, and will not affect the SNR reading that WSPR 
reports simply because measurements are being taken at a far narrower 
bandwidth than that to which a receiver is likely to be set..

The exceptions to this include:

- With a wider bandwidth, the probably of overload of the later stages 
of the receiver itself increases as most receivers have very poor 
"close-in" performance specs compared to wide-spaced: Bob Sherwood has 
some excellent documentation describing this issue.  Very strong signals 
getting into the latter stages (CW carrier from a PLC, perhaps) could 
push the limits of performance of some (particularly older) receivers.

- If there is a very strong signal within the passband, the AGC can 
desense the receiver, possibly putting the desired signal - if very weak 
- below the detection threshold of the data contained within the audio 
file that is generated during WSPR reception.  To be sure, amplitude 
disparities have to be quite severe as most people use 16 bit A/D 
converters and unless the input audio levels are set extraordinarily low 
(or the signal doing the desensing is extraordinarily strong) this 
should not be an issue.

There is one other issue that is more difficult to quantify - and that's 
the fact that the AGC's action will likely be keyed by energy high above 
that of the desired signal(s) - and varying.  On a passband with pure 
Gaussian noise in the background, this would likely result in pretty 
consistent results - but with a wider passband, there may be other, 
intermittent signals (not noise - which is likely to be consistent 
across any reasonable receiver passband, anyway) that may "key" the AGC 
in such a way that the results are skewed somewhat - but over the 2 
minute window - and certainly, over many receive periods, this effect is 
likely to average out - if it's even possible to discern it from normal 
level fluctuation changes in propagation, anyway.

Of course, the one thing that the WSPR program cannot possibly divine is 
"how noisy is your receive setup" set up as all it can do is compare the 
noise with the signal being received and there is no easy way of 
measuring your sites' performance compared to the natural background - 
nor is there an easy way to input that data, were it available.

73,

Clint, KA7OEI