[Knightsqrss] WSJT SNR figures

[Johan H. Bodin]

Hi Johan,

Joe has choosen to "normalize" the measured SNR to a 2500Hz BW but the 
WSPR decoder will see a much better SNR due to its narrow internal 
filtering. If the equivalent noise BW of WSPR is 6Hz, then the SNR 
improvement (power-wise) will be 2500/6 = 417 times = 26dB compared to a 
2500Hz BW. In other words, when WSPR reports -26dB SNR, it actually 
enjoys 0dB SNR, or so, in its decoder.

72
Johan SM6LKM

P.S.
I have added random timeslot TX with adjustable probability to the PIC 
based MEPT_JT beacon. A hardware sync input has also been added for 
those who don't want to use a GPS. I will post the design as soon as I 
have made some more tests...

/JHB


Johan_Smet at telenet.be wrote:
 > Hi All,
 >
 > I'd like to better understand the SNR values reported by WSPR. 
Apparently,
 > changing the RX IF-bandwidth (from 2.7 kHz to 500 Hz for instance) 
does not
 > affect the reported values. Why is that? Maybe because the WSPR noise 
power
 > sample is always recalculated into a 2.5 kHz bandwidth? Anyone for a 
sound
 > explanation?
 >
 > 73,
 >
 > Johan on5ex