John P-G wrote:
Graham is often difficult here too though, and I think this is related
to a soundcard problem at his end. The fact that you're nearer will
probably just exacerbate the effect.
In general however it does appear that the mode is very good on weak
signals and hit and miss on strong ones and one suspects that it's
very intolerant of errors in soundcard sample rate - which is masked
with weak signals to some extent.
Cheers,
John
Hello LF Group,
just to add more points to the ongoing disscussion about wspr
problem: to decode or not to decode ... :)
in my view John, GM4SLV is right when indicating that the soundcard
sample rate may be responsible for not decoding at all or, in part,
for errors.
some time ago i've posted to this reflector the following observation:
the wspr is using bin size of FFT 8192 and transmits 162 data bits
at 12000 Hz sampling rate so the wspr transmission lasts
162*8192/12000 = 110.59 seconds i.e less than 2 minutes. if the
11025 sampling rate was used than we would have 162*8192/11025 = 120.37
seconds - just over 2 min.
most of the soundcards have, say, master sapmling rate of 44.1 khz
therefore when downconverting this rate to the used sampling rate of
12000 Hz
some non-integer "offset" appears and how the drivers/chips deal with
this issue is an open problem
let me now add few other observations.. when i used wspr for the first
time ( on HF bands) soon i got interested in its performance in the
case of a strong signals so i have asked another ham ( we live in the
same city abt 3 km frm each other) to TX strong signal - indeed it
was red on the display and there was no problem
to properly decode his transmission. in the next 'experiment' i asked
him to do the same but at the same time i was trying to decode the
signal from distant transmitting stations ( i do not remember exactly
but i think it was a 7 Mhz band ) again, i was properly decoding weak
stations in a presence of a "red" strip on the waterfall. In some way ,
the last issue was also confirmed yesterday when i successfuly decoded
Jim , M0BMU and Gus, SM6BHZ: (frm MEPT.txt file)
in the same time slot:
090903 2242 22 -9 0.6 0.503956 SM6BHZ JO57 30 0 1 0
090903 2242 12 -19 0.4 0.503981 M0BMU IO91 33 0 1 0
Ok, it is only 10db difference but sure it is possible .
It seems that Jim, M0BMU has made a point when indicating the
narrow-band qrm's. if it is close to the 'signal' there is a false
decode or no decode at all.
here i have terrible and floating qrm on 500 kHz when it gets closer
to the signal wspr fails.
As for the PC clock synchronization..- again, i have looked into
all_mept file and i can see that DT can be as much as 3sec and still
yield a positive decode
090328 2244 12 -18 3.4 0.505094 DI2AG JN59 27 0 232
0 (DT=3.4sec)
yours, peter, sq7mpj
qth: Lodz /jo91rs/
|
