Dear Stefan, LF Group,
----- Original Message -----
From: "Stefan Schäfer" <[email protected]>
...> 0 dB above average is clipped, does this mean, that each sample that is
above 0 dB is just left away out of the whole integration that gives one
pixel in the spectrogram? ...>
I imagine Spectrum Lab must be doing the clipping in the time domain, i.e.
limiting the maximum amplitude of any particular sample that is above the
average level - perhaps Wolf could describe how this function works?
...>This clipping is done by SpecLab but you also implemented a narrow band
filter in your preamp, that is switchable. You said you want to show a
picture with the filter switched on and off (and display 0...20kHz). That
would really interst me! ...>
Well, the top image in the attachment shows a wide-band spectrum with and
without filter. Basically, at this scale, everything around 9kHz and below
is QRM...
...>Have you used compact inductors in a resistor housing or real wound pot
cores for high Q?...>
Small inductors on ferrite bobbins with radial leads were used - the 4.7mH
inductors had a Q only about 10 - 15. The 100mH was a bit bigger, encased in
a ferrite sleeve and had a Q of about 60, but lower Q would also have been
adequate - with 2.7kHz bandwidth, 9kHz centre frequency, very high Q is not
needed, although the resistor-sized inductors are a bit too small. The
self-screening of pot cores or toroids might be an advantage.
...>And what is the increase in the SNR @ 8,97kHz? ...>
The lower image shows the difference between a narrower-band spectrum with
clipping in use, and with or without the filter. The wideband noise floor
in-between the peaks is about 2 - 3dB less with the filter in use. Note this
is with a lower FFT resolution (about 0.3Hz) than used in the DK7FC tests -
because the noise spectrum changes quite quickly at my home QTH, it is
difficult to do these tests with 5mHz resolution. So you have to imagine the
noise peaks are "signals". The Spectrum Lab "hard limiter" does a
suprisingly good job even without the filter. From my earlier mail, the SNR
improvement with clipping and filtering was about 10dB. This test would
suggest that without the filter, using the SpecLab clipper alone would still
produce 7-8dB improvement. But if you live close to a VLF utility, the
advantage of clipping without the filter would be reduced, because then
SpecLab would have to set a higher clipping level, allowing more QRN energy
through the clipper. Also, the improvement obtained probably depends on the
level of QRN present.
Another reason for adding the filter was to enable use for audible reception
of 9kHz. I have added a simple 7.5kHz BFO, product detector and audio stage
to produce a complete SSB RX. I am not really expecting to hear signals, but
now I can listen to the various strange noises and use the system as a
portable D/F receiver to identify noise sources.
Cheers, Jim Moritz
73 de M0BMU
spectrum_3k_wideband.jpg
Description: JPEG image
|
