Hello Mike,
Am 29.09.2017 14:58, schrieb Mike Dennison:
I was using an "equivalent noise bandwidth" of 44uHz which meant
around 9 hours per pixel. If you look at the right hand side of the
picture the difference between the long term average on your
frequency (red line) and the average noise seems to be about 5dB
(10dB per div).
I have added quite a bit of hard limiting in SpecLab's Blackbox 1 and
this tends to smooth out the nighttime noise peak.
Also, I have increased the "internal average" in the FFT menu to '2',
which reduces the displayed noise but makes the start/stop edges
rather fuzzy.
Oh, did you switch a band pass filter in front of the noise blanker or
hard limiter? This is important.
I would recommend to send me a .usr file with your current settings
(Save settings as... then save as a usr file). I will have a look,
eventually do some dodifications and send it back to you.
With a more or less modern PC/notebook you can also run 2 instances over
the weekend and compare which one works better.
I will try two spectrograms later as you advise.
Just double-click a second time on the SpecLab icon to start a second
instance.
A question about SpecLab: When setting the bandwidth in the FFT menu,
what is the difference between the "Decimate.." and the "FFT input
size" settings. They both seem to produce the same result, eg
Decimate at 2048 and FFT at 262144 gives a bin width of 44uHz, and
Decimate 4096 and FFT 131072 gives the same. Why should I set one
high and the other low?
The higher the input size, the higher the frequency range that is
covered and can be analysed later, and the higher the CPU load will be!
Thus it is better to select a small input size and a high decimation
factor to obtain a lower FFT bin width.
I usually choose an input size of 8192 but one could even go lower. In
the information text field you can see which frequency region is covered.
73, Stefan
|
