Stefan wrote:

> www.iup.uni-heidelberg.de/schaefer_vlf/VLF/test.vt

Some good events on ch5 there, eg

 vtcat -S4 -E4 test.vt:1,2,3,4,5,6 | vtsgram -p100 | display

 http://abelian.org/vlf/tmp/170709a.gif

Now to measure the noise floor.  We can use vtstat -a option
for this.   It takes a time resolution parameter in seconds.

 vtcat -S4 -E4 test.vt:5 | vtstat -a r=0.01 > temp.1

temp.1 has 5 columns,

 col1 is the timestamp
 col3 is the mean amplitude (the 'DC' level)
 col4 is the mean squared amplitude
 col5 is the mean cubed amplitude

Using gnuplot commands

 set terminal x11
 set style data lines
 set grid
 plot 'temp.1' using 4

gives

 http://abelian.org/vlf/tmp/170709b.gif

where the x axis the the number of 0.01 second samples.

Using the mouse cursor in the gnuplot window you can read
off the mean square amplitudes.   Looks like it's normally
about 1.16e-6 give or take a bit, drops to about 3.7e-7
during the dropouts.

Your Octo, averaged over the 30 second test.vt
using

   vtcat -S5 test.vt:1,2,3,4,5,6 | vtstat -a r=25

(I skip the first 5 secs because of a start-up transient)

 ch1: 2.4e-6
 ch2: 3.3e-6
 ch3: 4.2e-6
 ch4: 1.4e-6
 ch5: 9.1e-7
 ch6: 1.8e-6

The low average on ch5 might be because it is dropping
out a lot!

The corresponding amplitudes from my Octo are:

 ch1: 3.9e-6
 ch2: 5.5e-6
 ch3: 3.5e-6
 ch4: 2.1e-6
 ch5: 3.3e-6
 ch6: 3.1e-6

Conclusion: away from the dropouts, the noise floor is
comparable with my Octo.  Noise dropouts are 3.7e-7/1.16e-6
so about -5dB.   But the signal not affected, or slightly
increases!

--
Paul Nicholson
--