search

lspestrip / striptease

Strip TEst Analysis for System Evaluation
MIT License
4 stars 3 forks source link

Missing factor 1/2 and 1/sqrt(2) ? #32

Open fincardona opened 4 years ago

fincardona commented 4 years ago

Hi,

Is there a missing factor 1/2 in the computing of WNL in the following lines of code:

https://github.com/lspestrip/striptease/blob/bccffbed11e9f023c69be54758ef86b1c3f8ef47/striptease/Spectrum%20tools%20help.Rmd#L102-L103

https://github.com/lspestrip/striptease/blob/bccffbed11e9f023c69be54758ef86b1c3f8ef47/striptease/Spectrum%20tools%20help.Rmd#L153-L154

and a factor 1/sqrt(2) in the following ones:

https://github.com/lspestrip/striptease/blob/bccffbed11e9f023c69be54758ef86b1c3f8ef47/striptease/Spectrum%20tools%20help.Rmd#L82-L83

https://github.com/lspestrip/striptease/blob/bccffbed11e9f023c69be54758ef86b1c3f8ef47/striptease/Spectrum%20tools%20help.Rmd#L133-L134

?

Furthermore, it is not to clear to me which are the real dimensions of WNL. Are they: amplitude / sqrt(freq) or power / freq ? I think one should decide to refer either to the first as WNL and to the second as WNL squared or to the first as sqrt of WNL and to the second as WNL.

Thanks

fincardona commented 4 years ago

I found also that you plot 2*noise_properties[0]: https://github.com/lspestrip/striptease/blob/bccffbed11e9f023c69be54758ef86b1c3f8ef47/striptease/Spectrum%20tools%20help.Rmd#L229 https://github.com/lspestrip/striptease/blob/bccffbed11e9f023c69be54758ef86b1c3f8ef47/striptease/Spectrum%20tools%20help.Rmd#L266 https://github.com/lspestrip/striptease/blob/bccffbed11e9f023c69be54758ef86b1c3f8ef47/striptease/Spectrum%20tools%20help.Rmd#L282 which compensates for the missing factor 1/2 in the computation of White noise level from spectrum but, then, what do you have on the y-axis of the spectrum?