[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[linrad] dBc/Hz

Hi All,

Is there anyone who knows the correction factor????

When we measure noise levels on a spectrum analyser we use
a logarithmic detector. By means of a video filter we can average
the detected voltage to get smooth curves. It may be a standard
procedure to take the reading for a carrier and the sideband noise 
from the spectrum analyser screen, correct for the bandwidth and
present the result as dBc/Hz but it is not correct. The definition
of dB says power ratio and the result obtained from the averaged 
logarithmic detector output is not power.

I can find empirically that the result obtained from a logarithmic
average is about 6 dB lower (better) than the true result from a
RMS detector. Obviously it is "simple" to deduce the difference 
between the RMS and the log detector results under the assumption
of white noise (within the passband). Unfortunately I am not
clever enough to do the mathematics myself but I think the result 
should be well known.

Is there anyone who knows how many dB (with decimals) one has
to add to correct the result from an oldfashioned spectrum analyser
to get the true dBc/Hz value for the sideband noise? 


Leif / SM5BSZ