[time-nuts] Noise source measurement
jherrero at hvsistemas.es
Sat Oct 22 15:17:28 EDT 2011
El 22/10/2011 15:13, Jim Lux escribió:
> Two things spring to mind:
> Spectrum analyzers don't have very low noise inputs, in general. Did
> you account for the additional noise from that source?
> The noise bandwidth is not necessarily the same as the resolution
> bandwidth (the shape of the filter isn't rectangular, after all).
> You'd have to look at the phase noise measuring ap notes from HP, I
> think they give the correction factor to use (it's more than a dB, as
> I recall)
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to
> and follow the instructions there.
Thanks, Jim. Yes, I've accounted it, the noise source is around 20dB
over the analyzer noise, so analyzer noise can be neglected. I was
missing the correction due to the filter form factor, and thanks,
Sylvain, for pointing me to the application note, that shows this
correction and other more important (in dB terms) due to the log
detection and averaging process of the video filter, so I must add
2.51dB, and substract 0.51dB due to the filter shape, so I must correct
add 2dB. I see that the shift-M function already provides this
correction (I was missing that! Thanks!)
I've make sure that no analyzer compression or overloading is taking
place. Also, the bandwidth of the noise source is limited to approx.
1.2GHz, centered at 8.2GHz, so the analyzer does not see too much noise
power (around -15dBm total noise power). I've checked changing the
attenuation values from 0, 10 and 20dB and the analyzer noise floor
varies as expected, but the noise source measurement does not change.
I was missing also to add a 1.2dB for the analyzer cable loss, and after
playing a bit around, the value I get at the center frequency is
-102dBm/Hz, around 8dB higher that I was expecting, but probably the
measurement is ok and the fixed losses due to the other elements were
somewhat overestimated during the design, and the gain of the amplifier
that is in the noise path is a bit higher than expected. I will measure
directly at the noise source output, and check losses and amplifier
gain, but to have a bit more of noise available is not a problem in this
case (in fact, it is an advantage). Mainly I was looking to be sure that
I was measuring it in the right way, so thank you very much for your help.
More information about the time-nuts