Steps in the Noise Floor
The noise floor will rise if signals are higher then the possible dynamic range.
In that case the SPECTRAN does not rise the complete noise floor. Instead it uses some tricky design so that only some parts of the noise floor needs to rise. This might look like steps but is no bug, its a feature
This would be nice if the step is stable. But you can see in the picture, between 2050MHz and 2125MHz, that the level of MAX is higher than the current level. The level is changing. If you look at the beginning of the spectrum at 1900MHz there is 10dB different between MAX and current level and there is now steps visible. What is the reason for that?
Is it possible to store a measurement and substrate it from the current measurement? If the noise floor stable finally we wouldn't have any more the steps.
Thank you for the info, but I have a comprehension problem.
In the picture between 2125MHz and 2170MHz the highest level is -50dBm and this is only 30dB higher as the noise floor. With a log-amplifier 90dB should be possible. But I don't know about your circuit.
Why is the step at lower frequencies where is no signal?
For me it is a problem if I measure a longer time and want to know was there any radiation. I had in mind to use the MAX hold, but if the noise floor also moving outside the frequency range where a bigger signal is, the small signals will be overridden.
Could you suggest a set-up.