I'm a little sleepy this morning but will try to answer this. If a CW (no modulation) signal just reaches the top of the display (on most models) you can consider that signal at the same level as the reference level. If the signal is modulated, especially wide digital signals like 8VSB digital TV or DVB satellite, you have to take the resolution BW of the analyzer into account.
The resolution BW on the analyzer is the IF filter and a CW signal will appear at the same level through all the filters where a wide signal will appear to reduce in level when a filter narrower than the signal is chosen. A simple way to get the correct level on a wide signal is to measure the BW of the signal at about 6dB down and I like to use a 1MHz resolution BW for this. Then take the log of the BW of the signal and multiply by 10 to get a correction factor for the wide signal. For example, a 24MHz wide signal would have a 13.8 dB correction that you would add to the measured signal (in 1MHz RBW) to get the actual level.
The leads into sensitivity and noise floor. The analyzer will have a specified sensitivity which will be measured with the narrowest resolution BW filter. One purpose of the RBW filter is to limit noise into the analyzer and the noise floor will change at a predictable rate, cut the filter BW in half and the noise floor will drop 3dB, cut the filter BW by 10 and the noise floor will drop 10dB until you reach the limits of the instrument. The trade off is the sweep speed will slow down considerably at narrow RBW settings, so your measurement parameters are dictated by what you want to measure, how much span you need and how much speed you need.
I should also mention that the narrower RBW filter settings allow you to see more details in the signal or to discern multiple close spaced signals. For the purpose of this discussion, a CW signal has no BW, its just an infinitesimally small slice of one frequency. Yet when you look at a CW signal on a spectrum analyzer it has BW and it gets wider with larger RBW settings. This is the analyzer tracing its internal RBW filter. If you are trying to see several narrow signals together like three 12.5KHz spaced UHF public service transmitters, a 1MHZ or larger RBW setting will show a single lump on the analyzer. As you narrow the RBW filter you will start to see the individual signals, especially with RBW settings like 1KHz (narrower than the transmitters) and the noise floor will be pushed down at the same time, exposing more detail and possibly more signals buried in the noise.
Some analyzers automatically add attenuation when changing the reference level. The RF section only has a certain dynamic range and changing attenuation is the way for the instrument to measure a very wide range of signals.
A 50/75ohm mismatch will incur a small loss, around .18dB if everything else is matched, but if there are other problems in the system like antenna to coax mismatch, the loss will go up. Some devices will show a frequency change if not terminated properly like sweeping a 75ohm filter with a 50ohm instrument.
The spectrum analyzer is your best friend in the RF business and the more you know how to drive it, the more places it will take you.
prcguy
For anyone that has had experiance with SA's, especially newer nodels, here are a few questions;
1. If I understand it correctly, the top of the dislay is the "reference level", correct? IOW's, if you set the level to say, 0dbmV (I use dbmV, not dbm), any peak that touches that top line would measure 0dbmV?
2. Is the sensistivity and the noise floor the same thing? Say the specs for the noise floor is -100dbm (a common spec), would that interpert into a sensistivity of -100dbm or is this different?
3. Where/how would a mismatch, say 50 to 75 ohms show up on a trace? Example, looking at the loss and trap depth through a series of filters that are 75 ohm using a tracking generator that outputs a 50 ohm signal into a 50 ohm input of the SA?
4. By increasing the reference level, does that attenuate the incoming signal as a attenuator would do?
