More Information - Keysight x-series User Reference Manual

5 Mode Functions
Mode Setup
Couplings
Preset
State Saved
Initial S/W Revision

More Information

The analyzer is characterized in the factory (or during a field calibration) with a model of the noise, referred
to the input mixer, versus frequency in each band and path combination. Bands are 0 (low band) and 1
through 4 (high band) in a 26.5 GHz instrument, for example. Paths include normal paths, preamp paths,
the electronic attenuator, etc.
In most band/path combinations, the noise can be well characterized based on just two parameters and
the analyzer frequency response before compensation for frequency-dependent losses.
After the noise density at the input mixer is estimated, the effects of the input attenuator, RBW, detector,
etc. are computed to get the estimated input-port-referred noise level.
In the simplest case, the measured power (signal plus analyzer noise) in each display point (bucket) is
compensated by subtracting the estimated noise power, leaving just the signal power. This is the operation
when the detector is Average and the Average Type is set to Power.
In other cases, operation is often not quite as good but still highly effective. With peak detection, the noise
floor is estimated based on the RBW and the duration of the bucket using the same equations used in the
noise marker function. The voltage of the noise is subtracted from the voltage of the observed signal-plus-
noise measurement to compute the estimated signal voltage. The peak detector is one example of
processing that varies with detector to give good estimates of the signal level without the analyzer noise.
For best operation, the average detector and the power scale are recommended, as already stated. Peak
detection for pulsed-RF can still give excellent effectiveness. FFT analysis does not work well, and does
not do NFE well, with pulsed-RF signals, so this combination is not recommended. Negative peak detection
is not very useful, either. Sample detection works well, but is never better than the average detector
because it doesn't smooth as well. The Normal detector is a combination of peak and negative peak
behaviors, and works about as well as these.
For best operation, extreme smoothing is desirable, as already stated. Using narrow VBWs works well, but
using very long bucket durations and the average detector works best. Reducing the number of trace
points will make the buckets longer.
For best operation, the power scale (Average Type = Power) is optimum. When making CW measurements
in the presence of noise without NFE, averaging on the decibel scale has the advantage of reducing the
effect of noise. When using NFE, the NFE does an even better job than using the log scale ever could. Using
NFE with the log scale is not synergistic, though; NFE with the power scale works a little better than NFE
with log averaging type.
The results from NFE with internal preamp can often be lower than the theoretical noise in a signal source
at room temperature, a noise density of –174 dBm/Hz. This is expected and useful behavior, because NFE
258
When NFE is enabled in any mode manually, a prompt will be displayed reminding you to perform the
Characterize Noise Floor operation if it is needed. If NFE is enabled through SCPI and a Characterize
Noise Floor operation is needed, an error will be entered in the system error queue.
Unaffected by Mode Preset. Turned off by Restore Mode Defaults.
No
A.04.00
EMI Receiver Mode Reference