40
Appendix V: Preserve Noise Feature
Figure 22: Effect of "Preserve Noise" feature
In order to accurately represent noise in a DUT waveform, a sufficient number of samples
must be acquired per symbol. In this case, when higher resolution is attained, noise can be
more accurately re-injected into the DUT waveform. Thus, the preserve noise feature in the
Low-Pass filter in the ML DSO compensates for the filter function
’s side effect of noise
reduction for adjacent measurement points. Note that an incorrect implementation of this
feature can negatively affect measurement accuracy. Some examples will follow:
▪
This feature is advantageous during mask testing, as the mask hits could be too low
(optimistic result) without noise preservation
▪
The accuracy of TDEC and TDECQ measurements is directly related to the existent
random jitter and noise in the DUT waveform. If noise is changed or compensated due
to the usage of a low-pass filter like Bessel-Thompson, then the measurement will
naturally be inaccurate.
▪
Setting “preserve noise” to 0% yields a purely deterministic signal, while 100% yields
the unfiltered signal with the complete noise profile.
