Afonso is correct.  Our frequency measurement sits on top of an edge extractor that runs an edge extraction pass on input waveforms at the mref value.  (It does not look at HREF or LREF and while it might give you warnings, it will even attempt to report frequencies if you drive the signal off screen both high and low).<br>
<br>
The edge exctractor requires the input signal transition completely through the hystersis band (top to bottom or bottom to top) and will then do a siinc-interp location back to the appropriate location of the edge crossing, potentially between samples.<br>
<br>
The resulting edge vector is processed to report every frequency seen on the input vector.    (just running through the edge list).   (Those generate 'current acq' stats).   multi waveform passes then generate 'multi acq measurement stats').. using every edge we've seen.<br>
<br>
Note that waveforms with large idle periods may cause confusing results depending on the nature of the crossing...   (e.g. a signal might be 0 level for a long period, but the edge extractor will still put an edge in the middle provided it passed through the hysteresis band).<br>
<br>
If you want to avoid potential confusion for such waveforms, it can be worth using gating to eliminate said areas.   (cursor gating, logic gating the measurement would both work)<br>