Simulate:
Note that at 160um turn-by-turn resolution, a 1024-turn average orbit measurement will yield 160/sqrt(1024) ~ 5um resolution, or what we've measured our repeatability to be.
Plot button signals b(2,3,4) against b1.
Amplitude of oscillations over these turns:
Odds are our actual resonant excitation implementation produces a much smaller amplitude than this; in theory the nonlinearities in button signals should be of less concern.
Amplitude of oscillations is similar.
Looking at AW's 'ping' data (+/- 2mm vertical amplitude), the coherence of actual TBT data is MUCH better than the 160-micron resolution scenario simulated above. This implies that the 1024-turn average orbit resolution is limited by something other than actual BPM resolution. Possible candidates: