151109-2356 EST
Common Sense:
Your down spikes in voltage are likely motor starting, or transformer inrush, or some other transient high current load. Unless there is oscillation within the spikes, then these would not account for your flicker. The recording instrument response time may not be fast enough to see your 3 to 4 Hz flicker.
For comparison see some of my plots of line voltage and short time average power vs time. Visit
http://beta-a2.com/EE-photos.html . My time resolution is 1 second. This means an averaging time of 1 second.
P27 --- The power spikes are not noticeable in the voltage waveform. I had a pretty stiff source to the main panel. The power spikes are from motors starting and their peak values are lower than the real peak because of the 1 second averaging of the power instrument (a TED 1000).
P29 --- Measurement at the outlet to a freezer. This circuit consisted of about 85 ft of #6 to a subpanel, 40 ft of #12 to an outlet, and 12 ft to an outlet strip ( the measurement point ). Only the freezer was monitored. The large voltage spikes are from the circuit impedance. These voltage spikes are very small back at the main panel. The spikes have no pulsed oscillation.
Can you measure two voltages simultaneously with your Fluke recorder? If so then record the input and output voltages of one of the problem dimmers when the flicker occurs, and with a time resolution of 16 milliseconds synchronized with the line if possible.
.