I also worked on a better mathematical analysis of the work done by drag, as shown below. (A stands for cross-sectional area and a will be amplitude) The size of the drag force varies with displacement from equilibrium because the speed varies with displacement. In my last post, I used an average drag force to get a ball park figure.
Here I've tried integrating between amplitude and -amplitude, half a time period.
So to reduce the work done by drag, amplitude is the key factor. A small amplitude is better. But I need an amplitude big enough to be able to tell that the direction has changed. Longer will also be better. That might just be possible if I can attach to the top rafter.