The CBA waged total war (130-day campaigns) on AM CVn through 1997 and 1998. I found the results very elegant, and here I'd like to share them.

Approximately 20 periodic signals are present in the star. But all are linear combinations of three basic frequencies:

w = 83.987 c/d
W = 1.795 c/d
N = 1.437 c/d

It's impossible to show the details briefly, but these figures give a whiff; they're power spectra of the most densely sampled part of each seasonal light curve. Some light spectral "cleaning" has been applied (removal of aliases of the stronger signals). Features are labelled with their frequencies in c/d, with the strongest signals cut off because they rise to power levels of 200-500. The weakest features labelled have a semi-amplitude of 0.0013 mag, and a power of 70 corresponds approximately to a semi-amplitude of 0.006 mag. Amazing what a networked telescope can do.

The orbital frequency w is easily seen, and 2w is also (very weakly) present. W and N are not seen, but are only manifest through sidebands of w and its harmonics. And there are intricate "selection rules" concerning which sidebands exist. The lower sidebands are all in the form nw - kW, where k=1,...,n; and within measurement limits, all such signals are present (no missing sidebands).

There are only a few upper sidebands, and they are all displaced by +N from another allowed signal (never 2N, 3N, etc., unlike the lower guys). There are many missing sidebands. For some reason, the star only likes w+N, 2w-W+N, and 3w-W+N.

Notice that the two years were quite different. The "nodal superhump" was absent in 1997 (no +N signals seen anywhere), but very strong in 1998.

The orbital signal itself is phase-stable over 1992-99. The exact period is 1028.7322(2) seconds.

I find it awesome to contemplate that all this comes from a star whose most powerful signal, the far-off-scale feature at 2w-2W, has a semiamplitude of 0.012 mag. What might we do with a very long campaign on a superhumping star with a truly powerful signal?