Thanks to everyone for their various ideas in support of this project, and of course a big welcome to Darryl too.
A few points:
I would have thought it virtually certain there will be some accumulated error. Darryl's allowance of 9 minutes of arc seems generous to me. I had been working on the basis of 1½ minutes (1 pixel of error). This equates to an error of just 0.05mm at the locating pins and this is 'pushing it', because it's about half of what I understand to be the working tolerance on the laser cutting. Moreover, (unless I can correct for this by adjusting the CAD files after seeing the test pieces) there may also be errors in the diameters of the locating holes caused by other laser cutting effects.
So getting this mirror screw to perform well in practice might involve some 'rough techniques'. In particular, at some point I can see myself having to judiciously 'tap' each of the individual slats back or forth by tiny amounts against their stops to achieve a smooth rendering of a test waveform. A laborious process - but not impossible. Even worse, tapping one could then affect its neighbours. But maybe this effect would prove useful when correcting for cumulative error. There will doubtless be problems, and perfection will be hard or impossible to achieve. But I won't know until I try!
Darryl's idea of allowing the user to shift the position of each line individually to correct for mirror position errors would be a wonderful feature to have, if this were indeed possible. Some raggedness at the edge of the raster would be a price well worth paying in my opinion. And a video-to-motor synchronising feature would be luxury indeed.
As for synchronising this heavy beast, apart from a stout motor I had been assuming using the 'club' circuit, a sync wheel somewhere on the shaft with 120 black and white marks and a opto pickup of some sort. Experience at 30-lines has shown me that if a few syncs (about 10%) are removed every frame, this circuit will not only lock to line but will automatically handle frame phasing too. Since (as I understand it) there are no integral syncs in Darryl's mechanical waveforms, I envisage a separate sync line with a bunch of the pulses missing every frame, just as I used in the Grosvenor and its mechanical camera.
Whatever source(s) are eventually used, I think we should base the signal spec on the assumption that Darryl's World Converter will be the 'fallback' standard, and any LED driving arrangements we incorporate should be based on the feeds available from his converter. Keeps things simpler!
One thing that is already coming across strongly is how much an ambitious project like this relies on the help of other members here, who are each contributing their own unique specialisms. Thank you all!
Steve O