>Try March and April 74 Sky and Telescope. I believe they used photoelectric
>tubes, but photodiodes could easily be substituted. Photodiodes have always
>impressed me as being relatively insensitive.
Of course CCDs are just collections of photodiodes and they get by. Photodiodes have really good quantum efficiencies, about 40%--better than tubes--but require external amplification. The trick to getting high enough output from them is to use them with an op amp that has an input bias current specification in the low picoamp range (this spec means the input transistors don't need much input current to operate) and, as with a CCD, to integrate.
You can wire opposing photodiodes--that is side-by-side diodes with a star image split between them, the configuration used for guiding--into the negative and positive inputs of a CMOS op amp configured as an current integrator (no input resistors, capacitor between the output and the negative input). When more light falls on one photodiode the capacitor charges; when more light falls on the other side the capacitor discharges. The integration averages out the noise and adds sensitivity. The gain is a factor the capacitor value, and you can put a voltage divider between the output and capacitor to vary the gain. An amplifier stage powers a DC motor. There's a bit of work in the details, like shielding the inputs from stray currents, etc. A decent library will have books with representative op amp circuits.
Four-quadrant photodiodes made specifically for optical guiding are available. Although they're used to center laser beams etc, they have enough sensitivity to use with starlight. I suppose an electronics supplier would help you find one.
Aart M. Olsen aart@uiuc.edu 217-333-7467 College of Veterinary Medicine Univ of Illinois at Urbana-Champaign