This is an input device controlled by a human being... For example, a DJ scratches a turntable controller to make certain genres of music. The wheel being turned is quite large so even very minute gestures can potentially be meaningful.
That said, as I think about it, my original goals may in fact be overly ambitious and impractical. For example, even with a 12" vinyl record, the circumference is about 1m, and 1/60 of a degree is only about 0.3mm of linear motion at the circumference. I have to do some research on what kind of precision does the musician expect of this, but they may be happy enough with 1/10 of a degree.
I am trying to design a control knob. The requirement is to give its angular position over e.g. I2C, with a precision of about 1/60 of a degree (something like a quadrature encoder with 1000 pulses/revolution counted at 4x). The knob is controlled by a human being and there is only one knob so speed is not a problem for any reasonable uC. (One example use of this is e.g. a MIDI pitch wheel, or making a DJ turntable scratch MIDI controller.)
This may become a product so I would not be able to use recycled parts, such as from an old-style mechanical mouse.
I searched and found products above USD100 advertised as "low cost", and wonder if this is just the way it is?
The mechanical encoders usually have only about 24 pulses per turn and I probably won't be able to gear it enough to satisfy my resolution requirement.
I also looked at magnetic sensors, they look cool because they can detect the orientation of a magnet from a distance. Resolution seems to be dependent upon A/D precision but I am not sure about the repeatability and stability for use as a human interaction device.
There seems to be optical encoder modules with etched metal wheels pre-assembled in a housing. Some of these can achieve 1000 pulses per revolution without a problem. They seemed to be geared towards motor control and they are expensive. Is there an inexpensive source?