Control knobs with scale Lets take dials as a starting point, and assume some kind of pointer is necessary as a knob (relative) position feedback.
Per knob and general textual or pictorial labels A legenda to indicate what the knob (dial) is for, and preferably a numerical indication of its (absolute) value.
Programmable knob functionality and high adjusting accuracy Most synthesizer have too limited front plate area to give every parameter a private knob (although some do), and especially the complex ones simply have too many parameters. The 168 parameters on a DX7 for instance would need the space of a average size mixing console, not very practical, and absolutely not cheap.
The problem lieas in the combinatino of the desirable properties,
which has been solved mainly in two ways.
One way is to alleviate knobs altogether, simply use a keyboard ranging from a parameter per key, or simiple arrow menu controls, to indicate which parameter is currently editable, and use up down controls to change them. A valuable extension is a global dial (usually of the >360 degrees range, only relative position type) to improve the capabilities for large changes and intuitive control. Major disadvantage: no overview of parameter values, fuzzy absolute and relative feedback feeling.
The other way is to have potmeter based (270 degree, absolute position) control knobs, a (usually global) display to indicate the absolute value, and to indicate functionality. For a limited set of parameters assigned to each pot, leds and a fixed text can be used. The major and quite significant disadvantage of this method is that reassigning a knob to another parameter will make its position indicator present the wrong value. The only way to correct for this is to either make the parameter match the value, which would instantaneously change all reassigned parameter values, or to update the parameter value as soon as the dial is touched, or to make the dial "pick up" on the current value as soon as its postion matches the parameter value. The latter is probably the most desirable one, but has the major disadvantage of always changing the parameter value a little bit at least before it is edited, and it also leaves the dial positions completely not matching the parameter values after a functional change. The latter can be improved on by having a "lock" indicator per dial.
The main idea is to have completely free rotating dials,
with a programmable scale around it, and textual display
next to it. So far, nothing new: use either 160 degree pots or
variable capacitors (or even coils?), which can also be AD-ed
fairly easy), a lcd display underneath it, a circular array
of small leds in the circumference, some analog and digital
multiplexing circuitry, and off you go. A touch expensive though.
I was thinking of double using an optical system, composed of the following components:
a dynamically directable, intensity modulable coherent
(or focussed) light source
I'm not sure lasers, or preferably laser diodes (the pen type)
can be modulated in intensity at high rates, but considering
there are fibre feeding modules in the gigabit/sec range so it
should be quite feasible, and probably cheap. The beam defection
need not be to fast, and if multiple scans or too many directions
are hard to fabricate, multiple beams (using the same lense system)
could be used for character generation (e.g. 8). I know that
at least there are magnetically controlled x-y deflected
mirrors to do disco laser show projection, alternatively
a rotating drum with correctly placed mirroring surfaces could
be used, as was my original idea a decade ago, combined
with an apropriate set of mirroring surfaces.
A light distribution system Most easily by mirroring surfaces (metal damped on plastic moulds should be easy enough to engineer and fabricate). It could also include fibres or simply perspex like light guides. I thought about a global, mildly tilted backplane running under the whole frontplate, possibly as an airtight comoplete part, to prevent smoke and condens problems, steered from a light beam with milt x-y deviation hitting a parabolic or semi-spherical mirror above the highest frontplate display position.
A frontplate with translucent areas around and under the dials Preferably mildly difuse material, maybe simply copy paper under perspex.
A combination of a dial position feedback mechanism combined with a global optical sensor.
The idea is to read a dial position by scanning its circumference and detecting wether a (in rotary sense non-symmetrial) mirror or non-mirror (black) part is hit by the scan beam, detected by some global photo transistor(s), linked with the beam position.
The whole thing together would be a fairly easy (cheap in mass
production) front plate composed of cheap parts: each knob only
needs and axis (for different knob types snap on new knobs
and couple them magnetically with the bottom axis part), and
a very simple disc with some mirroring (or just non-black)
and black sections on it. Every display only needs a piece
of transparent frontplate, and each the dial idem.
And to make various blocks, just make the frontplate transparent
at the desired places to indicate their boundaries,
and as far as intensity and deflection and modulation speeds
allow, more extensive text and graphics can dynamically be
displayed everywhere, nicely readable in dark performance
The whole thing is completely programmable, needs no maintenance, can even be multi color, allows programmable knob-scale ratios (like a ten stroke pod),and has none of the above problems, and it needs a not very fast microprocessor/controler to do both display and position sensing. And it is a very general and completely extendable (possibly even unit-wise) design.
A bit of a cach lies in the intensity problem, the idea is to only make vector type graphics and characters, and of course the dial is just a circularly moving beam with correctly timed modulation (for either bar or spot dial postion feedback), to make sure intensity is sufficiently readable (depending on dot size, of course). Principally, a heavy laser or other modulable focussed light source could be used, but when a bear bottle breaks the front plate, it's not so nice to gaze in an open laser beam of significant wattage. Giving the beam a very short defocussing range outside the frontplate-plane could be a solution, but this would either require a dynamicaly focussable beam, orr equal running lenght all over its (warped) x-y range, or a focussing front plate.
Needless to say, it is not too much of a problem to either dynamically or design time align and linearize the optical system by using lookup tables in the controlling microprocessor or include mirroring gauging/trigger points for instance in a regular grid over the front plate.