[sdiy] The Anything Box: Laptop as module?

[Grant Richter]

> I'm curious - are there issues with trying to quantize
> pots beyond 7 bits (maybe caused by jitter, etc)?
> Midi pitch bend is 14-bit, and most of my controllers
> *seem* to use all 14, although the Nord Lead only uses
> the MSB.
> 
> Also, for midi to CV - use of 7-bit midi CCs is really
> an issue if they directly control a CV; modulation of
> less than the full range is less noticeable, and
> software smoothing can be applied to help also.

If you consider the idea of encoding a 14 bit continuous controller, then
transmitting only the 7 MSB, then extracting all 14 bits. This is like data
compression over a network. You could look at the time interval between CC
packets as encoding the LSBs as a line segment.

This was the idea I proposed to Bob Moog for the Voyager instrument. I don't
know if he included in the instrument or not. He probably thought of it
himself, before I mentioned it.

If you could add a physically modeled component to the decoding. For
instance, momentum, the data is coming in at 7 bits, your outputting it at
14 bit and using computation to draw the "fine" straight line sections
between the 7 bit "coarse" points. Calculate the connecting line slopes by
the rate of receipt of the 7 bit CC packets.

When you don't get a packet when expected, you continue the line segment,
but coast the slope rate down to zero. Glide to a static value until the
next data point shows up.

A simple slew limiter will kind of do it, but by the time there is enough
smoothing, the phase lag becomes noticeable. The slew rate needs to be
modulated by the receipt rate of the CC packets. In hardware, this starts to
be come large, so a hybrid approach may be best?

Any of you fine PIC fellows want to make a MIDI to CV with extrapolated high
resolution CCs? There should be some marketplace for it.