zipper noise was: digital control

[Stewart Pye]

Hi All,

If you want 14 bits of resolution why not use pairs of midi controller
messages. Thi should work fine if you build the controller yourself. say
for control knob 1 use cc1 and 2. this will give 14 bits. Send both thes to
the sequencer and then to the synth. It will chew up twice the midi
bandwitdh, but not necessarily.....

If you use odd CC numbers for the LS 7bits and even CC numbers for the MS
7bits, send the most significant bits (even CC#) first. If the reciever
gets an odd CC number (LS) and a flag has not been set saying it got the
corresponding even number(MS), then use the last stored MS 7 bits and the
new LS 7 bits to update the DAC. Next time you get a new even number(MS),
store the value and set the flag and when the corresponding LS 7 bits come
along the DAC will be updated appropriately. 

This way you only have to send the MS 7 bits once in 128 steps. 


I hope all that made sense. It'd be a bit easier if we were talking bytes,
but they're not really.


Stewart 


At 11:38 AM 12/01/2000 -0800, Brousseau, Paul E (Paul) wrote:
>A lot of people have brought up glide curcuits, etc., but the high bitcount
>DAC seems glanced over.  What are the pros and cons of that?  The only
>immediate con I can imagine is that anything beyonf 7 (8?) bits is out of
>MIDI resolution.  I would propose that solved by using two MIDI signals, a
>"course" resolution and a "fine" resolution.  What am I missing?
>
>(After a moments thought...)  Using two MIDI signals would make things
>difficult for external control, I suppose.  It would be nearly impossible to
>spin two knobs to achieve a continuous sweep through a broad band.  Spinning
>just the fine control knob wouldn't do very much, and spinning just the
>course control knob would just introduce the zipper effect again.  The
>usefulness of two controls, then, would be limited to patch storage.  The
>module could be designed to use one knob locally.
>
>A relevant question, then, is "Is 14 (16?) bits enough?"  Also, how fast
>would one need to clock?
>
>--PBr