[sdiy] Digital ADSR - perceivable staircase?

[David G. Dixon]

> > Talk about hammering a nail with a pile driver!
> 
> Says the man who uses a Rube Goldberg contraption for what could be more
> succinctly done with a multiplying DAC?  :-)

Perhaps.  But the 2164 makes a pretty nice exponentially variable resistor,
which is in fact exactly what you find at the heart of every analog envelope
generator (usually in the form of an audio-taper panel pot).

> There is no right or wrong solution to this problem and everybody gets
> there from a different background and settles in a different spot.  

Amen to that!

> To take some sting out of my remark, your solution probably is the right
> one if you want to modulate the same signal with both a CV and a digitally
> generated signal.

... or to retrofit digital control onto an existing (and very nice) analog
EG design, which is in fact exactly what I intend to do.

> It certainly isn't if you need more than about 10 bit of precision (and to
> get there with the 2164 will require some heavy lifting).

Again, you must remember that my (suggested, but not tested) method involves
using the control voltage to control the rate of change of the envelope (by
controlling the resistance), and not the actual voltage of the envelope.
Hence, the envelope will be piecewise-continuous (i.e., a series of
connected lines) rather than a staircase.  Even with 8 bits, this would
sound as smooth as silk.

> As for Scott's digital EG: it is essential that the computation is done in
> high precision if you want to have a large range of timing values or
> envelope rates.  About 1 bit per 4 dB timing resolution is a good initial
> guess, so the 28 bit Scott uses should be good for around a 400,000:1
> range.

The 2164 has 120dB of useful control.  That corresponds to a 1,000,000:1
range.  However, the real beauty of it is that this range is greatly
expanded at the fast end of the spectrum, where you want fine control over
the shape of the envelope.  (I personally almost never put the attack or
decay knobs on my ADSR beyond about 1/3, and I never (ever) use the "long"
setting of the range switch...)

> That is certainly nothing too exalted to ask from an EG.  How many bits
> you use at the target of that modulation is an entirely different
> question.  Same thing as with phase accumulators, really: you need more
> internal precision than you will use externally.

Yes, but I honestly think it's a moot point with a properly designed analog
circuit, even one with a digital interface.  But as you said before,
"different strokes..."