[sdiy] samp. freq for ADSR ?

[Oren Leavitt]

One other little DAC resolution caveat:
Quantization granularity may be especially noticable on VCAs (or anything) that have an exponential response as the envelope approaches the maximum level the input accepts.

Oren

-----Original Message-----
From: Richard Wentk <richard at skydancer.com>
Sent: Apr 2, 2004 7:35 AM
To: synth-diy at dropmix.xs4all.nl
Subject: Re: [sdiy] samp. freq for ADSR ?

At 17:09 02/04/2004 +0200, jbv wrote:
>Hi all,
>
>OK, 48 KHz is a theorically fine sampling freq for audio signals.
>But right now I'm brainstorming on a project in which multiple
>ADSR signals would be generated by a uC / DAC / S&H.
>What is the optimal freq at which each signal should be refreshed /
>incremented ?
>At first glance 48 KHz seems a luxury, but what about very fast
>AD ?

The shortest useful time for an envelope attack is 1ms. You can go for 
100us if you want to get really obsessive, but 1ms will do for most things.

Sampling at 4KHz would give you 4 samples for the attack phase. That 
doesn't like a lot, but it's likely to be enough.

4KHz would also give you LFOs up to 2KHz, which should be enough for most 
things.

More relevant might be DAC resolution. With 8 bits you'll effectively be 
overlaying some hints of quantisation noise onto filter and envelope output 
- not a lot, but I think it will affect the sonic character of the output, 
and will make it dirtier than it would be otherwise. If you're planning to 
do accurate pitch envelope modulation of oscillators 8 bits definitely 
isn't enough. 12 would be about right, and 16 would be ideal for a top 
quality design.

Richard