[sdiy] uLaw DACs in the modern era?

rburnett at richieburnett.co.uk rburnett at richieburnett.co.uk
Thu May 24 12:15:21 CEST 2018

Yes, they apply a gain that increases with time to the analogue sound 
when they recorded it to compensate for the natural decaying envelope of 
the cymbal sounds.  This approximately flattens out the natural 
amplitude variation over the duration of the sound.  The much reduced 
dynamic range then makes better use of the limited number of 
quantisation levels available at the time (6-bit.)  It essentially makes 
sure that even the otherwise quiet tail of the sound is still of 
sufficient amplitude to use up most of the quantisation levels and avoid 
suffering excessive distortion.  The original envelope is then 
re-applied by an analogue VCA when the sound is played back.

I suspect that the resulting dynamics of the sound are not exactly 
authentic to the original though.  The hi-hat samples in the TR-909 get 
exponential decaying envelopes applied to them upon playback, whereas 
the ride and crash cymbals get *linear* envelopes applied to them that 
are actually generated from the ROM address counter outputs!  Anyone 
who's ever used a 909 will recall how "compressed" the cymbal samples 
sound.  I think this is largely due to the replacement of the 
instruments' natural envelopes with linear ones!


On 2018-05-24 01:17, John Ames wrote:
> On 5/23/18, Adam Inglis <21pointy at tpg.com.au> wrote:
>> .. and in the case of the first Roland "all-digital” drum machines 
>> (TR-707
>> circa 1984 onward), wasn’t there a little analogue envelope shaping 
>> trick at
>> the end of the chain to really give them some extra whack?
> Mm, I don't know about the all-digital ones, but I know I've read that
> the 909 (which used sampled cymbal sounds) turned up the gain as it
> played through the samples so that the quiet tail-off had better
> fidelity with its limited sample width. Wouldn't surprise me if they
> used a similar trick on the digital ones until (presumably) they went
> full 16-bit and it wasn't so much of an issue.
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy

More information about the Synth-diy mailing list