[sdiy] Oberheim Xpander Envelopes and LFOs etc
sounddoctorin at imt.net
Sun Sep 30 23:32:41 CEST 2007
It depends on a lot of things as to how low one can go with modulating
voltage updates. One way of course to simply/cheaply get away with
lower update rates is to switch in capacitors of various values to
correlate roughly to the rate of slew being approximated. For limited
ranges one capacitor is sufficient but if one wishes to approximate a
WIDE variety of envelope rates for instance, the smoothing filters
should be switched by s/w in accordance to the rate being demanded. I'm
not sure if they did that on the matrix or waldorf but it's one way of
achieving perceived adequacy for sure at low update rates. Still a
dynamic staleness will result in rapid transients unless this scheme is
taken to an extreme of course.
And often the real trouble is sweeping at high resonance rates also.
Most filters enter self-resonance fairly abruptly. If a synth is set up
more like the Crumar trilogy and some others' CEM3320's, then obviously
it will reduce the brackish transition state that smacks of digital
control in most older digital envelope analogs (eg. sequential's Prophet
600 and later CEM3394 based machines/Akai AX60/73, etc.) .
Anyway, I'll leave the rates to people who have done more playing
around with that aspect. I'm not sure where the trained ear begins to
really whine in terms of raw sample rates. My wild and crazy guess on
envelopes, to cover it well with one 1ms filter, it's going to have to
be at least 18 bit resolution at 1khz. I know you can hear it at 16
bits on the prophet 600 in certain applications. An extra two bits
should resolve that to close enough I would think. Those envelopes just
stone wall at the end just a little too much :-). -Bob
Tom Wiltshire wrote:
> Hi All,
> I've been spending the weekend doing a bit of research into the
> oberheim Matrix synths and the Xpander.
> The Xpander's multi-mode filter design is really something special,
> and has been mentioned here before (check the archives - http://
> search.retrosynth.com/synth-diy/search/ mar 97, sept 04) but
> one thing that I can't find the detail on is the envelopes and LFOs.
> The Xpander uses a pair of processors, with one of them being
> dedicated to generating all the envelopes and LFOs (and lag
> processors, and ramp generators, and tracking generators and, and,
> and..!!). Basically, all the slow control voltage processing is done
> in software. This processor is (I read somewhere) a 6809 running at
> 16MHz. This isn't clear at all from the schematics I've seen,
> although 68000 series is a possibility.
> What I want to know is what output sample rate is this processor
> managing to produce. It is dealing with 5 envelopes and 5 LFOs per
> voice, plus all the other stuff, for each of 6 voices. This is 30
> LFOs and Envelopes, which has got to take a while to work out. I've
> done similar stuff on more modern processors, and felt like I was
> running out of time generating only 4 envelopes. However, I was
> working with a 25KHz output rate.
> I suspect this is one of the areas where corners can be cut far more
> than most people would accept. The Waldorf Wave also uses software
> envelopes, and these are only updated at about 53Hz! (http://
> www.unofficial.waldorf-wave.de/wavetech.html). Yet the Wave is held
> in high regard. If anyone could do a similar test on a Oberheim
> Xpander, I'd be interested to know the result.
> So the question is "How low can you go?". 53Hz might be really silly,
> but would 1KHz be enough? 5KHz? I know some of the software sound
> languages (csound or puredata) use rates of 6KHz or so.
> What counts as a reasonable sampling rate for LFOs and Envelopes?
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
More information about the Synth-diy