[A6] Internal electronics, Found it!

[dj]

Seth Stroh wrote:

> The "obvious" engineering reasons for multiple DACs:

Theres never any obvious reasons in engineering,
merly many solutions to ugly problems! ;-)

> 1. A single DAC isn't fast enough to update all of the CVs for 16 voices

Absolutely shure on that?
The PCM54 has a settling time of 3u sec (Voltage mode) fast enough to let
every channel have a resonable low droop rate at the mux ,one DAC could
do quite well infact, but as i didnt say anything about LFO's and other
modulators
with update rates around 5Khz per channel!! Thats a different matter, but if one
studdies the data  sheet of the PCM 54 then we find that the current output has a

settling time of 350n sec, now this is fast fast enough to run the entire machine

from one DAC and having modulation at high update rates, the OPAMP
buffer for such a DAC current output will be pretty tight, these days its
a easy matter with OP AMPS available in the 100Mhz GBW range.

> 2. The mux/s&h inside the ASO/ASF doesn't switch/charge fast enough to
> be controlled by a single DAC

Thats probably the main reason, the CPU has to latch the DAC value, then
wait for the mux cap to settle the charge, and in between CPU takes the next
DAC and so on. There is always a trade off of mux cap values update rates
and settling times.

I think Xpander and Chroma solwed this in a neat way!

> In fact, it is probably a combination of both!  And the solution to both
> problems is to control groups of voices "in parallel" using several
> DACs.  (So, in an update cycle, you are simultaneously charging the s&h
> for pitch CV on VCO 1a, 5a, 9a, 13a - then pitch CV for VCO 1b, 5b, 9b,

Possibly so , but still, most times in a multitimbral setup for example the
chance
of CV's  being completely different due to different pitches, filter cofficients
etc etc are very high.!

> The not so obvious stuff
> is all of the VCAs - everywhere you can set a level like VCO1 out to osc
> mix, that is a VCA controlled with a CV.  VCO1 sub-oscillator out to osc
> mix, another VCA and associated CV.  Filter 1 bandpass to filter mix,
> another VCA controlled with a CV.  Filter mix out to main mix left is
> another VCA controlled with a CV.  Well, you get the idea :)

Yes i suspected that theres plenty of VCA in the A6!

> This wouldn't be very useful.  It has been stated in previous
> discussions that the CVs used by the ASO and ASF chips do not follow any
> "standard" synthesizer signals like 1V/oct or V/Hz.  Instead, the
> voltages are tailored to the inner workings of the custom chips.

It was just an bad idea of mine, btw up or down scaling can easily be done
externally as long as the CV's sends a linear 1V, 830mV, 1,12V "whatever" CV,
there is "obvious" (bold statement :-) a expo-converter somewhere since
direct linear control of a linear oscillator will generate unaccseptable error
over the Oscillators frequency range. 16 bits cant possibly cope 10
octaves with needed accuracy in a non expo driven VCO!
(Even the manual states this!).
Besides there is no need to cover a 10 octave range, i mean how many
musicans play solo tunes where fundamental is 19,5khz? My ears stop
hearing "pure tones" at at around 15,8khz. First harmonic is then at 31,6Khz!!
Well dogs and such will hear that! :-)

> In fact, the situation is even worse!  What do you think the Andromeda
> is doing during an autotune?  Well, for example, it is building lookup
> tables for the voltages to supply a particular voice chip to get a
> particular pitch.

> Because the capacitors for the oscillator timing constants and the filter

> integrators are on chip, and because on chip...............snip

Was "just" comming to what you say here, in the case of  "not so very linear"
VCO's a tune table is a neat solution to a design problem, its also a short cut
to easyied up silicon design requirements and design time who equals to cost in
a very unpleasant way!! :-:

> I think the Andromeda hardware is fine as it is!

Agrees in full! A very fine job they did! Oh, the software people too! :-)

TJ