[sdiy] POLY 800

[Bob Weigel]

There are several basic architectures of concern

1) the Digitally 'controlled' polysynth.

These instruments afford us the convenience of patch storage 
primarily,**  In order to interface the digital control with the analog 
voltages needed to control the analog circuits, these synths usually use 
a scheme were the CPU steps through addresses at some point in it's 
routine.  These addresses open different lines on analog switches like 
CD4051 chips.  Each input is hooked to a different control voltage from 
the panel (Like the cutoff amount pot for instance, or the VCF EG 
attack).  As the CPU selects the voltage it currently wants to evaluate, 
(this whole thing is skipped obviously for synths that have no knobs.  
If data entry wheels are used then the CPU is told by interrupt or 
polling that this is what you are trying to do.  It then usually feeds 
back to the display the value that you are currently selecting and 
....does the rest of what I'll describe here..)  then the 4051 chips 
common line appears with that voltage and it is wired to a comparator.  
The other leg of the comparator is wired to the DAC output.  Either 
right before or after the CPU is asking the value stored in RAM to be 
sent to the dac.  So now you get the picture..if that DAC value is too 
high then the comparator flops one direction...too low..if flops the 
other.  The cpu reads the output of the comparator and then begins 
incrementing or decrementing the value on the data bus that is being 
read by the dac until the comparator reverses it's output!  Now there is 
a threshold built in of course.  If it only inc or dec'd once it ignores 
the alledged 'change' because that's beyond the logical limit of 
resolution in this app.

Ok so also the DAC output is buffered to the common line(s) of a bunch 
MORE 4051's or whatever.  These analog switches output when selected to 
the particular sample and hold cell that is being called up (same one we 
were currently working on above for example cutoff or vcf eg attack.)  
The DAC charges the s+h with the right voltage and when all the 
correcting of any knob movement is done, the cpu goes to the next 
function or the next step in it's routine if that was the last function 
in the sequence.

So if you look at the DAC output on a scope, you'll see these 'random' 
voltages at set time segments since there is always some time allotted 
to make sure the S+H is at least replenished so it doesn't sink over 
time even if no knobs have moved.  And as you move a knob you can see on 
the scope which time segment relates to that particular function.

So anyway the outputs of the S+H's go to the actual analog 
circuitry..for instance the control voltage for the filter's cutoff, or 
the EG's attack.

2) Synths with digital wave generation and / or envelopes and /or LFO's  
but otherwise similar to 1).  

    Various schemes are used for the generation of waveforms in 
non-analog oscillator synths.  Some like DK600 use an analog HF 
oscillator then divide that signal down to produce note frequencies.    
This synth has no individual pitch envelopes obviously.  It's 6 voice 
only because of the polysynth architecture that follows what could 
potentially be a fully polyphonic synth.  The polymoog for instance uses 
this same type of waveform generation but has a voice card for each 
note.  The DK600 only has 6 of these though they are much more 
elaborate.   The Crumar Triology is another divide down based synth but 
it maintains full tonal polyphony but only has 6 actual synth cards like 
the DK600.  It routes all C's and F#'s to one card, all D#'s and G's to 
the next, etc.  12 notes /octave so 2 notes per card requires indeed 6 
cards.   In this way the envelope retriggers when you hit another note 
that is in *that* group.  Interesting concept.  Anyway that machine had 
no patch storage on the panel. Like Polymoog it had some hard wired 
presets and a 'panel' selection which makes the knobs live.  But it does 
use a form of digital wave generation as do many of the polystring 
machines like Omni, Crumar Performer etc.  These usually had just one 
synth 'voice' and are referred to as 'paraphonic' in that while all 
notes sound, there is only one envelope and filter available to modulate 
in the course of play.  Some like the Triology and Korg Delta allow you 
to select whether you want single or multiple triggering for the limited 
number of EG's.

Anyway other machines like the Roland JX series use digital signals to 
approximate sawtooth waveshape and of course to produce square and pulse 
waves as they are natually inclined to do already :-). 

It and other otherwise fully analog signal machines like Prophet 600 and 
Chroma polaris also have digitally aproximated envelopes.  It shows more 
on some than others. hehe...I was thinking..it would be *kind* of a cool 
feature to ...have a 'bit trimmer' knob on a synth so you can go from 
high res and gradually bring it down to produce an 'analog to digital ' 
kind of transition. Cool?  Anyway Roland went from real EG's to digital 
between Juno 60 and 106 also but they are pretty decent resolution on 
the 106 so you don't notice that much.  On the P600 it's a little 
annoying especially at the end of long sweeps where the sound just cuts 
off stone wall.  Oh also of interest I know Roland pulsed capacitors in 
an actual EG circuit on the JP4 rather than trying to convert D/A.  That 
machine is a bit of a mystery because when you are on a preset, the 
sliders have no affect on anything except the globals of course.  So, I 
mean they COULD have just gated between the physical voltages for the 
'manual' selection, to DAC S+H buffers for the stored settings and 
presets.  That way they could have had nothing digital involved at all 
when you are in "Manual" mode. However..that isn't what they did.  They 
followed the standard polysynth digital control method in 1) 
above...except obviously didn't bother reading knobs while they were on 
presets.  Which is kind of a bummer but oh well.  That's what notebooks 
are for :-)

Anyway..sort of in between here and the next step appeared many 'hybrid' 
synths which usually had analog filters with some kind of direct digital 
control and digital 'wavetables' so that the entires signal was process 
then just run through the analog filter basically.  The Ensoniq ESQ1, 
Kawai K3, Korg DW series along with many samplers like the DSS1 fit into 
this category.  The poly 800 is not far from here though I can't 
recall..it may actually 'let the bits out' before the filter for some 
reason :-).  That machine only has one filter like the Siel DK70 btw.  
So again envelopes retrigger when notes are let up or every time you hit 
a note depending on the mode selected I believe on that one.

The DK80 is a splittable dual DK70 basically and they both use analog HF 
oscillators again but have a chip which takes it from there.  It 
receives analog voltages for the envelopes...but handles all the 
waveform generation for the 6 voice polysynth.  SO MUCH of what you'll 
see is clearly examlified by these synths.  Digital signals from the 
keyboard matrix and CPU just...go inside this poorly documented chip and 
you kind of have to look at the layout and guess what goes on inside to 
some degree sadly. 


3) Digital signal processing synths-

    The DX7 uses no analog chips in the signal generation at all and was 
a total breakaway from the rest of the industry in this way.  I mean 
there are op amps after a dac to produce analog output but..I mean 
nothing besides that.  It and the machines that followed used custom 
IC's which did the signal processing tasks of that particular 
architecture design.  FM modelling on the DX's.  Harmonic 
additive/subtractive modelling on the Kurzweil K150, Kawai K5 and later 
K5000, "Phase distortion" on the casio CZ series, and a plethura of 
sample playback based synths.  The Kurzweil K250 was the first real 
'workstation' that included a sampling interface, hammer action keys and 
high quality sounds using part of the bits as dynamic level shifters 
(because 12 bits with a 4 bit exponential 'scale' is a LOT better 
sounding that raw 16 bit.) (Ray built this thing in 1984!)






**( since note triggering can be done without having to control the 
synth in any way except making the key contacts close or putting the 
proper code in the data stream wherever that needs to be done depending 
on how much digital stuff is already going on there.  For example the 
arp omni has individual switch lines which trigger analog circuitry to 
cause the note to go on.  The Yamaha SK-20 for instance has a key coder 
chip that logically assesses the states of the switching lines which 
interact with diodes to minimize the number of physical wires needed 
from the keyboard.).  Were it not for this the controls could simply be 
fed thr

Dirk Vanden Berghe wrote:

> That's excactly what I mean, as far as the Formant is concerned I 
> understand everything. Once MIDI and any digital stuff gets in the 
> picture, I seem lost.
>  
>
> */jure zitnik <kokoon at gmail.com>/* wrote:
>
>     but poly-800 is pretty much all-analogue (except the DCO and the
>     chorus...)
>
>     On 3/17/06, Dirk Vanden Berghe wrote:
>     >
>     > Hi all,
>     > My knowledge on synths never got any further than the analog
>     monofonic stuff
>     > - starting with Formant back in the seventies.
>     > Could I find tutorials on the web or buy any books that would
>     allow me to
>     > step into digitalised synths from the eighties. Poly 800 and so
>     on...
>     > thanks
>     >
>     > Dirk
>     >
>     >
>     > ________________________________
>     > Yahoo! Mail
>     > Use Photomail to share photos wit! hout annoying attachments.
>     >
>     >
>
>
> __________________________________________________
> Do You Yahoo!?
> Tired of spam? Yahoo! Mail has the best spam protection around
> http://mail.yahoo.com
>