On Sat, 25 Jan 2003
media.nai@... wrote:
> Based on your descriptions, regardless of what Yamaha might have
> called these, more accurate names for these functions would have been
> Attack Rate and Decay Rate.
>
> (As you know, this is the case with almost all analogue ADSR's -- the
> knobs control current sources charging a cap, so their settings only
> translate to time values if the attack level and sustain level are
> fixed, and unlike the CS-80 the sustain usually isn't fixed.)
Yes, the should be called rates. Yamaha called them times, however,
which led me to believe that they used some sort of analog multiplier in
their EG chip to achieve a constant time for a given setting of the rate
controls regardless of the initial or attack level settings. After a
bunch of tests on a CS-50 voice board, though, I determined that the times
were not constant as they were affected by the moving of the IL/AL
endpoints. The Yamaha "time" moniker stuck in my head, though. I will
call them rates from now on. (I was very happy to learn that the times
were not constant, as building my equivlant of the EG became much easier).
When I refer to times, I am usually stating the minimum/maximum times
attainable when the levels are the farthest apart. So, a max. attack time
of 30 seconds means IL is as low as possible and AL is as high as possible
to get the most "vertical" distance the slope has to traverse.
> If the CS-80 EG is analogue, wouldn't there would be a bit of a slope
> as it drops from the sustain level to the initial level?? I also
> wonder if Roland were looking at the CS-80 when they designed the VCF
> EG for the TB-303.
Well, yes, a slope defined by the release rate control.
> Do the CV inputs completely replace the knob settings, or do the CV
> inputs modulate the knob settings??
Modulate. Replacing the knob settings would by far too awkward to be
useful. Much more intuitive to set a level or rate, then "move it around"
with a little +/- CV. The "0 to +V" I mention is an extreme, such as when
a knob is fully CCW and the full range of the parameter is to be adjusted
with a CV.
> While 30s is certainly long enough, I don't know if 3ms is punchy
> enough for driving a VCA. Or is this being designed primarily as a
> filter envelope??
Well, my original spec. was to do 1ms to 10s (nominal) but it was
suggested that the times/rates be made longer. It is just changing the
timing capacitor; perhaps the assembly instructions can suggest more than
one value so a user can get what they want.
> > There is a 3-position switch to select the point at which a TRIG OUT
> >pulse is generated: when the attack peak is reached, when GATE IN goes
> >inactive or when the end of the cycle is reached.
>
> Wow!! That gives it a whole mess of new functions!!
>
> However, I do notice that it does not have a trigger input. If it
> did, it could be used as a trigger delay, a VC LFO, and not the least
> expected feature, an envelope generator when using a trigger!!
>
> So how does it respond to gates that drop before the attack or decay
> phase is over?? I guess that's the question.
>
> As you know, the 800 has three different modes, including a
> trigger-only mode that it produces Attack-Decay envelopes using the
> Attack and Release knobs. The UEG has several options how it handles
> dropped gates, including a one-shot mode. Whether this is
> implemented using two switched jacks, or one jack and a toggle
> switch, the ability to respond to a trigger is essential for a
> full-featured EG.
This is one of those things that I'll determine when tweaking PIC code.
Originally I wanted to have separate gate and trigger inputs, but I can
program the PIC to do an A/R function for a dropped gate, etc. I've also
considered replacing the inverted output with a trigger input.
> >PIC control, so if they need tweaked later, it is no problem.
>
> Is it all digital??
No, only the logic circuit to step through the cycle is digital. I
decided to use a PIC part to allow for some flexibility in the way
gates/triggers are handled, the level and rate analog switches are
operated, the LED is handled and trigger outputs are handled. All the
rest of the EG is analog.
Crow
/∗∗/