a friend sent me this, its been around for awhile. i can't remember where it originally was created. I'm sure someone here will be able to share insight into that:
DSG's: 2 is never enough. My favorite DSG patch is to get two SG's to
trigger themselves at different rates. Use one as a sample-and-hold
source, the other as a sample-and-hold clock. You get some amazing
melodies this way.
Any time you set up a feedback loop in control signals you can get interesting results -- for example, set up a patch using a DSG+UAP to control amplitude,
and then run the output audio into a DSG patched as an envelope follower, and CV the TG attack withthe follower output, and mess with knobs until it's meta-stable.
As a Serge enthusiast since 1977, I can but only agree with what Chris said
about the DSG. I believe, that the DSG is the ultimate 'floating module'in a
synthesizer: you program it within itself for what you want it to be (a VC
LFO, A VC Transient Generator, an envelope follower, a VC Pulse delay, A sub
harmonic series generator (divider), an audio oscillator and a non-linear
audio processor (filter)).
Used as a transient generator, I have discovered that using "feedback"
(reinjection of DSG output into the Rise/Fall VC input) is very useful.
For example, by adjusting the VC input potemtiometer clockwise to 2 o'clock,
one obtains a "release" shape with a sort of logarithmic twist: the ideal
percussive envelope for xylophone type of sounds.
/ \ Log Release
/ - \
By adjusting the VC input potemtiometer counterclockwise at 10 o'clock) the
feedback creates a "release" with a sort of resonance peak (an upward bump),
then falls suddenly vertically.=20
/ \ =20
/ \ R
I have also experimented with passing a DSG output into Serge's VCM module3.
By adjusting potemtiometer fully clockwise and taking the rounded sinewave
output: the envelope becomes much more percussive.
Usually, for envelope shaping I use two DSG to create composite envelopes: I
reinject the output of DSG 1 into the input of DSG2 and adjust the settings
The DSGs make great lag processors, especially when you modulate the
rise/fall. It lags the up differently than the down. DSGs also make
cool filters, they pull out transients first (because they lag the
sharpness off, round it out to synewavishness). especially good to take
the edge off a waveshaped crunch on audio signals.
The Slope Generator's output is linear, but by patching the output back into the attenuated VC input you can create exponential waveforms (with the attenuator knob set after 12 o'clock) or log waveforms (with the knob set before 12 noon).
harmonic series generator (divider)
You connect an external clock to trigger input. With the "Rise" potentiometer fully clockwise (5 o'clock) and the "Fall" potentiometer to 9 o'clock. Let's assume that the external clock is very fast.According to the position of that pot (between 9 o'clock and 10 o'clock) the DSG will act as a Divide by n Counter. For example, you can program it so that it triggers every 32,16,8,4 bpm....
This division process is very interesting if you want to divide a master clock between four DSG used as dividers/transient genberators: program the first one to trigger at 32 bpm, the second DSG to 16bpm, the thrird DSG to 8bpm, and the fourth to 4 bpm. Very complex percussive envelopes can be
generated that way and everything is in sync....
You can also use the same patch to process audio sounds in the 'IN' input: you can emphasize or subtract harmonics (filtering). To do that, turn the "Fall" potentiometer slowly from 5 o'clock to 9 o'Clock.
How can this module act as a filter? It slew-limits the high frequency components of the incoming signal, at a rate set by rise and fall knobs, and VC. If the rise and fall knobs are set fully clockwise (fastest rate), filtering will be audibly minimal, seeing as the Slope Generator is responding rapidly to input. Turning rise and fall knobs counterclockwise increases the response time, and high frequency content at the output will decrease. Turn them fully counterclockwise and audio output will be minimal. So feed this thing a low harmonic content waveform like a sine or triangle, and you've got a rough equivalent VCA!
This sort of filtering action is more subtle than the regular VCFs, though a bit more obvious than a 'tone control.' Good for mellowing out raspy and buzzy tones.
Cross-connecting two Slope Generators together generates interesting 'chaotic' waveforms. Connect the output of one to the attenuated VC input of the other, then take the output of the second and patch it to the input of the first. Play with the attenuator settings.
The DSG has a 'Rise-Both-Fall' switch. This determines whether the control voltage coming in at the 'VC IN' jack controls the rise time, the fall time, or both. Since you can use feedback to create output exponential or logarithmic output curves, you can use this switch to create combination curves where the rise is linear (set the switch to FALL) and the fall is exponential or logarithmic, or the rise is curved and the fall linear, etc. The 1V/OCT jack is not affected by the switch.
Voltage controlled waveshaping is possible when DTG/DSG Rise/Fall/Both switch is in Rise or Fall mode
The LED's on this module and a few others act as a level indicator, dark=off, bright=full on, or in-between.
OK, so this thread is almost a month old, but I think this is a simpler patch, that actually works better than the others proposed. The question was, how to use a DSG/DTG to get a true attack/decay envelope, which can be retriggered before completely decaying.
trigger in -> DSG #1 trig. in
DSG #1 fall full right
DSG #1 output -> DSG #2 input
DSG #2 rise full right
DSG #2 output -> envelope out
Then set your envelope rise time on DSG #1 rise, and your fall time on DSG #2 fall. Since DSG #1 falls instantly, it can be retriggered as soon as it finishes rising, and DSG #2 will obligingly follow it up as
fast as it wants to go. (DSG #1 could be any DTG, and DSG #2 could be a DTG with the signal input jack option.) You can also easily VC the rates.
Re John Lofflink's patch, in addition to needing an extra module, setting rise time is more fiddly: you have to set it the same on both DSG/DTG stages. (However, that patch will work with two vanilla DTGs, since it doesn't need the signal in.)
Re James Coplin's (modified) patch, I don't understand why you want a rising edge out of DSG #1 into DSG #2 input - only the trig. in would care about a rising edge. With the nonlinear rise on DSG #1, you either get a nonlinear envelope, or again you have to set DSG #2's rise time to be the same as the effective rise time on DSG #1. Actually the above patch is the same as your original (linear) patch, with the knobs set differently.
True AD EG
Here's the patch. It will take two stages of a DTG or DSG with the post 1994 GATE OUTs as opposed to the old END out, as well as a scaling processor stage.
GATE to TRIG IN on first DSG stage, set Rise/Fall to short values. GATE OUT gives an inverted trigger.
GATE OUT of first DSG stage to Scaling Processor INPUT2, Bias with +5V on Input 1 and full invert of INPUT2 (you could also use a Dual Processor). This gives a positive biased trigger.
OUTPUT of Scaling Processor to second DSG stage INPUT. OUTPUT is your triggered AD. Set Rise and Fall to taste, but if Fall exceeds the time between GATES then you'll get envelopes that retrigger at the falling slope point, which is normal retriggerable envelope action.
Using the regular output is fine for percussive (fast) attacks, but what if you want to do the same thing with longer attacks? The solution would seem to be to use the Loffink's gate & inverter method. For the inverter you could use just about anything, incl. the BLOG inverter.
The attack & decay time of DSG #1 in the chain would have to be longer also to 'feed' the attack buildup of DSG #2. For added fun, VC the attack & decay times of DSG #1.
In this case, I think the DSG #1 would be called a 'one-shot': hit it with a pulse and it gives you one pulse out, with 'on' time you can vary if you want.
You can take the output of the first section in the USG into the VC of itself and set the VC control all the way left and set to modify the rise time. This will still create a fast enough rising edge to trigger the input section of the other half of the USG. Adjust the rise time on the first half to taste as you would with the inverter. The VC adjustment makes the rise time an extremely rounded squared off wave. This allows for long attack times except in cases where very long attack times are needed.
The main problem is trying to change the clock as a gate stream to a pulse stream to avoid the sustain stage of the USG. I just took the gates from the clock or whatever into the trigger in on one part of the USG with the attack and decay set to zero. Take the output of this input to the input of the other
part of the USG and set your slopes accordingly. The attack time of the first usg seemed plenty fast to not have to use the gate and go through an inverter etc. Any objections to this approach?
As an aside, if you are using the USGs as a envelope with a sustain stage, the level of the sustain is equal to the voltage present at the input. Sticking an attenuator on your gates gives nice control of the height. I'm particularly fond of running the gates through a vca of some sort using a sin to set the envelope height.
Although Bob's AD patch was new to me, I have used a similar patch for a more complex EG. For those of you who don't have a Serge ADSR (or could sometimes use an extra) you can create a close approximate using a DSG/DTG and an ACPR Active Processor. Keep in mind this
requires an actual *gate* source - such as TKB KP - to function properly.
Gate in -> SG #1 trigger in AND SG #2 signal in
SG #1 out -> ACPR input #1
SG #2 out -> ACPR input #2
Attack time set by combination of SG #1 rise and SG #2 rise
Decay time set by SG #1 fall
Sustain level set by ACPR 1/2 pot*
Release time set by SG #2 fall
* full left produces an ASR, as with any ADSR at max sustain level