Author |
Message |
paul e.
Joined: Sep 22, 2003 Posts: 1567 Location: toronto, canada
Audio files: 2
|
Posted: Sun Feb 12, 2006 9:30 pm Post subject:
|
|
|
interesting, but not continously variable?
sounds like the waveform would suddenly change when tha VCA is opened? _________________ Spiral Recordings |
|
Back to top
|
|
|
morbius
Joined: Feb 12, 2006 Posts: 95 Location: Great Smoky Mountains - USA
|
Posted: Sun Feb 12, 2006 9:59 pm Post subject:
|
|
|
paul e. wrote: | interesting, but not continously variable?
sounds like the waveform would suddenly change when tha VCA is opened? |
Ummm... workin' with ya here... tweaking should smooth-out the variables. I'll give it a shot tomorrow.
At worst, I think adding EG's to each of the four VCA's would take care of it. I'm also thinking of another way... still using four VCA's, but instead of a sequencer, using two oscillators for modulation... one with a sinewave, one with a squarewave. The '+' & '-' of each would be used to cyvle thru the four VCA's. We'll see.
Ya know... that's one of the really fun things about analog modulars... there is often more than one way to approach goals... some work, and some work better. _________________ ~Morbius~
http://www.MusicByCybertron.com
morbius001a@yahoo.com |
|
Back to top
|
|
|
ian-s
Joined: Apr 01, 2004 Posts: 2669 Location: Auckland, New Zealand
Audio files: 42
G2 patch files: 626
|
|
Back to top
|
|
|
morbius
Joined: Feb 12, 2006 Posts: 95 Location: Great Smoky Mountains - USA
|
Posted: Sun Feb 12, 2006 10:19 pm Post subject:
|
|
|
g2ian wrote: | Cross fading works in some cases, sine > triangle for example. Cross fading between a square and narrow pulse does not sound like changing the pulse width however.
Some oscillators have a variable slope triangle that can be modulated between saw through triangle to negative saw. This produces an effect similar to PWM but the drastic difference between the spectrum of a saw and triangle gives it a sort of filter modulation effect. I've experimented with a saturation module, which squares up the triangle in the center of the modulation range, the result is a little more useful I think. |
And I was just thinking about the use of a clipper/rectifier to work with what you are talking about. _________________ ~Morbius~
http://www.MusicByCybertron.com
morbius001a@yahoo.com |
|
Back to top
|
|
|
morbius
Joined: Feb 12, 2006 Posts: 95 Location: Great Smoky Mountains - USA
|
Posted: Mon Feb 13, 2006 3:05 pm Post subject:
|
|
|
morbius wrote: | paul e. wrote: | interesting, but not continously variable?
sounds like the waveform would suddenly change when tha VCA is opened? |
Ummm... workin' with ya here... tweaking should smooth-out the variables. I'll give it a shot tomorrow.
At worst, I think adding EG's to each of the four VCA's would take care of it. I'm also thinking of another way... still using four VCA's, but instead of a sequencer, using two oscillators for modulation... one with a sinewave, one with a squarewave. The '+' & '-' of each would be used to cyvle thru the four VCA's. We'll see.
Ya know... that's one of the really fun things about analog modulars... there is often more than one way to approach goals... some work, and some work better. |
Ok... so- I did as mentioned above. I used one oscillator as the source... took the outputs for the sine, tri, sawtoth, and square waves, and patch each to it's own VCA. Each VCA was controlled by it's own EG, and each EG was fired in sequence by a Q960 sequencer.
I added a little attack, and release.... and each had it's sustain up full. The only thing that was NOT identical was the 'amount' of EG that was going to the VCA's for the saw and square (mainly because of the higher overtones).
I started the 960 at a slow rate, then increased it using it's 'range' switch. Of course, I realize that if you truely want that effect (a lot) buying a module that will do it (or plug-in) makes more sense. But, that is part of what makes modulars so much fun... you can build the patches anyway you want to.
Here's the link to the mp3:
http://www.unisynth.com/user/morphingWAVES.mp3 _________________ ~Morbius~
http://www.MusicByCybertron.com
morbius001a@yahoo.com |
|
Back to top
|
|
|
paul e.
Joined: Sep 22, 2003 Posts: 1567 Location: toronto, canada
Audio files: 2
|
Posted: Mon Feb 13, 2006 9:21 pm Post subject:
|
|
|
yeah, that's pretty cool!..thanks for the example _________________ Spiral Recordings |
|
Back to top
|
|
|
Somnium7
Joined: Jul 08, 2006 Posts: 4 Location: USA
|
Posted: Sat Jul 08, 2006 11:02 am Post subject:
|
|
|
ndkent wrote: |
Finally Serge makes a Waveshaper that inputs an external saw and VC shapes that into a sine. Their "entry level" osc, the PCO doesn't have a sine output so this module might have been a unique way to produce one. At the moment no one has a similar or copied module.
|
You mean the Serge Triple Waveshaper.
http://www.cgs.synth.net/synth/serge/pic/schem_tri_wave.gif
This is a simple circuit and does a great job of morphing from saw to sine but the sonic results are rather boring. I found that this circuit can be modified to behave more like the NTO. By removing the diode in the feedback loop and increasing the 100pF cap to 4700pF the circuit will now morph from saw to sine to trapezoid to square. This sounds much more interesting. |
|
Back to top
|
|
|
amnesia
Joined: Jul 29, 2006 Posts: 149 Location: amnesia
|
Posted: Sat Jul 29, 2006 6:49 am Post subject:
|
|
|
You mean the Serge Triple Waveshaper.................
Wow I didnt know there were Serge mods to be had:-) Do you know of any more? |
|
Back to top
|
|
|
Kerr_Avon
Joined: Jan 09, 2006 Posts: 43 Location: UK Aldershot
Audio files: 3
|
Posted: Tue Aug 08, 2006 4:52 am Post subject:
Varying Waveforms Subject description: Cardboard Required |
|
|
I can do that but it's a bit manual and one requires pre-cut cardboard to move the faders.
http://homepage.ntlworld.com/henry01/waveform_synth/waveform_synth.htm
Alternatively, moving from 'anything' to sine and vice-versa can be done with a sharp enough LP filter. |
|
Back to top
|
|
|
State Machine
Janitor
Joined: Apr 17, 2006 Posts: 2809 Location: New York
Audio files: 24
|
|
Back to top
|
|
|
Rob
Joined: Mar 29, 2004 Posts: 580 Location: The Hague/Netherlands/EC
G2 patch files: 109
|
Posted: Wed Aug 16, 2006 2:05 pm Post subject:
Re: continuosly variable oscillator waveforms |
|
|
paul e. wrote: | i may be asking a painfully obvious question..
would anyone have any insights on how to programme an oscillator to 'morph' from a sine to a square, ramp etc over time? |
Feedback linear FM can do an interesting job here. Direct feedback of a sine will increase the amount of harmonics up to an approximation of a sawtooth. Feedback of 'the square of a sine' (sin^2) will build up a series of odd harmonics approximating a square or triangle, depending on the polarity of the feedback.
To make it work some simple rules. First, any DC component must be removed by a highpass filter to prevent the oscillator from detuning. This filter should be set to a relatively high frequency. With analog oscillators I have good results by simply using a 3300pF capacitor in series with a 10kOhm resistor in the feedbackpath from the output of the oscillator to its LinFM input. If the modulation is not deep enough the resistor value can be lowered until it works, but that 3300pF invariably seems to be the magical value over here.
An odd harmonics only wave can be produced by feeding the output of the oscillator into the two inputs of a ringmodulator and feeding the output of the ringmodulator back into the LinFM input, again with a 3300pF capacitor in series with a 10kOhm resistor. By using an inverter module after the ringmodulator it is possible to change from a square approximation to a triangle approximation.
You will need a very good quality analog ringmodulator with very little 'leakage' to get a pure odd harmonics only result. But the sine of an analog 'multiwave' oscillator is hardly ever of reasonable quality anyway, so this will impurify the results no matter how good the ringmodulator. One should really use a good quality 'dedicated' sinewave oscillator with a stable output amplitude, e.g. of the biquad type, together with a good quality ringmodulator to get a pure 'odd harmonics only' result.
Of course one needs to use a proper LinFM input, the 1V/Oct input will not produce the effects described.
With a VCA between the oscillator or ringmodulator output and the capacitor/resistor an envelope can be applied, the effect is similar to the opening of a very sharp filter without resonance.
The sawtooth, triangle and square waves are not as 'geometrically perfect' as the standard saw and square. That is because the wavefrom gives itself a 'sweep' to approximate the flanks in the final waveform, but it takes a little while before this sweep builds up. This causes the saw to have only one sharp corner while the other corner is a bit blunt. But the result definitely gives a warm sound which is imho very useful for e.g. bass synth sounds.
Digitally it is a bit more tricky, as the feedback will most certainly produce partials that lie outside the bandwidth of the system, which often causes a nasty oscillation at half the samplerate and/or produce chaotic noise. The trick is to limit the bandwidth of the feedback loop with a soft LP filter in the loop, set to e.g. 8kHz.
When the oscillator output is squared with a multiplier module to get only odd harmonics an extra lp filter must be applied after the multiplication, as the multiplication involved will also produce partials outside the bandlimit for the feedback signal, which again could produce nastiness when the feedback is deep.
It is a good idea to crossfade between the direct oscillator output and the squared output of the multiplication. 'Crossfading through zero' (the G2 'Fade2-1' module) gives a single knob to morph between a saw through a sine to a square. Standard A<->B crossfading will produce a soft PWM-type effect, not as deep as PWM on a standard pulse, but quite pleasing.
Here is two patches for the G2 demo software that examplify the matter. The first patch is the basic principle. There are two knobs, one morphs between a saw and a square, while the other defines whether the crossfade is direct or through zero to give the sine halfway the other knob.
Note that the feedback is much deeper as e.g. on a DX-type machine, the extra bandlimiting modules make sure that this gives no problem and no noise or nasty very high frequency oscillations are produced.
The other patch has some envs, lfo and seq added to give an idea of the sonic effect when everything gets modulated.
The theory:
Applying LinFM this way creates a recursive situation, the sine itself will produce the second harmonic in the output, the second harmonic will then create the third harmonic, the third the fourth, etc.
When the output of the osc is squared the sine will produce a second harmonic in the ringmodulation. This second harmonic at the output of the ringmodulator will actually create the third harmonic in the oscillator, so the oscillator will not produce a second harmonic. The third harmonic wil now create the fifth harmonic, the fifth the seventh, etc. So, in this particular case the ringmodulator or multiplication acts like a very special type of comb filter that 'blocks' even harmonics in the generation of harmonics in the oscillator. Don't expect a ringmodulator to work like this in other situations.
The amplitudes of the harmonics will gradually decrease for the higher harmonics. The total amount of harmonics is defined by the amount of feedback. Mixing the oscillator output with the squared output in a crossfader will create all harmonics, but the amplitudes will not gradually decrease, a bit like the harmonic series for pulses.
The word morphing is a tricky word to use, as I don't think there is a proper definition that defines what morphing actually is. In visual graphics it is not simply a mix or crossfade, but a deformation based on a freely definable trajectory. The trajectory is often an interpolation between an outline in one picture and an outlike in another picture. When the morph is applied together with a crossfade between the pictures one gets the well known 'cat becomes dog' thingy. To apply this principle on waveforms seems quite another matter, as what is the freely definable trajectory? A crossfade is not enough as in graphics morphing is the combination of crossfading plús the deformation over a given trajectory. My example is also not morphing in that sense, as it is in fact a recursive build up of a harmonic series with no sign of a freely definable trajectory. I think the closest thing to real morphing of sounds can be found in some functions of the Kyma system, where two sound recordings are analysed and their spectra stored. Then a trajectory to 'travel' from one spectrum to the other can be defined and the produced intermittent spectra used to reconstruct a sound that 'morphs' from one to the other.
Still, using VCA's to crossfade between waveforms, possibly combined with waveshapers, can produce waveforms with very interesting sonic properties and I would not object at all to call this 'morphing'. For me personally it is close enough.
/Rob
Description: |
Example of an oscillator producing a wave that can be smoothly faded from a sawtooth to a square, and optionally through a sine in between. The oscillator uses LinFM feedback and provisions are made to make this feedback stable, so the results are predict |
|
Download |
Filename: |
WaveMorphing.pch2 |
Filesize: |
1.46 KB |
Downloaded: |
1665 Time(s) |
Description: |
Example of an oscillator producing a wave that can be smoothly faded from a sawtooth to a square, and optionally through a sine in between. The oscillator uses LinFM feedback and provisions are made to make this feedback stable, so the results are predict |
|
Download |
Filename: |
WaveMorphing2.pch2 |
Filesize: |
2.92 KB |
Downloaded: |
1723 Time(s) |
|
|
Back to top
|
|
|
Gregg Hermetech
Joined: Nov 23, 2005 Posts: 35 Location: Paris, France
|
Posted: Wed Aug 16, 2006 11:21 pm Post subject:
|
|
|
very interesting theories rob! i'm gonna try out some of your suggestions on my frac rack modular later, with various sine waves!
the blacet VCO has a very nice VC WAVE out, which can go from a pseudo triangle to a pseudo square wave. it sounds very nice and creamy/phat/tube like, and is achieved with LED's i believe! |
|
Back to top
|
|
|
Jari Jokinen
Joined: Sep 11, 2006 Posts: 87 Location: Finland
Audio files: 1
G2 patch files: 1
|
Posted: Mon Oct 09, 2006 3:49 pm Post subject:
|
|
|
I am looking for an analog oscillator, which can do the triangle/saw morph (and perhaps other tricks too), like the Nord Modular Shape Oscillator does.
Analogue Systems rs95 and rs953 seem capable:
http://www.bigcitymusic.com/product.asp?cat=new&pid=1000187
But: "The output from the SAW-OUT socket is -2V to +8V when a true sawtooth wave is produced, and ±2.5V when a triangle wave is produced." So, I should expect a volume drop, when going to triangle...?
Plan B Model 15:
http://www.ear-group.net/model_15.html
What else is available?
Best regards
Jari Jokinen |
|
Back to top
|
|
|
ian-s
Joined: Apr 01, 2004 Posts: 2669 Location: Auckland, New Zealand
Audio files: 42
G2 patch files: 626
|
Posted: Mon Oct 09, 2006 5:45 pm Post subject:
|
|
|
Plan B model 15 looks like a VC crossfade so that won't replicate the G2 shape oscillator.
The RS ones are more like it.
There is an apparent drop in volume even when the Triangle and Saw are the same level. Which makes the tri > saw morph less useful than I imagined. |
|
Back to top
|
|
|
Jari Jokinen
Joined: Sep 11, 2006 Posts: 87 Location: Finland
Audio files: 1
G2 patch files: 1
|
Posted: Tue Oct 10, 2006 12:40 pm Post subject:
|
|
|
Indeed, Plan B oscillator does cross fading. Last edited by Jari Jokinen on Wed Oct 11, 2006 3:11 am; edited 1 time in total |
|
Back to top
|
|
|
ian-s
Joined: Apr 01, 2004 Posts: 2669 Location: Auckland, New Zealand
Audio files: 42
G2 patch files: 626
|
Posted: Tue Oct 10, 2006 1:43 pm Post subject:
|
|
|
I didn't intend to imply that the Plan B oscillator was not a great product, simply that it does not replicate the G2 shape oscillator.
FWIW the Moog Voyager just uses cross fading for it's triangle to saw and it is still a useful effect.
I have however seen circuits which are claimed to shape a sawtooth into a variable duty cycle triangle under voltage control. Trying to figure out how they work makes my brain hurt.
I imagine that the reason it is not normally implemented is that PWM is so much easier, and produces a similar but better (as in the tonal balance is more constant) effect. |
|
Back to top
|
|
|
Jari Jokinen
Joined: Sep 11, 2006 Posts: 87 Location: Finland
Audio files: 1
G2 patch files: 1
|
Posted: Wed Oct 11, 2006 12:32 pm Post subject:
|
|
|
g2ian wrote: | I imagine that the reason it is not normally implemented is that PWM is so much easier, and produces a similar but better (as in the tonal balance is more constant) effect. |
You are propably right.
Apparently no analog oscillator does tri/saw like the Nord Modular Shape Oscillator. I made a nice sound with it, which I intend to do with an analog sytem. Changing to PWM doesn't seem to work in this case.
Best regards
Jari Jokinen |
|
Back to top
|
|
|
ian-s
Joined: Apr 01, 2004 Posts: 2669 Location: Auckland, New Zealand
Audio files: 42
G2 patch files: 626
|
Posted: Wed Oct 11, 2006 4:30 pm Post subject:
|
|
|
Found one of the shapers I was talking about.
here
I tested the basic idea with a G2 patch and it seems to work. Maybe a DIY project for you.
I also read that the CEM3396 chip had a wave shape with VC triangle to saw. |
|
Back to top
|
|
|
davep
Joined: Jul 05, 2004 Posts: 467 Location: Oakland, CA
Audio files: 10
G2 patch files: 73
|
Posted: Thu Oct 12, 2006 5:25 pm Post subject:
|
|
|
FWIW, the Oberheim Matrix6 has DCOs that have a waveforms that can be modulated from saw to triangle. Not by crossfading, by changing the up & down slopes like the G2 shape osc. _________________ Dave Peck |
|
Back to top
|
|
|
Jari Jokinen
Joined: Sep 11, 2006 Posts: 87 Location: Finland
Audio files: 1
G2 patch files: 1
|
Posted: Wed Nov 15, 2006 5:32 am Post subject:
|
|
|
Regarding RS95:
> Here goes the sweep test...
Thank you. Just as mentioned in the RS95 manual, the triangle waveform is much quieter than saw.
Furthermore, when morphing, there are three segments in the waveform: Raise, jump up, and long fall - or alternatively long raise, jump down and fall. This makes it sound more like a cross-fade.
From:
http://www.thereminworld.com/forum.asp?cmd=p&T=2169&F=1&p=3
Best regards
Jari Jokinen |
|
Back to top
|
|
|
|