Analog modeled waveforms. Who's doing it?

[aciddose]

I understand your post as satire of course. I'm also very much aware that you are most likely also aware, but I'll state it none the less.

I suspect that a thread like this will never come to any reasonable conclusion. The blind leading the blind.

[eatyourguitar]

the moog modular 901 cloned in the lab using a real unijunction transistor integrating the power supply to a timing capacitor. after I did some tests I parted it out. dust to dust.

[xalama qo]

I often think I hear something in synths in the attack portion of the note. Almost as if the waveform has something different going on in it's first few cycles before it stabilizes. I wonder if this is what I'm hearing. It's subtle but there.

Hi zero, it's possible you're hearing the difference between having the oscillators free-running, or set to restart their cycles in sync at the start of a note. The free-running option can sound a bit less defined (didn't want to say mushy, cos that's not accurate at all), but there are certain sounds which sound more organic with the subtle note-on variations. Remember when it was taboo to sample 808 and 909's because each kick sounded 'different' and that was the appeal of the real thing. How times have changed.

I coincidentally downloaded the user guide for Diva this evening and it describes the option to phase-sync or free-run oscillators......yup, my terminology's screwed, but I mean at note-on, not sync as in cliche Prophet 5 or Odyssey.
Anyway, the Diva manual is a good read, and IIRC you said you have Diva?
But it is subtle stuff indeed

Perhaps I misread, I thought he was talking about hearing something in analogue synths. (Note: When I say analogue, I don't mean plugins, I mean actual analogue synthesizers made with real solder and silicon.)

He did go on to ask if any devs are replicating the 'imperfect waveform' phenomena, so I assumed VST devs. Then he mentioned the Venom using single cycle samples from various real analogs. So I thought the 'anyone doing it" part of his question was aimed at guys involved in VA development, both hard and soft.
As far as I know it's been a key element in analog modelling since day one...VAZ had pretty unique sounding osc's in the 90s, ReBirth sounded like it should. Pro53 always sounded like stock Reaktor stuff with a fancy (at the time) filter attached...I'd be surprised if they modelled a P5's osc's at all.
I always thought the guys who didn't model osc's were taken out back and beaten up! Scope's Minimax isn't perfect, but was the de facto ITB mini for a long time. Then didn't Arturia bungle a bit on their Mini osc's...the sawtooth ramped in the wrong direction, or something like that, which they corrected in a later revision.

Anyhoo... I took the OP's question as aimed at software devs and DSP devs. But this thread has gone very analog and above my head in the meantime, so I'll bow out and just lurk.

[mystran]

I understand your post as satire of course. I'm also very much aware that you are most likely also aware, but I'll state it none the less.

I suspect that a thread like this will never come to any reasonable conclusion. The blind leading the blind.

Well .. I don't know about conclusions, but it sort of feels like people have a dozen different definitions of what exactly does constitute an oscillator; some people like to look at the "core" while others like to take a complete module and look at the output jack. Yet others might look at the output of a full synth with filter set to "neutral" (as if there was such a thing).

I actually seriously thought of suggesting that anyone making any claims and/or posting pictures specify the actual circuit, the specific point it was probed at to obtain the measurement and preferably the impedance of the probe/scope too (just in case).

But then I realized this was in Instruments rather than DSP dev, so I just figured I wouldn't bother.

[dumbledog]

Hey guys! So I'm stuffed stupid full of Turkey and the Niners are getting their butts kicked so I figured I'd take us all on a magical journey to Sawtoothland! Let's marvel at all the sawtooths these marvelous VSTs and AUs can produce! Gasp in delight as Gordon Reid has fits over which ones are actually ramp waves! Why? Honestly I don't know I just started and followed through I mean I really should have been watching Blade Runner instead but oh well

Let's start with Arturia's collection:











Want more? I got like a million of them

[Urs]

Well .. I don't know about conclusions, but it sort of feels like people have a dozen different definitions of what exactly does constitute an oscillator; some people like to look at the "core" while others like to take a complete module and look at the output jack. Yet others might look at the output of a full synth with filter set to "neutral" (as if there was such a thing).

The easy way would be to look at the circuits and figure out what the manufacturer says about it. I think that in all Roland Service Manuals the capacitor before the filter is depicted as part of the filter board. Furthermore, if we look at, say, the SH-101 or the Minimoog, there's no capacitor between VCO outputs and the mixer section. As the mixer is clearly (I hope) some part that is "beyond oscillator", I think in many cases we can state "no, the capacitor isn't part of the VCO". Lastly, in most of these cases the waveform looks completely different than anything one can sample from such a synthesizer's regular output jack. Hence presenting a belly-shaped waveform as "sampled from an analogue oscillator" borderlines deception.

[Sendy]

[aciddose]

http://imgur.com/a/k6RcS



Of course I don't have a whole lot of synthesizers to provide direct measurements from the oscillator.

I will tell you this though: Every single waveform will look exactly the same at this resolution. Always.

For example are you aware:

• The peak level of this wave is 5v
• Some complex mixture of noise (not merely white) is at about 500uv peak (-80db) with Gaussian distribution
• Some PSU hum is present at an even lower level, I'd estimate 200uv (-90db)
• Some of the digital clock leaks on to the output at about 1mv (-70db) (~2mhz)

None of that is audible. You have to use some of the advanced features of the scope to ever see it. None of it has any effect on the signal you actually hear.

If you want to see something nasty take a look at the output of your DAW.

Actually I should post some scope photos of that to freak you all out. You'll start some sort of cult where you believe a DAW can never reproduce any signal without "unholy corruption" and so therefore must sound terrible.

[JPQ]

Tone2 Saurus has weird waveforms. i dont post shots test itself. and this is reason why i really love it and makes even harder wait my real DCO synth(s).There reads synth(s) i dont yet know how many i need...

[ghettosynth]

Perhaps I misread, I thought he was talking about hearing something in analogue synths. (Note: When I say analogue, I don't mean plugins, I mean actual analogue synthesizers made with real solder and silicon.)

He did go on to ask if any devs are replicating the 'imperfect waveform' phenomena, so I assumed VST devs.

Yes, replicating the phenomena believed to be in analogue oscillators. That part isn't confusing.

Anyhoo... I took the OP's question as aimed at software devs and DSP devs.

Of course it is aimed at those devs, that wasn't at all what I was talking about. ZC stated that he thought that there was something going on in the first few cycles of a waveform. That makes little sense in an analogue context which was the essence of my post. Although I have my doubts, beyond what was already suggested with respect to phase reset (in plugins), if there is something in plugins, it's not modeling an analogue phenomenon, hence, that wouldn't make much sense in the context of this conversation.

[JPQ]

Also nice see arturia waverforms actually differ in different synths. at least sawtooth btw i like some their sounds not most analog maybe but still.

[aciddose]

ZC stated that he thought that there was something going on in the first few cycles of a waveform.

This one is easy.

The VCF cutoff will be set to sustain at some level as well as the VCA amplitude in between notes. For example it is possible to apply the gate to the VCA (or a 2nd envelope) with a quick release and have a slow release applied to cutoff.

Now depending upon what the cutoff is set to before the attack starts, the attack will of course be different.

Not just that though! ...but wait! there's more!

There can be CV feed-through in the VCF and VCA as well as DC offsets.

For example the VCA might have a constant offset depending upon the CV input. This is normally filtered out by a high-pass filter inside the VCA, but some offset can still leak through when the amplitude CV changes. So let's say you get 1/10th CV leakage. (this is super high and unreasonable, but just for example...)

The envelope jumps to 10v during the attack, let's assume it's instantaneous. This gives ~1v appearing inside the VCA. If the high-pass filter is set to 16hz (typical) it will take 23ms for this 1v pulse to decay.

So we now have a 23ms decay at the start of the note. If there is any non-linear effect taking place in the VCA or any stages afterward, the whole signal will be shifted 1v positive. If the signal is normally 5v wide, this is 20%, which is a significant asymmetry.

The result might be that for a short period, the first few cycles will have a lot more even harmonic than otherwise.

While 16hz is typical (10k, 1u), 1.6hz is also typical (10k, 10u). This would be 230ms decay.

These sorts of CV feed-through pulses are extremely important to what makes the "analog" sound. The MS-10/20 (the KORG-35 filter) is an excellent example of this. Even the OTA filter uses the same transistor VCA which is extremely non-linear, so revised/late model MS-20s definitely exhibit exactly the characteristic zerocrossing describes.

(The solution is to reverse the collector/emitter and the resistor which is normally on the collector. This brings the leakage down to nearly zero. This is a "bug" in the original designs that lends some interesting buggy flavor to the sound. Often undesirable but cool in a mad-max sort of way.)

[ghettosynth]

ZC stated that he thought that there was something going on in the first few cycles of a waveform.

This one is easy.

The VCF cutoff will be set to sustain at some level as well as the VCA amplitude in between notes. For example it is possible to apply the gate to the VCA (or a 2nd envelope) with a quick release and have a slow release applied to cutoff.

Yes, I was simply saying that it has nothing to do with the oscillator, as I said:

What you're hearing is not likely in the oscillators.

'

What ZC posted seemed to imply that he thought that analogue oscillators start oscillating with each note in an analogue circuit, which, of course, is generally not true. When I said that I might have misread, this is what I was talking about.

There are any number of explanations for why an attack sounds different note to note, but they are unlikely to have anything to do with how the oscillator "starts up."

[aciddose]

That's actually a possibility too, although a lot less likely.

The oscillator could take time to settle out at a new frequency assuming there is something weird going on.

Yes the oscillator would generally not "start up" at any point other than when you first turn on the power, so any theory like that must be very technical and specific to the weird oscillator it applies to.

[ghettosynth]

That's actually a possibility too, although a lot less likely.

The oscillator could take time to settle out at a new frequency assuming there is something weird going on.

I thought about that and decided that it wasn't worth bringing up. I still don't think that it is. You would be able to tell if that was the case because you wouldn't hear the effect when you repeatedly played the same note.

In any case though, the point I was trying to get across was that, in general, analog oscillators do not stop and start with played notes.

Yes the oscillator would generally not "start up" at any point other than when you first turn on the power, so any theory like that must be very technical and specific to the weird oscillator it applies to.

Right, and more likely, as I said, anything heard has nothing to do with the oscillator.