Entertaining "Diva or OB8" poll happening on GS

[ghettosynth]

Oh... it came out as I had hoped - Diva first, Obie second. Certain people who wanted the Obie to sound better got it wrong

Hi Urs. I did the test - glad you found it - and got it right

Big thanks for adding the Uhbie filter - I think it makes a tremendous difference. I will probably be doing a part 2 to these tests - further investigations - but I pledge to make fair tests...

If you do, push the filters into resonance and sweep them quickly.

[SWAN808]

Urs or can anyone confirm - was the Dual VCO modelled off a JP6 or JP8?

[Urs]

Urs or can anyone confirm - was the Dual VCO modelled off a JP6 or JP8?

Well, both, more or less.

The Jupiter 8 uses VCO1 as CrossMod slave while it is master to Sync VCO2. This IMHO is a fundamental conceptual flaw, so we did it the other way round, VCO1 is master for both Crossmod and Sync (which is possible in the JP-6 of course, both ways).

The shape of the triangle differs between the two and both are far from perfect. Hence, Diva has three options for the Dual Osc triangle.

[SWAN808]

Urs or can anyone confirm - was the Dual VCO modelled off a JP6 or JP8?

Well, both, more or less.

The Jupiter 8 uses VCO1 as CrossMod slave while it is master to Sync VCO2. This IMHO is a fundamental conceptual flaw, so we did it the other way round, VCO1 is master for both Crossmod and Sync (which is possible in the JP-6 of course, both ways).

The shape of the triangle differs between the two and both are far from perfect. Hence, Diva has three options for the Dual Osc triangle.

interesting..but I thought there was quite a difference between the JP8 (discrete?) and JP6 (CEM) oscillators?

[Urs]

The main difference between vintage quality oscillators from Moog, Roland and Curtis was the long term time varying behavior (drift). The sawtooth was and is a perfect sawtooth, the slopes of the square and pulsewaves are perfect. The common triangle was the biggest concern, as it could easily glitch into some underlying sawish spectrum.

Other than that, whenever people talk about the importance of whatever "belly" or else a sawtooth has, they talk about the circuitry that follows it, and maybe the levels at which they run into the mixers and filters.

Therefore, apart from triangle and maybe smoothness of sine it's really just gain staging, drift and general features (e.g. modulation ranges) that mark the difference of oscillators built with quality parts.

Put an oscilloscope right behind the VCO in a SH-101 and you'll get an absolutely perfect, linear "no belly at all" sawtooth.

The only analogue synths we have where oscillators matter is the MS-20 and 800DV. Those can be quite "off" of what one would expect.

[pdxindy]

It also took me awhile to figure out where the transitions are. Last night I ever so slightly preferred the first entry, but would very happily use either (and I know that if I listened again, I could easily prefer the second). But I knew that would be the case before I listened and the poll bears this out: there's no discernible difference. Analogue synthesiser sound fetishists increasingly sound like oenophiles or audiophiles (and we know how those worked out). Loving an analogue synthesiser as an instrument is another matter entirely, though, and I totally understand that.

Don't be so hasty to dismiss analog fans as audiophiles. It is often the case that there is no usefully discernible difference, as in this case. What is very common about these examples, however, is that the filter is not being pushed in any way and the oscillator is not being modulated in ways that matter. You don't need Diva to reasonably emulate most basic Oberheim sounds as many satisfied customers of a very popular Oberheim emulation will tell you. The reality is that this has been true for years. For most audiences this has been true for almost two decades. My audiences were quite satisfied with the sound of the Nord Modular filter in 1998.

I'm not a purist, at all, and I think that Urs really has pushed the envelope, I'm a satisfied customer. There is still some actual value in real analogue circuitry, however, and tests of simple sounds doesn't refute that.

I agree... for lots of 'calm' sounds there is little difference, but the more you push the synth the more analogue still has something to offer.

Also, softsynths are a bit like samples... play a sample of a trumpet and it does sound exactly like that trumpet that was recorded. But try to play a song and there the difference is obvious... a real trumpet has a nuance and expressiveness that samples cannot meet... especially live

To me, the 'magic' in analogue is in the playing... the subtle character in the transition from one note to another... the variation in how the envelopes and filters retrigger while playing... in the tone of the resonance as it is turned up and how it interacts with the noise... and how all those things combine in unexpected ways... especially played live.

Besides generally sounding gorgeous, Diva is a step forward in capturing some of that... Diva holds up well when pushed and has some lovely non-linear responses. You get something for that cpu use!

[SWAN808]

Other than that, whenever people talk about the importance of whatever "belly" or else a sawtooth has, they talk about the circuitry that follows it, and maybe the levels at which they run into the mixers and filters.

very interesting thanks Urs. So are you saying the circuitary after the Osc saturates the wave or adds some EQ change? That is generally responsible for the character difference in the different manufacturers?

[Urs]

Other than that, whenever people talk about the importance of whatever "belly" or else a sawtooth has, they talk about the circuitry that follows it, and maybe the levels at which they run into the mixers and filters.

very interesting thanks Urs. So are you saying the circuitary after the Osc saturates the wave or adds some EQ change? That is generally responsible for the character difference in the different manufacturers?

Well, I think that most quality oscillators sound alike. They may vary in gain of one waveform to another, a bit of triangle shape here and there, but a sawtooth of one is indistinguishable from a sawtooth of the other, if taken straight from the oscillator. If we tapped in there before the filters and did a comparison, no-one would even recognize the sawtooth of his own synth. Too much of the character is contributed by the following circuitry.

The typical belly-shape of the sawtooth often assiciated with a Minimoog VCO, or anything else fwiw, is created by nothing more than a capacitor that's put between the oscillator and the filter. This acts as a DC-blocker (essentially a highpass filter) which transforms the unipolar (DC) VCO signal into a bipolar (AC) one. There's nothing mythical about it. Add a highpass filter behind Zebra's otherwise perfect sawtooth and you get a very similar looking waveform.

In addition, the input distortion of a filter *can* (as opposed to "always does") muffle the spectrum. While it's true that for a guitar a distortion most often *adds* harmonics, for a signal like a pure sawooth most often the opposite is true. And so it comes that the gain stage of a mixer or a VCF contributes to the perceived sound of a VCO when in fact it's something else that's doing it.

However, I hope it's clear that not all oscillators sound the same. Of course they don't. However, companies like Roland, Moog and Oberheim had it sussed out in such perfection, they've really been engineered close to perfection in the 1970ies. Cycle to cycle, yes, but not hour to hour, not even minute to minute.

- Urs

[david.beholder]

Urs, I get it right in that GS test.

The only noticable difference was: Diva is too bright and crispy clean. Can we have something like "dark and muddy" button in settings page?

[Howard]

Can we have something like "dark and muddy" button in settings page?

[Urs]

The only noticable difference was: Diva is too bright and crispy clean. Can we have something like "dark and muddy" button in settings page?

Hah! You don't need a "dark and muddy" button. You need Satin.

sorry couldn't resist

[SWAN808]

hi Urs

Thats interesting. What you said re cycle by cycle reminded me about a chat I had with a analogue developer. I was wondering if you had found the same with Diva?

I analyzed waveforms and transistor specifications from original Oberheim discrete & Curtis Chip VCO schematics. In fact, the Field-Effect Transistors (FETs) which trigger the reset of each saw wave cycle are imperfect, and have approximately 1% variance from one oscillation to the next. Therefore, while one "saw blade" is 440hz, the next may be 442.8 or 438.5, etc. These minute variations are present even in a VCO with a perfectly stable exponential voltage converter (exp converters are typically the source of instability and 'drift'), but they average out over dozens of oscillations to sound like a consistent frequency.

Nevertheless, this adds a very noticeable sonic quality somewhat reminiscient of "fuzz" or "warmth" due to the added variance.

Moreover, this effect is very different for different frequencies due to the non-linear characteristics of the JFETs — just like a real guitar, piano, trumpet, or saxophone, each note & octave has a distinct harmonic content.

Many DCOs add 'slop' or 'synthetic drift' to attain a VCO sound, but this only imitates the long-term (>1/2 second) inaccuracy of the exponential converter, and doesn't come close to the beautiful per-oscillation variations of VCOs.

Filters are undeniably crucial to the analog sound, but I have methodically and experimentally found VC Oscillators to be just as critical.

[Urs]

Interesting! That's news to me, we have never analysed the OB-X or SEM discrete VCOs, nor did we perceive this with the CEM3340s. One percent is quite a bit... it'll sound much like noise modulation, no?

[Gamma-UT]

Followed by "it's the wrong sort of sound, post some other sound that I'll nail instantly [except I won't]"

But it is. Look, I'm not an analog purist, but, we don't need zero delay filters for many sounds.

It was more a reflection on the people who always turn up in threads saying "it's the wrong sound" who often disappear when the supposedly right sound comes up. They are often the same people who declare that the difference between analogue and simulation are night and day – up to the point where they are asked to choose blind.

[SWAN808]

Interesting! That's news to me, we have never analysed the OB-X or SEM discrete VCOs, nor did we perceive this with the CEM3340s. One percent is quite a bit... it'll sound much like noise modulation, no?

yeah it is quite interesting...I was wondering if that perhaps was why people percieve DCO to be a bit harsher/harder...perhaps even digital models also? Is this something you could verify with all your clever gear?