Improved Drift

I love that the Super 6 Drift feature works across a number of parameters: tuning, envelopes, filter, but it’s useable range has two major issues:

1. Improved scaling: The sweet spot for Drift mode is only ~5% of the fader. I wish the scaling of this fader could be improved.

2. Improved key tracking High notes seem to drift more than low notes. If I want big, fat, phasey low octave notes, the amount of drift necessary makes chords up the keyboard so out of tune that it is unplayable. I wish that the key tracking were reversed, so that I could have vintage sounds all over the keyboard

4 Likes

I think what we need here is the drift parameter to modulate the osc. frequency in a linear manner. This would have the effect of making bass more detuned than high notes (ie constant beat frequency). This would have the secondary effect that more of the fader travel would be “sweet” as the obvious pitch detune would be less prevalent. It’s a good observation thanks

10 Likes

RE keytracking,I think that would be a perfect fix. I would also significantly vote for a larger parameter range. As much as I love getting sort of a broken drift sound, it would be cool to scale the parameter differently on the fader so that more real estate is in the subtler territory.

4 Likes

I asked for this before too, when Drift first arrived, would love more range in the subtle area.

1 Like

Amazing! Thank you for the reply. Hope this can make it in a future firmware update. As it stands now I can get close to some big beautiful analog sounds, but I’ve never been able to get the low end right on big chords. This change would be super meaningful and game changing!

2 Likes

Maybe a good idea would be introducing smt like note/key modulation as in REV2 - you could then choose tracking direction with negative value. And again, can we possibly have NOTE number as a modulation source on keyboard model?

5 Likes

I like how Sequential made their “Vintage Knob” in their later synths. I have had alot of old VCO polys, and none of them really drift like S6 drift does, or those "slop* synths.
It’s more about single voices being out of calibration with tune/envelope times/filter, and when playing round robin, you hear it clearly. This is what makes those older Oberheims ect. sound so great, not that they are constantly drifting like some old cassette.

The guy who is the master of this is Creative Spiral here https://www.youtube.com/watch?v=g8nVxciNSr0 he did the algorithm for the 3rd Wave. It picks up on the fact that it’s an elderly circuit on a particular oscillator/envelope/filter combination and that the defect in tolerances will be repeated every 6 notes proportional to round robin or whatever other system of note recycling is used. It is certainly not random. This effect work brilliantly on the 3rd wave and I strongly suspect the same technique is used on the prophet 5/10. This would be optimal on the S6 and SG.

3 Likes

UDO describes their drift as working somewhat in this “Voice Component Modeling” methodology. It randomizes pitch, phase, filter cutoff & frequency, ENV stages, and LFOs. The VCM model doesn’t use randomization, but rather a lookup table per voice and component to create randomizations that mimic vintage analog gear. The 3rd Wave implementation is clearly the most stunning. Drift on the Super 6 definitely needs reconfiguration has a very small useable range in the 30-40% of the slider range. Anything below is unnoticeable, everything above puts your chords firmly out of tune, unpleasantly in random way. Unfortunately, I am never able to fully get the drifty analog sound that I’m hoping for with this current implementation. I am really hoping UDO improve the drift function in the update they are promising.

4 Likes

Agreed, drift/vintage mode Sequential does is much more faithful representation of the “drift”. On Prophet 10, and OB6, it not only sets oscillators to drift, but messes up the envelopes and filter quite audibly. I am struggling to get this effect - it is either too subtle or it goes bonkers after the PW slider has been pulled up above 50%.

I’ve just been listening to the drift on my new Super 6 and, although I’m not too bothered about it not doing vintage voice modelling in the same way as the Prophet 10, I think it might matter sometimes that the drift is not quite randomised in the way the manual suggests. Instead, the parameters seem to track a smooth-shaped LFO (perhaps a slewed random LFO but the effect is ultimately like a modulated sine wave).

This is not the best for pitch drift because you end up with a constantly changing frequency (like the old Sequential slop modes) when a fixed random offset per key press might be preferable, as this would better ensure the subtle chorusing effect we want without having inconsistent changes to beating during sustained notes.

But the effect is perhaps worse on the envelope stages: if you listen to, e.g., variation in decay timings, you hear consecutive sequences of longer decays followed by shorter ones instead of true random values. The effect of drift at 100% is so small on the envelopes that you’re unlikely to hear this in a mix but in principle it doesn’t seem very musical.

I didn’t spend a huge amount of time reverse-engineering it but it did also sound like the lows of the pitch drift coincide with the smallest envelope drift values, while the highs of the pitch drift coincide with the largest envelope drift values, suggesting that all of the modulated parameters uniformly track a single LFO waveshape. I would think for ‘random’ drift that the parameters should be treated independently even if their values are LFO-controlled.

1 Like

If it helps to add some background, drift is a slowly moving random wave. Each parameter that is affected is affected by a random amount and in a random direction. So for example, voice one might go up a few cents while voice 2 goes down in pitch, and voice 3 might be unaffected. Other parameters will correlate, as they are driven by the same random wave, but in random amounts in random directions. Of course one limitation is when the random wave is 0, every modulation drift is also 0. Of course the oscillator mixer, HPF, low pass filter, final vca and envelope amount control VCAs are all analogue, so they will have their own inherent voice to voice (day to day) variation

2 Likes

Thanks, that’s very helpful to know. I shouldn’t make out like there’s only one way of getting drift right - one approach to it is never going to be right for everyone in all situations but it’s helpful to know how it works just to predict what a patch will do, so the information is appreciated :slight_smile:

I really hope this improved drift is coming to V1.0!

1 Like