Question about Fine Tune Modulation, DCOs

[creativespiral]

Is there any limitation of the Rev2 DCO architecture that prevents it from being able to have MODULATION of FINE TUNING?   Or is this just left out in the firmware/software layer?

I know we can modulate the OSC Freq and that LFO modulation provides finer resolution (down to 12.5 cents / 25 cents as triangle), but am wondering why Fine Tuning is not offered as a mod destination?   12.5/25 cents is still way too much frequency modulation for many situations.

I've been researching raw VCO and DCO spectral graphs, putting together some analysis, and it seems that there is a high rate frequency jitter that is present in VCOs, but with a total frequency range that is extremely small (ie:  2 cents or less)... While this effect on the fundamental frequency is nearly indecipherable, this high rate, low freq range jitter produces more of a blur in the high order harmonics...  I think if we had Fine Tune modulation as a Mod Destination, we could mimic that same sort of upper harmonic wash - which could produce a perceived softer / warmer sound in the upper frequencies.   Also, besides the VCO mimicry, it would allow more fine tune control of oscillators from a variety of other sources.

Is there some hardware reason why OSC FINE TUNING is not offered as a Mod Destination?   I just realized that Osc Fine Tuning even has MIDI CCs associated (CC 22, CC 25)... so it seems like fine tuning is mostly set to be a destination, but was just left out of the Mod Destination table list for some reason?   

If its just a matter of firmware/software, it would be really great to have Osc Fine Tuning added as a Mod Destination! 

[panic]

Unfortunately I don't know the answer to your question, but surely, an easy way to have finer control over the modulated pitch (be it by finetune as mod destination, or by a rescaled LFO amount) would be handy.
Until then, of course, there are a number of possible workarounds for this problem, however sacrificing mod sources and mod slots. You are very familiar with this synth, so I guess you know the techniques of using mod wheel, env 3 or the gated sequencer to provide LFOs with a smaller range? Otherwise, I would be happy to explain this.
Also, have you tried modulating pitch with small amounts of noise?

[creativespiral]

Hey panic - thanks for the ideas!.. yeah, I'm familiar with the workarounds for smaller modulations... I've been using the mod wheel for testing purposes, but I don't like the idea of giving up performance controls, and I almost always use Env3 for other stuff. 

I hadn't considered sacrificing the gated seq though...   that's a good idea!, as long as not going for an animated patch.   Thanks! 

Still would love to know if Fine Tune Mod Destination is a possibility...  I would think it is, and would be useful if implemented.

Cheers, J

[maxter]

I'm intrigued. Sounds to me like this eventual possibility and approach (hopefully along with lfo slew) could be the best way to mimic VCOs, according to your analysis. Being able to control the rate of jitter and finetuned modulation would even surpass VCOs, in that you'd have a lot more sound character control of the simulated "unstableness". Wow, I REALLY would like to see this implemented. Can't really think of anything with a higher potential of raising the bar and putting an end to the criticisms of the Rev2 being "cold" etc.

Considering all the work you put in, and contributing all these great ideas, Sequential ought to implement it and give you cred, if not bring you aboard! It would obviously be for the better if they hear your ideas, at least!

[MPM]

OSC SLOP ?

[creativespiral]

The OSC SLOP algorithm that DSI/Seq has implemented is very interesting, and great for quickly "roughing up" your oscillators and getting some complex detuning behavior.   I did a bunch of testing with OSC SLOP, and detailed findings in the "Appendix F" thread, if you want to read specifics about what it does... It's definitely a useful feature and from what I can tell some good thought went into its design (it appears to give you the equivalent behavior of three extra free running LFOs working on each each oscillator).

OSC SLOP is applied to all of the oscillators, and provides a relatively slow drift over a medium to large frequency range.   

The VCO behavior that I've been studying is a bit different.   I've been looking at it from a harmonic diagram / spectral analysis standpoint, of a single oscillator at a time.   For every type of waveform (beside a perfect sine wave), there are many harmonic partials that are output along with the fundamental frequency.  VCOs seem to produce a high frequency, low range frequency jitter... at the fundamental, the effect is very mild, but in higher level harmonics, the jitter produces more of a blurred frequency output... its very subtle, because of amplitude falloff, but I suspect this is one of the reasons why many people think that DCOs may sound more harsh/cold... because high order harmonics are too precise and stable.

I'm writing up my findings and will post them soon, along with audio/video showing comparisons.   It may be another type of detuning control that is worth adding along with OSC SLOP in future boards... or at least it might be worth noting how to accomplish it via modulation if you want more of the VCO type of sound.   

OSC SLOP is more for major "virtual aging" of your synth sound to produce noticeable slew and drift of oscillators, whereas the OSC JITTER I'm looking at is more for just softening / washing out high order harmonic content. 

[panic]

You are on a noble quest, but you are not the first to be on it (although you might just be the most thorough). The way you describe your findings (high frequency low amplitude pitch jitter) is what you get when you modulate pitch with noise at a very low amount. I do that quite often, and like the effect, but I doubt you will be satisfied. Also, lots of us already used the downscaled LFOs to modulate pitch of the 2 osc separately in various ways. Try it, it sounds nice, that’s for sure…

Still, the features you request in this topic are useful, I think so too. They would make life easier. However, I also highly doubt that they ever will be implemented. It are not new requests, they were already there in rev 1 times.

(hopefully along with lfo slew)

You do know you can make your own slewed LFO using the gated sequencer? Put it in no gate mode, use the slew parameter provided there, and you can pretty much draw any LFO shape you’d like (not free running though, always key-synced).
Still, I would love to have a slew limiter! I hate the fact that the filter only tracks keynumber, and not actual pitch. I really hate it when I do a very slow glide over several octaves when the filter is keytracking, cutoff jumps right up. I suppose that has not been remedied in the rev 2?

[maxter]

(hopefully along with lfo slew)

You do know you can make your own slewed LFO using the gated sequencer? Put it in no gate mode, use the slew parameter provided there, and you can pretty much draw any LFO shape you’d like (not free running though, always key-synced).

Hmm... In no gate mode, isn't the sequencer completely free running? If so, that would work, sort of, but a minor hassle to set up, and it wouldn't really be random. My personal main reason for wanting LFO slew is for using on the S&H wave shape, which of course would make it more random, and without the stepping.

[guyaguy]

(hopefully along with lfo slew)

You do know you can make your own slewed LFO using the gated sequencer? Put it in no gate mode, use the slew parameter provided there, and you can pretty much draw any LFO shape you’d like (not free running though, always key-synced).

Hmm... In no gate mode, isn't the sequencer completely free running? If so, that would work, sort of, but a minor hassle to set up, and it wouldn't really be random. My personal main reason for wanting LFO slew is for using on the S&H wave shape, which of course would make it more random, and without the stepping.

It’s always controlled by the clock either with divisions or without. No gate just means no gate for the envelopes on each step so you can get more slewed or evolving sounds without triggering the VCA or VCF on each step.

[philroyjenkins]

The OSC SLOP algorithm that DSI/Seq has implemented is very interesting, and great for quickly "roughing up" your oscillators and getting some complex detuning behavior.   I did a bunch of testing with OSC SLOP, and detailed findings in the "Appendix F" thread, if you want to read specifics about what it does... It's definitely a useful feature and from what I can tell some good thought went into its design (it appears to give you the equivalent behavior of three extra free running LFOs working on each each oscillator).

OSC SLOP is applied to all of the oscillators, and provides a relatively slow drift over a medium to large frequency range.   

The VCO behavior that I've been studying is a bit different.   I've been looking at it from a harmonic diagram / spectral analysis standpoint, of a single oscillator at a time.   For every type of waveform (beside a perfect sine wave), there are many harmonic partials that are output along with the fundamental frequency.  VCOs seem to produce a high frequency, low range frequency jitter... at the fundamental, the effect is very mild, but in higher level harmonics, the jitter produces more of a blurred frequency output... its very subtle, because of amplitude falloff, but I suspect this is one of the reasons why many people think that DCOs may sound more harsh/cold... because high order harmonics are too precise and stable.

I'm writing up my findings and will post them soon, along with audio/video showing comparisons.   It may be another type of detuning control that is worth adding along with OSC SLOP in future boards... or at least it might be worth noting how to accomplish it via modulation if you want more of the VCO type of sound.   

OSC SLOP is more for major "virtual aging" of your synth sound to produce noticeable slew and drift of oscillators, whereas the OSC JITTER I'm looking at is more for just softening / washing out high order harmonic content.

Very curious about the effect your describing. You saying it could probably be achieved with lfo control over fine tune?

Just as a proof of concept, do you have any settings with the LFO on osc freq as is (even with several modulations wasted)  as an example of the washing out thing?

Also the guy posting osc slop may have meant as an LFO source. I'm experimenting with that right now and getting some very nice results that are different enough from a simple LFO on pitch.

[creativespiral]

I uploaded a video to YouTube that briefly discusses what I've been experimenting with on the Rev2:
https://www.youtube.com/watch?v=Oy9_ShMl16U

It's a subtle effect due to amplitude falloff of higher harmonics, but it washes out / blurs some of those high frequencies, instead of providing perfectly tuned harmonics.   It may be one of the subtle characteristics that lead to people characterizing DCOs as "colder" or "harsher" than VCOs.   

The use of Gated Sequencer actually provides a great amount of control over the effect, although being able to have FINE TUNE as a MOD DESTINATION would be great to quickly approximate the effect with a single LFO.   

Maybe it would be worth adding as an additional control like OSC SLOP to future products.  As mentioned, OSC SLOP is a quick way to virtually age your synth character with noticeable detuning on all the oscillators, whereas this is more focused on subtly blurring the higher harmonics.

[creativespiral]

Here's a video with another look at the spectral graph, set to FFT Bins rather than Line mode, and changed from Log to Linear scaling...  just a different spectral view of same effect from first video.  (Watch that video first for context)

https://www.youtube.com/watch?v=MbHKgr5LbsI

[panic]

Hmm... In no gate mode, isn't the sequencer completely free running? If so, that would work, sort of, but a minor hassle to set up, and it wouldn't really be random. My personal main reason for wanting LFO slew is for using on the S&H wave shape, which of course would make it more random, and without the stepping.

Ah, yes, there you are right. Having no real smoothed S&H is a pity indeed, and using the gated sequencer will always be less random, and is more of a pain to set up.
My approach for this in the past was the following:
- Program “ad random” your two gated sequences (you only have 2, because adding slew effectively reduces the 4 sequences to 2), with a different length, and assign them to the same destination. This adds greatly to a more random like effect since the sequences loop add a different point, and the summed result will vary more in time.

- Use “no gate, no reset” mode. As said, the sequencer is not free running, it will only start to run when a key is pressed. But “no reset” will make it restart at the point where it was stopped, not at step 1.

A problem might be on patches with a long release. Since the sequencer is key-activated, in theory it stops running during the release phase (can’t remember if it really does, I guess I mostly use it on short release programs).

Edit: and of course, if you have one around, you can always connect an external LFO. I never really tried connecting my cp-251, but it should work (no keysync unfortunately).

[maxter]

Hmm... In no gate mode, isn't the sequencer completely free running? If so, that would work, sort of, but a minor hassle to set up, and it wouldn't really be random. My personal main reason for wanting LFO slew is for using on the S&H wave shape, which of course would make it more random, and without the stepping.

Ah, yes, there you are right. Having no real smoothed S&H is a pity indeed, and using the gated sequencer will always be less random, and is more of a pain to set up.
My approach for this in the past was the following:
- Program “ad random” your two gated sequences (you only have 2, because adding slew effectively reduces the 4 sequences to 2), with a different length, and assign them to the same destination. This adds greatly to a more random like effect since the sequences loop add a different point, and the summed result will vary more in time.

- Use “no gate, no reset” mode. As said, the sequencer is not free running, it will only start to run when a key is pressed. But “no reset” will make it restart at the point where it was stopped, not at step 1.

A problem might be on patches with a long release. Since the sequencer is key-activated, in theory it stops running during the release phase (can’t remember if it really does, I guess I mostly use it on short release programs).

Edit: and of course, if you have one around, you can always connect an external LFO. I never really tried connecting my cp-251, but it should work (no keysync unfortunately).

Great tips, thanks! External LFO doesn't work per voice, unless mono, but otherwise that's the least hassle to set up. I do believe the sequencers don't stop with key release on the no gate no reset mode. That's the mode I was thinking of earlier, got them mixed up. I do believe they're actually free running on no reset mode. On my P'08 I used to put it in key trig mode, trigger the voices with an external sequencer at various synced rhythmic intervals, and once set I'd throw it into no reset mode, for each voice to keep it's intended rhythmic interval when shifting chords or whatever. If I recall correctly. Don't have the opportunity to check this out right now, but will have to do later.

[panic]

By the way, you can also add to the pseudo-random effect by modulating the gated sequencer with things like velocity or keynumber. External LFO is indeed global. God, I love the modulation flexibility of this machine, it's the reason there will always be something of the p8 family in my arsenal, despite the fact I might osc and filter on other synths better.

[creativespiral]

Very curious about the effect your describing. You saying it could probably be achieved with lfo control over fine tune?

Just as a proof of concept, do you have any settings with the LFO on osc freq as is (even with several modulations wasted)  as an example of the washing out thing?

Yeah, if Fine Tuning was a Mod Destination, you could approximate this effect by directly routing a single LFO to Osc Fine Tune.   The key is that you need to be able to swing frequency by very small amounts (a fraction of a cent).   Currently, with a triangle, you're swinging 12.5 cents in each direction (25 cents total range)

The example I posted on the video is a bit more complex, using both triangle and random LFOs, in conjunction with animated gated sequencer and slew.    But IN ITS MOST SIMPLE FORM, an approximation would be to setup the following:

A. LFO 1
Type: Triangle
Freq Value: 50-70 (somewhere in that range)
Destination: OSC Freq
Amount: 0

B: MOD SLOT 1
Source: Gated Sequencer
Destination: LFO 1 Amount
Amount: 1

C: GATED SEQUENCER SETUP (In most basic form, just used to scale Mod Slot 1 Amount)
Set to NO GATE mode, so envelopes don't retrigger
Set Step One to a very small value (like 3), and set Step Two to reset (so it's just always playing Step One)
Set Destination to Nothing (Leave it blank... using the mod matrix to route it instead)

In that simpler setup above, you're just using the Gated Sequencer to scale down the Modulation Amount from '1.000' to '0.023'  (ie: 3/128)... that small fraction is sent along to the LFO via mod matrix, and instead of getting a 12.5 cent minimum modulation (25 cents bi-directionally),   Now you're modulating a total range of just 0.58 cents motion.  (ie: 0.023 * 25)

This ultra fine modulation of less than 1 cent is barely noticeable at the fundamental, but results in motion in higher harmonics, like seen on VCO oscillators.   This example could be reproduced using just a single LFO routed to Fine Tune Freq (if that was an option)... so you wouldn't have to waste an extra mod slot and the gated sequencer slot to scale the data.   

Using the gated sequencer with some animation and slew to it, and adding a second LFO set to random s+h, just allows for fine tuning of this harmonic motion, and adding some glitchiness, as that seems to be present in VCOs, in addition to the undulation generated by the triangle.   

If a dedicated control over this type of fine tune harmonic jitter behavior were to be added in future boards, it would be good to consider combining some triangle/sine motion along with some randomness, and have it applied on a per osc basis, controlled by an overall amount potentiometer...  Similar to how OSC SLOP seems to be the equivalent of several LFOs modulated to each osc separately. 

[guyaguy]

By the way, you can also add to the pseudo-random effect by modulating the gated sequencer with things like velocity or keynumber. External LFO is indeed global. God, I love the modulation flexibility of this machine, it's the reason there will always be something of the p8 family in my arsenal, despite the fact I might osc and filter on other synths better.

Hmm, what do you mean by modulating the gated sequencer?

[panic]

I mean routing a modulation source to affect the amount of the sequencer. I know the sequencers are not mod destinations, but you can work around it:
- leave seq destination blank
- assign the sequencer as source for a modulator in mod matrix, and set your destination there
- now modulate that modulator

[philroyjenkins]

Honestly, try the osc slop as LFO destination if you haven't yet.

I set up the modulation you described with the gated seq and I feel like this is achieving very similar but in a simpler way.

Tri LFO on Osc Slop with an amount of 1 and speed on the 30-40 range. It sounded a little better with key sync on but YMMV. I've also had good results at higher rates like 70-90.

Its basically acting as a very subtle lfo with a little bit of variance across the keys. 

E: I totally agree though, adding fine tune would be a simple but effective addition. (I'd love some deeper lfo options like slew as well.)

[creativespiral]

Yeah, I've tried with the OSC SLOP / LFO modulation as well... that was the first method I tested, but wasn't able to match the same sort of behavior observed with VCOs.

There may be a few reasons:

1. Using the Mod Matrix to OSC SLOP, at a value of 1, you will still get total frequency range of around 6 cents swing maximum.  Even dividing that by four using the LFO, the range may be as high as ~1.5 cents.   The total range that I've observed with VCOs seems to be smaller on average (ie: 0.5 cent total range).   Note: the OSC SLOP algorithm generates a complex motion, so sometimes it will be less, and sometimes more.   This method may generate up to 3x that of what I have observed.

2. This Complex Motion of the OSC SLOP creates more of a random effect, and while my observations of VCOs have shown some randomness/glitchiness, there is a good amount of order to the harmonic motion overall.  ie:  the undulation has a pretty standard motion, but with some slight variation.  This osc slop method produces too much randomness over time.

3. Since this OSC SLOP modulation method is using a LFO to modulate another "Complex LFO macro", there are crossovers between the two where you get no motion... ie: if the osc slop is hovering around / crossing over the 0 point, then the LFO modulating it is going to modulate between 0 and 0.

It's really getting down into fine details... but that's sort of what this is all about...  DCOs produce a sound/character that is 95% of that of VCOs... its just a matter of figuring out that last 5%.

On a side note, I had a breakthrough (related to this, but slightly different), that has totally changed my thinking about VCO synth emulation.   I'm gonna do some more work on it... currently applying it to a bunch of my custom patches.  I'll have some thoughts (and patches) to share soon