[Resolved in 2.05B4] Pitchbend response?

[cube48]

Hi,

I've built my second PreenFM2 kit, this time no single issue. Thanks to Xavier and Van Daal Electronics!

I'm using it with Linnstrument which allows Pitchbends up/down to +24/-24 semitones. My finding is that PreenFM2 has non-linear Pitchbend response. It bends down much faster then bending up from center position.

My setting in Modulation matrix:
Pitchbend > 10.00 > o*Fq

Linnstrument Pitchbend setting:
+-12 semitones (I've tried 10 semitones and other values too but not getting me any closer)

I still have to experiment more with this. Could it be that the operator's frequency (PB * Matrix multiplier) just doesn't correspond to any musical intervals (i.e. notes)?

Is there a way to set the Pitchbend response to a standard musical range in semitones?

How exactly is the Pitchbend value coming from the controller used in the Modulation matrix formula?

Thanks a lot in advance for any insights.

P.S.: PreenFM2 already is awesome with expressive possibilities of Linnstrument! FM comes alive under your fingers (Velocity+ModWheel + Aftertouch on every single pad!)

[Xavier]

Hi,

The bitchbend is a midi value not linked to any frequency and is interpreted by the synth.
So i don't really understand what the Linnstrument (midi controller) change when you specify the number of semitones for the pitchbend.

But what you say is intersting and true.
The preenFM2 pitchbend is different from the standard pitchbend effect. And maybe in a not-so-good way.
That's something i'd like to work on (when i have free time).

Best,

Xavier

[cube48]

Xavier, thanks for your reply.
I understand that Pitchband is specific MIDI control type with higher resolution then regular CC's. Different synths have different maximum PB ranges. Some have only 2 semitones, others up to 24 semitones PB range or more, often definable as a part of individual patch/sound. Linnstrument has no PB wheel but you can slide horizontally with your fingers over several pads, generating standard PB control messages. Linnstrument allows you to define the maximum PB range and if it corresponds to the synth range setting you'll get consistent PB response - 1 pad = 1 semitone so you can play continuous PB slides chromatically.
I was kind of suspecting that the PB response of PFM2 is not a simple affair as it is in the very nature of FM to produce inharmonic frequencies easily.
Thanks in advance for any effort!

[cube48]

Xavier, I know it's not a lot but I would pay 50 EUR for just for getting the Pitchbend properly implemented on PreenFM2. Would that motivate you to spend some effort on this? Thanks for all your great work!

[Xavier]

Xavier, I know it's not a lot but I would pay 50 EUR for just for getting the Pitchbend properly implemented on PreenFM2. Would that motivate you to spend some effort on this? Thanks for all your great work!

That's very nice of you. But you can keep your money 
I'm trying to find a solution that doesn't kill the performance.

[lokki]

the usual way is something like this...
bendfactor = ((bendrange * 100.0) / 8192.0);
bendrange would be the semitones that you want to bend up and down

cents = bend * bendfactor;
bend is 14 bit between -8192 and 8192

Final Frequency = Frequency * 2^(cents/1200);

for a bendrange of 12 this would give you a factor of 0.5 to 2.

i'm sure you already figured out that much though :-) the difficult part will be to implement this with pm, right? do you have to shift every single oscillator?

[Xavier]

I have something working.
I thought all the frequency modifications that come from the matrix should modify the frequency that way.
Whatever source you send to the o*fr target, the matrix multiplier is now the number of tone you want.
lfo1 > 0.5 > o*fr : 1 semitone oscillation around the main frequency
PitB > 6 > o*fr : The pitchbend shift one octave up and down.

"2.05 Beta1" firmware attached,
Let me know if you find any problem.

Xavier

[lokki]

thanks xavier, this is great news!! i will try tomorrow and let you know. may i add that the current o*fr modulation implementation can be VERY nice for modwheel. you can get some stop and go sounds from it (going all the way down to zero hertz and up again). if there is a possibility to choose the mod behaviour, that would be fantastic. pitchbend of course should always be "tone/note" dependant.

[lokki]

i works as expected! cool. one thing i noticed is that it is quite "steppy", when i bend faster. i have set the pitchbend to +- one octave and it is really noticeable, sounds like it's running a chromatic scale :-)

[lokki]

actually scratch that! it works perfectly fine!!! I was using one of my very cheap controllers, which only outputs 7-bit pitch bend (high nibble). sorry for the noise..

and also, nice smoothing of the mod wheel.

great, thanks Xavier!

the only thing now is, that it would be nice to get the old behavior back as an option for everything but pitch wheel...

[lokki]

one other suggestion, if you don't mind

standard on synths is +-24 semitones maximum pitchbend range. so the modmatrix scale should either go to 12, or a halfstep should only be 0.4 :-)

[Xavier]

Thanks for your feedback Lokki. 

What you can do is to use 2 matrix rows with PB as source, they will be added : 10 + 2 = 12.
You can even do : 10 + 8 = 18 = +-36 semitones 
(I'm expected it to crash after 51 semitones as there's no checking for reading outside the exp2 array in this Beta)

I may have an idea in the following days, but for the moment, i have no solution to make the harmonique shifting for the PB only.
I'll  add an option but that will be ALL or NONE.

Xavier

[lokki]

ah neat! i did not know, that two sections would add up. cool. ALL or NONE is already quite good i think. will it be per patch?

[Naive Teen Idol]

My finding is that PreenFM2 has non-linear Pitchbend response. It bends down much faster then bending up from center position.

Hi Xavier,

I wasn't an owner back when this thread was started, but I'm finding that this odd pitch bend response is back in the v2.12B. When I put Pitchbend at +10, it does a perfect octave up but then so far down it's actually out of audible range.

Is there any way to fix this going forward? Thanks so much.

[Xavier]

Make sure you use "O*Fh" and not "O*Fq" with the latest firmware.
The first one should apply a "symetrical tone pitch bend".

If you still have a problem, please try to explain that clearly.
What do you use ?
What you are doing + settings ?
What is the behaviour and what would be your expected behaviour ?
Thanks