Few observations

[kuzma_p]

First of all. I've found some quirks and wanna know, does this behavior is something intended to be.

1. When i change algorithms and available voice numbers in instruments overall loudness jumps significantly. Yes, 1-carrier Algo sounds louder than 2- or 3-carrier. And 1-voice instrument sounds even louder. Don't you think that this needs to be compensated or fine-tuned somehow? Nevertheless, this does not cause much inconvenience.

2. When i modulate OP1 by OP2 everything sounds fine, but when i connect OP3 to OP2 such way for example: OP3 -> OP2 -> OP1 with any modulation index greater than 0, timbre starts to 'beat' like there is 'detune' somewhere between OP3 and OP2. This kinda differs from 'classic'  FM behavior where modulating OPs with exact tuning produces static tone.

3. When i tweak IMs while holding key on external midi keyboard i get sound with one timbre and when i retrig this note i can get significantly different tone.

[Xavier]

Hi,

Thanks for the feedback.

1. This is how it works. To avoid saturation and to laverage the room in the sample as much as possible for each algorithm.
That's the same with the number total voices used by all the timbres.  The final mix stage divides the final sample by the number of voice so that it fits the 18 bits of the converter.

2. I'm surprise by the >0.. That's true that upper operator cannot have large IM. And it works better with small frequency in the upper op... I'll give a depper look.

3. I need more information on this one. The IM is multiplied by the input operator which has en enveloppe. So the begining of the sound can differ from the end, which is maybe what you hear. There are 48 seconds (16 seconds * (Atk +  dec + sus)) = during which the enveloppe can vary.
Let me know with more details.

[kuzma_p]

3. Ok, i'll try to tweak and record some output today

[kuzma_p]

Then i want to make few suggestions 

4. I'm noticing that KSyn 0.02 in LFO settings producing some delay (or maybe it is placebo?) or feels not fas enough. Maybe you can set 0.01 as really instant LFO retrig? There is use case i'm exploiting alot on SW or HW synths to alter every triggered note, something like S&H: i use random LFO wave with really slow rate (or even 0) which retrigs with new random value each time key is pressed. When played fast with such LFO assigned on IM or Cutoff you get instant random mental acid line, AFX utilizes this trick a lot. Or when such LFO affects Tune slightly you can get something with analog feel.

5. Can you add (up to) X10 modulation multiplier in Tools>Set for someone who need fast sharp blips and zaps when modulating pitch with Envelope? And anyway this can be cool experimental feature.

6. What about introducing 4 additional modulation sources, W/X/Y/Z linked to CCs, which you can set in Tools>Set ? Exact thing is already implemented in Waldorf Blofeld.

7. And some kind of exorbitant feature request / idea:
Macro control for common sound characteristics that is accessible via hold key combination Eng + Op for example.
With this you can control 4 sound characteristics: Attack (simultaneously in respect of already set OPs attacks) , Length (simultaneously in respect of already set OPs decays and releases), Brightness (IMs), Transpose. It is just idea, so this could be discussed.

Thanks for your attention.

--
Xavier's modification: renumbered points

[Xavier]

Seems like you spent time with your Preen 

4. Great trick ! I'l think about a solution. Maybe Adding a "Off" value then starting at 0.00.

5. I 'm not sure i get it. x10 on the LFO env timing ?  You meant /10 on the timing right ? Strange to be in the settings, it's more a preset parameter.

6. It was in the PreenFM1 and i removed it... i'll definitely add it back. CC1, CC2, CC3, CC4.

7. Would that work to be able to access the 4 CC of previous point from the encoder ? You could route CC1 to 4 target in the Matrix.

As you seem to be really inside the engine...
I'll share this week a beta version of the "work in progress" firmware: arpeggiator, IM/velocity per voice, cheap but usefull filter section (LP, HP, Bass boost).

Xavier

[kuzma_p]

5. I 'm not sure i get it. x10 on the LFO env timing ?  You meant /10 on the timing right ? Strange to be in the settings, it's more a preset parameter.

6. It was in the PreenFM1 and i removed it... i'll definitely add it back. CC1, CC2, CC3, CC4.

7. Would that work to be able to access the 4 CC of previous point from the encoder ? You could route CC1 to 4 target in the Matrix.

As you seem to be really inside the engine...
I'll share this week a beta version of the "work in progress" firmware: arpeggiator, IM per voice, cheap but usefull filter section (LP, HP, Bass boost).

Xavier

5. No, i mean multiplication of existing Mult parameter in Modulation Matrix. It can be presented in settings and should be set to X1 by default. But if someone wants to exceed existing -10 ... 10 range, to make extremely deep modulations (for percussion, for example) he can tweak this setting. Yes, this can affect all existing presets, so user should go to settings and set depth back to X1, if he wants to get his common sounds back. X10 can be too much, tho. Something like X1 -X4 is enough i guess.

6+7. Yeah, i think that would be great.

Wow, WIP FW. Sounds interesting, looking forward for this.

Seems like you spent time with your Preen 

Hehe, yeah. Then, i'm really interested in FM so i patiently edit my tx81z from panel sometimes. PFM2's UI is a breeze and fun to edit in compare 

[6581punk]

Just a note, but "classic" FM (ie. Yamaha) is phase modulation not frequency modulation

Preen does FM not PM right?

[Xavier]

Yes the preen is FM... 
Which is why of the reason i cannot implement feedback. It changes the pitch with FM. It works with PM.

[kuzma_p]

Yeah, Blofeld has FM abilities too. You can do 3OP FM pretty well with it. And you can FM Osc by itself so it starts to sound funnily unstable. It could be useful somehow with Blofeld, but in context of PreenFM it is kinda useless.

[Xavier]

Pitch change is expected with FM when 2 or more cascading modulators.
http://www.musicdsp.org/showone.php?id=160

As soon as there's a DC in the modulating signal, there's a pitch shift, which is the case with cascading modulators.
I'll try to add DC filter to see how it sounds but it won't accept the CPU overload at MAX polyphony.

[kuzma_p]

Maybe DC in the modulating signal introduces this 'beating' effect too?

https://www.dropbox.com/s/aldhoj04nw25fcl/PFM2.flac

It is init patch, based on algo11
OP3 -> OP2 -> OP1

I'm increasing IM1 till ~0:28 mark ( IM1= ~2.30) , since this mark i start increasing IM2 and since 0:40 'beating' is audible. From 0:40 to 1:03 i'm increasing IM2. Then i retrig key and get kinda different sound — this is behavior from #3 of my initial post.

This is how TX81z behaves with similar OPs setup:
https://www.dropbox.com/s/lmnqj3zlmz4cu6f/tx81z.flac

Amazing:
For TX81z i have to retrig key for IM to take effect, but i can tweak ratios in realtime. And for PFM2 vice versa — ratio tweaking is non-realtime but IMs tweaking is.

[kuzma_p]

Found graphics bug:
Algorithm diagram on PFM2 doesn't show OP5  for Alg26

[Xavier]

Fantastic day for the PreenFM today.. 
Thanks Kuzma, you really put me on the right tracks.

Adding a DC filter on the right operators  makes a huge difference.
And does not add that much CPU.. 
In the worse case i will have to remove 1Khz to the sampling rate.

And this DO fix the pitch shift that can be small or bigger depending on the parameters:
It does fixes the beating you ear and add a LOT of stability to the sounds.
It makes chords much clearer in many situations.

I Added the DC filter thing in ALGO1 to ALGO9.. And not the other, don't try on DX7 that uses ALGO10 and more.

Here is a WIP beta version --
---EDITED : will open beta thread soon --
( Put it in pfm2/ beside the last stable firmware so that you can quickly switch : Power on + Op)

Problem with "szer" oscillator waveform please don't use it.

Please don't report any problem yet with the following but  you will also have :
. velocity IM beside each IM per voice (not per timbre as in the matrix)
. one arpeggiator per Instrument (engine section)
. Filter slot (engine section). Frequency of LP and HP available as destination in the matrix. Try "Bass 0.00 1.00" on bass :-)
. OP5 of ALGO26 now show up ;-)

Xavier

[kuzma_p]

Cool, gonna try this today!

[kuzma_p]

Huh, this filter slot sounds promising indeed.
Looking forward to this firmware.