Tal J-8

[zvenx]

Very interesting.
Thanks for doing this.


i had this convo recently with someone else.

I had brought it up to Patrick when ED first brought it up a few years ago and he said basically it was so slight to the point of negligible..

rsp

I have been comparing the various JP8 emulation filters

Familiar face from GS, hehe!

Wanted to ask you, since you were comparing the various JP8 emulations (and IIRC you owned or still do own a real one), what say you on the complete lack of phase noise in J-8's oscillators? This seems like quite a weird design choice for an emulation of a vintage analog synth...

To showcase this:

1. Load default patch
2. Go to SC panel, set Intensity to minimum and press Random. This will reset all the trimmers to the default
3. Set the VCO-2 Fine Tune knob to default as well

Now play any key repeatedly and every note will sound exactly the same without any phase noise whatsoever.

I don't think this should be possible even if all the trimmers are set to the default. You cannot get any analog VCO to behave like this, ever. And IIRC you cannot get this sort of deadness from either Roland's or Arturia's emulations either (so they got it right at the very least in that regard).

Yes I thought it was weird, but upon testing mine, my earlier JP8A model drifts far less than you think.
I believe Tal's JP8 is a later 14 bit model which has improved tuning accuracy

My Roland Jupiter 8a - warmed up and after autotuning upon boot up.
Dual sawtooth oscillators, with minimal detuning to start.
Even my earlier 12 bit model doesn't drift much, and the later revision drifts even less.

https://soundcloud.com/coolcolj/roland- ... al_sharing

[Funkybot's Evil Twin]

But it's not tuning right...it's phase? You can simulate the audible effects of phase beating by slightly knocking the tuning out slightly in the SC panel, but it's not quite the exact same behavior as fully free-running oscillators where the phase won't be perfectly locked even if the tuning is spot on.

Anyway, it's not something I worry too much about, knocking the OSCs slightly of tune does a good enough job approximating the result, but it's one of those small things where it's about emulating behavior.

[Sawtooth Fairy]

You're missing the point. A vintage analog emulation should always have phase noise in the oscillators. Always! The above scenario should not be possible, trimmers or no trimmers.

I was just playing around with the trimmer panel and noticed that the problem seems to be that as long as you do not play more notes than you specified in the voice selector, the synth assigns the same voice to the same note again and again. Thus there is no pitch variation as you play the same note again and again, ultimately turning the repeated notes into a single long note because of the pick-up mode of the envelopes.

If you exceed the number of allowed voices, though, it assigns another voice to that note and you get the variation.

In Retrologue2 that is rather different. There is only a single pitch randomization knob for the entire synth. But it works fine. I set it to like .5 and the same note sounds slightly different each time I play it, leading to pleasant overlapping with longer release times. Regardless of how many voices I have allowed.

Sometimes less is more...

Use the Jp8 oscillator mode or Round robin, if you want J-8 to cycle through the voices

Where is that setting? Both Poly modes seem to work the same way, no cycling...

[CoolColJ]

Where is that setting? Both Poly modes seem to work the same way, no cycling...

Under Voices, on the far mid left of the synth

But upon testing with the SC panel open, it does not actually round robin like I thought, if you press the same note repeatly.
It will only cycle if you press different notes...

Bringing up the oscilloscope plugin in Reaper, and it does look like the phase of the oscillators are locked if you press the same note. Only on different notes will the phase change, while still using the same voice

[EvilDragon]

Yes I thought it was weird, but upon testing mine, my earlier JP8A model drifts far less than you think.
I believe Tal's JP8 is a later 14 bit model which has improved tuning accuracy

My Roland Jupiter 8a - warmed up and after autotuning upon boot up.
Dual sawtooth oscillators, with minimal detuning to start.
Even my earlier 12 bit model doesn't drift much, and the later revision drifts even less.

https://soundcloud.com/coolcolj/roland- ... al_sharing

Tuning accuracy and drift is NOT the same thing as phase noise, though! Basically TAL's oscillators don't seem to be truly freerunning, as they would be on an actual JP-8...?

[Sawtooth Fairy]

I think free-running only means that they are running all the time, i.e. not started upon playing a note.
Do oscillators have independent "clocks"? I think on the Roland Juno all waveforms are derived from the same signal, hence it sounds rather static, albeit it in a positive way in this case.

[EvilDragon]

Juno is a DCO synth, Jupiter-8 is a VCO synth. VCOs are freerunning all the time.

[CoolColJ]

Yes I thought it was weird, but upon testing mine, my earlier JP8A model drifts far less than you think.
I believe Tal's JP8 is a later 14 bit model which has improved tuning accuracy

My Roland Jupiter 8a - warmed up and after autotuning upon boot up.
Dual sawtooth oscillators, with minimal detuning to start.
Even my earlier 12 bit model doesn't drift much, and the later revision drifts even less.

https://soundcloud.com/coolcolj/roland- ... al_sharing

Tuning accuracy and drift is NOT the same thing as phase noise, though! Basically TAL's oscillators don't seem to be truly freerunning, as they would be on an actual JP-8...?

Tal J-8 oscillators are free running.
I recorded repeated key presses of the same note. And here are the waveforms.
You can see the start of each keypress is different with a few starting on the negative side of the sawtooth

Screenshot 2023-06-30 071952.png

[EvilDragon]

OK, randomized phase then I suppose? But if it is randomized, it is not randomized per oscillator. It seems that the phase offsets between the two oscillators are identical regardless of this (hence, lack of phase noise). You can see the exact same patterns on the waveforms emerging.

[Funkybot's Evil Twin]

The issue, as I reacall from 3 years ago, is that when using OSC1 and OSC2 with no tuning drift applied in the SC panel, both OSCs will be and remain perfectly phase locked with each other. They're pretty much sync'd. As ED points out, that shouldn't happen with free-running VCOs. It seems like there's a random phase that may be triggered on keypress, but the same random phase offset gets applied to both oscillators equally.

EDIT: cross-post.

[CoolColJ]

Waveform from my Dual Sawtooth Jp8 audio clip - it looks about the same.
This is with dual VCOs, whereas the Tal one above is just a single sawtooth

Screenshot 2023-06-30 065834.png

[CoolColJ]

J-8 with poly aftertouch controlling the filter/volume is an amazing thing on Arps, it almost makes it sound like a TB-303 pattern with the fluctuating variance and accent per note filter

Coming soon tm

[perpetual3]

Is it usually for a dual unison patch to completely take up the entire cpu on an i9 MBP?

[zvenx]

J-8 with poly aftertouch controlling the filter/volume is an amazing thing on Arps, it almost makes it sound like a TB-303 pattern with the fluctuating variance and accent per note filter

Coming soon tm

Ahh you have it too..
It is amazing.
Can't wait till Patrick releases it.
I assume it fixes the bugs you reported as well?

And that new _____ makes it so much easier to work with for me...and that you can change the size of the _____... Brilliant.

rsp

[zvenx]

Is it usually for a dual unison patch to completely take up the entire cpu on an i9 MBP?

How many voices you have assigned to that patch? 12? Then yes probably.
rsp