PG8X (inspired by the JX8P): new beta version uploaded

[ENV1]

You say, that OSC1 is affected by the CROSS MOD settings, even if you set the levels of OSC2 to 0?

No, it was the octave slider of Osc1 that would fade Osc2 to 0 volume. (I have just reproduced it.)

Check this audio and the picture below for the settings. What you hear in the clip happens when you move the Osc1 range slider to 8' and above.

Osc1 volume=0, Osc2 volume=99.

I think Roland found quite a sweet spot for those filters.

And so did you. Yes, there are certainly cases where a ZDF filter can give very good results, but all in all i think they are hopelessly overrated because usually they sound utterly vitreous at high resonance and that kind of sound just makes me want to run outside and.....kill somebody. One exception to what seems to be more or less the rule would perhaps be MauSynth, which actually does sound rather good even at very high resonance. But that one has other issues in that department that need urgent fixing.

Non clicking envelopes are easy, as they are (as many people will say) quite slow, and not as punchy as other synths.

Punchy enough for my taste. If a synth can do the Let the Music Play bassline as good as yours and without any clicking (and you can hear how good it is in the clips) im a happy camper because it means that i will have no problems with 16th sequences no matter what other basslines i might want to sequence. Thats all i could ever ask for.

Well, the hardware does not have it, but of course, you can get PWM type sounds by using a square wave on OSC2, SYNC it to OSC1 and then modulate the pitch of OSC2.

I know. (Thats how i stumbled across the fadeout thing.) But if you want to recreate basses like in Let the Music Play (which was made on a 303) you cant use 2 oscs or you will get phasing that ruins the character. What you need here is a single square osc, set to circa 60 (or 40) percent duty cycle, i.e. just slightly off symmetrical. Only then will it sound right.

There are quite some interesting developments on the hardware side with an firmware update (plus some extra hardware) for the MKS70, which introduced PWM and adds some extra envelopes. I might include that in an extended version of the planned MLS-70, but it is not yet an ongoing project.

Just though id suggest it since the synth would profit so much from it. There is something about that squarewave i find intoxicating, and im a huge fan of PWM sounds, so of course i would love to be able to make them with PG-8X.

[EvilDragon]

Yes, there are certainly cases where a ZDF filter can give very good results, but all in all i think they are hopelessly overrated because usually they sound utterly vitreous at high resonance and that kind of sound just makes me want to run outside and.....kill somebody.

Wow, isn't that harsh a bit? All ZDF filters I've checked out so far (Diva, TAL, Oxium, Monark, The Drop) behave much better, sound sweeter and more precise at high resonance than any other non-ZDF filter, really. It's not even a contest.

[martin_l]

Yes, there are certainly cases where a ZDF filter can give very good results, but all in all i think they are hopelessly overrated because usually they sound utterly vitreous at high resonance and that kind of sound just makes me want to run outside and.....kill somebody.

Wow, isn't that harsh a bit? All ZDF filters I've checked out so far (Diva, TAL, Oxium, Monark, The Drop) behave much better, sound sweeter and more precise at high resonance than any other non-ZDF filter, really. It's not even a contest.

About ZDF (and I definitely do not want to derail this thread...): In the many threads about ZDF, it seems to be clear, that for filters without non linearities, and with non audio-rate frequency modulation, as here in the PG8X, the bi-linear transform is perfectly fine. ZDF become important for non-linearities or fast frequency modulation.

[AusDisciple]

Wow, isn't that harsh a bit? All ZDF filters I've checked out so far (Diva, TAL, Oxium, Monark, The Drop) behave much better, sound sweeter and more precise at high resonance than any other non-ZDF filter, really. It's not even a contest.

I haven't read all 99 pages of this thread!!!... but I just wanted to chime in here and say I totally agree with you on that ED.
For example, DIVA's Roland modeled filters sound every bit as good as the filters on my Roland analogs (SH09, System 100, Juno6, Juno106, JX3P).

[ENV1]

Wow, isn't that harsh a bit? All ZDF filters I've checked out so far (Diva, TAL, Oxium, Monark, The Drop) behave much better, sound sweeter and more precise at high resonance than any other non-ZDF filter, really. It's not even a contest.

Matter of preference, as with all things.

As i said, there certainly are things about a ZDFF that can be considered superior and i understand that. And in certain situations this sure can help to make a sound a little better than with a 'regular' software filter, i understand that too. But in general i just dont like the character they (or most of them anyway) exhibit at very high resonance, so personally i can live just fine without them. A good filter as the one in the PG-8X is all i need to be happy.

[martin_l]

You say, that OSC1 is affected by the CROSS MOD settings, even if you set the levels of OSC2 to 0?

No, it was the octave slider of Osc1 that would fade Osc2 to 0 volume. (I have just reproduced it.)

Check this audio and the picture below for the settings. What you hear in the clip happens when you move the Osc1 range slider to 8' and above.

Osc1 volume=0, Osc2 volume=99.

I can't see the pictures here at work, and I don't have the plugin at hand. But I guess what you find is the normal behaviour:

If you have Oscillator 2 with a square wave at (say) 8'', and you sent Oscillator 1 to 4'', then you reset Oscillator 2 before the square can reach the low phase. Hence, the signal vanishes. You basically have driven the pulsewidth below 0.

Even if this behaviour sounds odd, it is the normal behaviour of hard sync.

[ENV1]

If you have Oscillator 2 with a square wave at (say) 8'', and you sent Oscillator 1 to 4'', then you reset Oscillator 2 before the square can reach the low phase. Hence, the signal vanishes. You basically have driven the pulsewidth below 0.

That is of course very possible.

But shouldnt the signal vanish instantly then? Because thats not what happens, as you can hear in the clip the higher octave(s) can be heard for a short moment before the volume of Osc2 is softly fading down to 0.


Heres the picture attached via the forum:

8x.png

[AdmiralQuality]

The real JX-8P's envelopes actually click. Particularly when you apply a lot of Dynamics to the VCF or VCA, and re-trigger the same note but at a substantially different velocity. The change in velocity scaling causes a sudden jump in the envelope. A bug that's a feature.

[martin_l]

The real JX-8P's envelopes actually click. Particularly when you apply a lot of Dynamics to the VCF or VCA, and re-trigger the same note but at a substantially different velocity. The change in velocity scaling causes a sudden jump in the envelope. A bug that's a feature.

Actually, no. They don't click. Believe me, I have that thing sitting next to me, and just tried it.
Also, if you get yourself a service manual, which has the circuits of the voice boards in it, you will see that there is a low pass filter between the demuxer and the VCA input of the IR3R05. Without that, you would always hear the zipper noise, as the envelopes are software generated, an only updated every 5ms. Hence the slow envelopes and the slightly sloppy timing of the JX8P. Also the velocity scaling is done in software, and these jumps go through the same filter.

[AdmiralQuality]

The real JX-8P's envelopes actually click. Particularly when you apply a lot of Dynamics to the VCF or VCA, and re-trigger the same note but at a substantially different velocity. The change in velocity scaling causes a sudden jump in the envelope. A bug that's a feature.

Actually, no. They don't click. Believe me, I have that thing sitting next to me, and just tried it.
Also, if you get yourself a service manual, which has the circuits of the voice boards in it, you will see that there is a low pass filter between the demuxer and the VCA input of the IR3R05. Without that, you would always hear the zipper noise, as the envelopes are software generated, an only updated every 5ms. Hence the slow envelopes and the slightly sloppy timing of the JX8P. Also the velocity scaling is done in software, and these jumps go through the same filter.

I'm sure they do, Martin. I remember it well from when I owned one.

On the real JX-8P, make a sound with maximum sustain level, long release time and maximum VCA dynamics. Play a note HARD. Release it and let it enter its long release stage. Now immediately play the same note (in regular Poly1 mode, not in cycle assignment mode) but VERY softly. Hear the click when the new note starts?

I solved that in Poly-Ana by always having the attack stage start from the current position of the envelope (which yes, can actually make an attack that goes DOWN in cases like I described above). But of course that's not a feature of the JX-8P.

[AdmiralQuality]

(And maybe "click" is too strong a word to describe it. But it is a jarring discontinuity. Used to drive me nuts.)

[martin_l]

The real JX-8P's envelopes actually click. Particularly when you apply a lot of Dynamics to the VCF or VCA, and re-trigger the same note but at a substantially different velocity. The change in velocity scaling causes a sudden jump in the envelope. A bug that's a feature.

Actually, no. They don't click. Believe me, I have that thing sitting next to me, and just tried it.
Also, if you get yourself a service manual, which has the circuits of the voice boards in it, you will see that there is a low pass filter between the demuxer and the VCA input of the IR3R05. Without that, you would always hear the zipper noise, as the envelopes are software generated, an only updated every 5ms. Hence the slow envelopes and the slightly sloppy timing of the JX8P. Also the velocity scaling is done in software, and these jumps go through the same filter.

I'm sure they do, Martin. I remember it well from when I owned one.

Make a sound with a long release time and maximum VCA dynamics. Play a note HARD. Release it and let it enter its long release stage. Now immediately play the same note (in regular Poly1 mode, not in cycle assignment mode) but VERY softly. Hear the click when the new note starts?

I solved that in Poly-Ana by always having the attack stage start from the current position of the envelope (which yes, can actually make an attack that goes DOWN in cases like I described above). But of course that's not a feature of the JX-8P.

That's what I just did. And they don't click. At least, I would not consider that a click. You do hear an abrupt change. I

You can get a slightly more clicking if you use the DCO2 ENV MIXER with strong dynamics, instead of the VCA. This mixer input is not smoothed but goes straight into a transistor, controlling the level of the oscillator output.

[AdmiralQuality]

What about the VCF envelope input? It could be that I'm remembering.

I'm quite sure on this. But without a JX-8P handy I can't demonstrate it and when I tried to reproduce it with PG-8X I couldn't.

[martin_l]

If you have Oscillator 2 with a square wave at (say) 8'', and you sent Oscillator 1 to 4'', then you reset Oscillator 2 before the square can reach the low phase. Hence, the signal vanishes. You basically have driven the pulsewidth below 0.

That is of course very possible.

But shouldnt the signal vanish instantly then? Because thats not what happens, as you can hear in the clip the higher octave(s) can be heard for a short moment before the volume of Osc2 is softly fading down to 0.


Heres the picture attached via the forum:

8x.png

No, the signal does not vanish instantly. The reason for that is, that there is a DC killer (a low frequency high pass filter) between the raw oscillators and the ring modulator.

The signal of DCO2 is a pulse with varying duty cycle. In your extreme case, it goes from a normal square to a constant 1 if DCO1 is at 4' and DCO2 at 8', i.e. DCO2 generates only a DC signal, which then slowly decays to zero. That slow decay then multiplies the amplitude of DCO1, giving rise to that fading out. If you switch to SYNC1 instead, you will notice that the signal vanishes instantly.

I just tried your example on my hardware, and it does the same fading out for SYNC2, and none for SYNC1.

There is one situation, which I am currently getting wrong: If you have 0 release and play very short notes, this fading will take longer, as the voices might go to sleep in between, while the real DCO's, of course, run constantly. But, that situation is quite rare, and I don't think it justifies ditching the CPU reduction of shutting down silent voices.

I might look into a mechanism to prevent them from sleeping for the duration of that level adjustment. However, the effect is only ever audible when ring modulation is on.

[martin_l]

What about the VCF envelope input? It could be that I'm remembering.

I'm quite sure on this. But without a JX-8P handy I can't demonstrate it and when I tried to reproduce it with PG-8X I couldn't.

You do hear some discontinuity, a short zapping of the filter.

You are correct that the PG-8X currently does not do that. I just noticed that I did not LP filter the frequency before sending to the filter. It looks like the hardware has a similar capacitor as for the VCA.