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

[bM3w]

Lovely synth

Two things I'd like to see fixed whenever possible:

1) the switches should actually act like switches and not faders. For example on DCO Range & DCO Wave Form the switches behave like faders. Although it works it just "feels" weird.

2) The outer ring of the Tune dials have some minor graphic glitches when they turn. Again no big deal, but since everything else seem so perfect you should be able to get this little issue right easily

[martin_l]

Lovely synth

Thanks!

Two things I'd like to see fixed whenever possible:

1) the switches should actually act like switches and not faders. For example on DCO Range & DCO Wave Form the switches behave like faders. Although it works it just "feels" weird.

It is intended since the hardware controller works the same way. In order to show more clearly which setting is selected, I used the colour coding of the labels.

2) The outer ring of the Tune dials have some minor graphic glitches when they turn. Again no big deal, but since everything else seem so perfect you should be able to get this little issue right easily

I know. So far I am still concentrating on the sound and functionality. Graphical improvements might come later, if everyting else works.

Cheers,
Martin

[bM3w]

1) the switches should actually act like switches and not faders. For example on DCO Range & DCO Wave Form the switches behave like faders. Although it works it just "feels" weird.

It is intended since the hardware controller works the same way. In order to show more clearly which setting is selected, I used the colour coding of the labels.

Really? So you can basically have a setting in-between? That's odd, but if you say it's just like on the real thing, there's nothing to complain about

[martin_l]

1) the switches should actually act like switches and not faders. For example on DCO Range & DCO Wave Form the switches behave like faders. Although it works it just "feels" weird.

It is intended since the hardware controller works the same way. In order to show more clearly which setting is selected, I used the colour coding of the labels.

Really? So you can basically have a setting in-between? That's odd, but if you say it's just like on the real thing, there's nothing to complain about

No, there are no settings inbetween. If you imagine the slider giving values from 0-99, then 0-24, 25-49, 50-74 and 75-99 would correspond to the 4 different settings (e.g. the waveforms). You can see this behaviour when you slowly move the slider and watch the labels change colour (on my plugin). In the hardware, you sill see the display change abruptly from SAW to PULSE (for instance) when you move the slider. So, it does not 'morph' between the settigns, but just switch.

[martin_l]

the0.99 did load here in reaper on Win7 64bit,
but irt crashed while changing presets,
and now it doesn't load again,
I get a runtime error and programm termination
same in XT2 here

might need to reboot my system,

but this shouldn't happen imho

Are your crashes similar to the ones described above by s_t?

I just installed the demo of Reaper 3.75 on my XP SP3 laptop, and found no problems. It loaded, played and also did not crash on patch change.

It looks like the problems occur only on 64-bit machines. I don't know how reaper is treating 32-bit plugins in a 64-bit environment. Does it use some sort of internal 32->64 bridge?

Regards,
Martin

[AdmiralQuality]

It looks like the problems occur only on 64-bit machines. I don't know how reaper is treating 32-bit plugins in a 64-bit environment. Does it use some sort of internal 32->64 bridge?

Regards,
Martin

It does, but the fellow above said he was using 32 bit Reaper on x64.

[martin_l]

I had a look through the latest posts again, to see whether there is any pattern in the crash reports. Here is a quick summary:

• s_t: 32-bit Reaper on 64-bit Win7: Crash
  wwjd: 32-bit Reaper on 32-bit Win7: OK
  32-bit Reaper on 32-bit XP3: OK
  Tale: 32-bit Reaper on 32-bit XP3: OK
  edteneyck: 64-bit Reaper on 64-bit Vista: OK
  fateamenabletochange: 64-bit Reaper on 64-bit Win7: OK
  rsmus7: ??-bit Reaper on 64-bit Win7: Crash
  32-bit XT2 on 64-bit Win7: Crash
  martin_l: 32-bit Reaper on 32-bit XP3: OK

So far, the clear cases for a crash were running a 32-bit host on a 64-bit systems, while running a "bit-consistant" setup seems to be fine.

wwjd and rsmus7: Could you, please, let me know more details about your setup?

If my suspition is correct, it might help to narrow down the source of the problem.

Another cause or trouble _could_ be the asio4all drivers. On my laptop, this driver immediately crashed both Reaper and FL Studio demos. Reaper even gave me an imediate blue screen when starting the audio engine.

I wonder whether asio4all could also be involved in other crashes. Therefore, I would also be interested in whether anybody is using asio4all.

In general, please, give me as much information as possible, when you report a bug, because only those infos can give me a clue as to where to look for the problem.

Cheers,
Martin

[wwjd]

Ok, test of brand new Lenovo laptop, a few lame business apps installed, Windows 7 64 bit, tried 32 bit Reaper on it, all presets, all voices, no problems at all for about 10 minutes messing with it.

PLatforms I've tried without issue:

winxp 32bit sp3, reaper 32bit
win7 32bit, reaper 32bit
win7 64bit, reaper 32bit

[martin_l]

Ok, test of brand new Lenovo laptop, a few lame business apps installed, Windows 7 64 bit, tried 32 bit Reaper on it, all presets, all voices, no problems at all for about 10 minutes messing with it.

Thanks for trying that. Did you install asio4all, or did you just use the native Windows sound engine?

Martin

[wwjd]

just the default. I normally do use asio though

[Evan]

martin, I use ASIO4ALL on my laptop (W7 64bit OS), with Reaper 32bit... it's completely stable, no crashes at all. I have not tried your plugin yet! But this setup has been stable for me with other plugins.

[s_t]

I'm currently on vacation, so I can't do any further testing at the moment. I'm using RME HDSP9632 audio interface. The sample rate has been 44.1k all the time.
I'd like to see more reports from similar systems too (W7 x64, Reaper 32). It should be noted also that There was a short period when it worked fine in my system. So it could be possible that if there is a problem within this synth, it can appear randomly with other users as well. Yet to be seen, though.

older versions of pg8 worked fine too. You said you started using new version synth edit with the 0.99? Could that be something to investigate?
Synth edit plugins have caused relatively very little troubles to me in the past.

cheers

[brok landers]

martin_l:

ok i've only briefly tried this, but from the first impressions this thing really does sound like a j8xp! though i can only judge from memory, i sold my jx8p 1996 and my jx10 with the pg800 1998, so my memory hopefully serves me right. but if i had to judge your synth without knowing that it's a jx8p emu, just from sound, i would have had guessed jx8p right away...

so, good job, really good job!
only thing i don't really understand why you limit it to 6 voices only, just for the sake of emulation... as i understood it, you use synthedit only for the gui, so i would, if dare to suggest, make the voice count freely adjustable in steps of 1 up to at least 12 voices, which is what is enough even when you play pads. i always hated this on the jx8p and i bought the jx10 just because of this reason...

oh, and i know, you've involved a lot of oversampling in order to avoid aliasing from the crossmod/osc sync/filter (though, as the filter isn't really saturating and the resonance is tamed from above ca10khz, as well as the resonance isn't so loud anyway, you could perhaps not oversample the filter at all), but could you maybe optimize the synth in terms of cpu usage? even if i set the oversampling to quality 1 and downsampler stages 2, it still uses a _lot_ of cpu.
are you calculating the lfo and the env's sample accurate? i'm asking only because the env's as well as the lfo of the jx8p were slow anyway, so every 64 samples would perhaps not only do, but also come closer to the rather crappy env's the jx8p had...
i'm just shooting inot the dark here though...


however, this is really impressive! great work!

[fateamenabletochange]

alot of the crashes might be Reaper ?? ....Reaper is unpredictable with some stuff, like the Korg vsti's, different peeps getting different results with similar setups.

[martin_l]

martin_l:

ok i've only briefly tried this, but from the first impressions this thing really does sound like a j8xp! though i can only judge from memory, i sold my jx8p 1996 and my jx10 with the pg800 1998, so my memory hopefully serves me right. but if i had to judge your synth without knowing that it's a jx8p emu, just from sound, i would have had guessed jx8p right away...

so, good job, really good job!

Thanks for the kind words.

only thing i don't really understand why you limit it to 6 voices only, just for the sake of emulation... as i understood it, you use synthedit only for the gui, so i would, if dare to suggest, make the voice count freely adjustable in steps of 1 up to at least 12 voices, which is what is enough even when you play pads. i always hated this on the jx8p and i bought the jx10 just because of this reason...

So far, I did not add more voices because of the high CPU load. I am still thinking about making 8 or 10 voices an option. The reason which stops me right now is that I have to find some clever way to determine the unison detuning values for arbitrary number of voices.

Since I am not using the SE polyphony, but wrote my own voice allocator, unfortunately it's not quite as easy as to add a control for the polyphony. In fact, if I used the SE mechanism, it would be as easy as that...

oh, and i know, you've involved a lot of oversampling in order to avoid aliasing from the crossmod/osc sync/filter (though, as the filter isn't really saturating and the resonance is tamed from above ca10khz, as well as the resonance isn't so loud anyway, you could perhaps not oversample the filter at all), but could you maybe optimize the synth in terms of cpu usage? even if i set the oversampling to quality 1 and downsampler stages 2, it still uses a _lot_ of cpu.

Well, I first started out without oversampling, but the standard state variable filter is only stable up to about Nyquist/3. Limiting the filter frequencies to that value was clearly audible at 44kHz sample rate. In order to be able to open the filter wide enough, I had to go to 4-times oversampling.

The 'secondary' oversampling for the oscillators is necessary for getting fine enough frequency resolution for high pitch notes. That's why I left it as a user parameter.

are you calculating the lfo and the env's sample accurate? i'm asking only because the env's as well as the lfo of the jx8p were slow anyway, so every 64 samples would perhaps not only do, but also come closer to the rather crappy env's the jx8p had...

While the LFO itself is just a standard SE oscillator (which is quite cheap, and there is only one in the synth), the envelopes are recalculated only every 5 ms (that's the resolution I found on the JX8P). The ramp increment of the oscillators, and the filter parameters are only calculated every 10 samples (with respect to the sampling rate, not the oversampling rate). Reducing it further did not give me any more speedup. The oscillators and the filters are already written in one module, which is coded in assembler, using SSE instructions, so that the two oscillators are calculated simultaneously. I already thought very long and hard about where I could squeeze out another clock cycle, but at least with the current general algorithm, I can't find anythings else to optimize.

The problem really is the generation of the peculiar pulse waveform, and also the hard sync and the cross modulation (there seems to be a high pass to eliminate the DC component before the cross modulator).

i'm just shooting inot the dark here though...

however, this is really impressive! great work!

Thanks...