Alternative phase invert problem

[wartaler]

Hmm there goes my theory,

I actually never recorded the ND2 to test this, I did test it with ND1 and found that starting phase did vary - however not sure now whether I looked at fully isolated notes - I think I did but...

hmmm then indeed envelope and pitch bend are the only two parameters left, I will have a look this weekend, see if I can find some tricks to reveal differences between hits

Thanks,
Albert

[gzifcak]

I think actually that the square wave does not reset, after a quick test. I've verified that the sine, triangle and t1 waves reset. I haven't looked at the resonance modeling waves.

As for the inconsistencies that seem to be caused by the bend envelopes, I will try using a sine set to the lowest frequency, and maximum bend depth with minimum time, to see if i can see distortion in the waveform.

[gzifcak]

OK more findings (based on listening with very low pitch, very short decay, no punch and no bend, with waveform recording tests to verify in case I couldn't tell by ear)

phase resets at zero for the following waves:
A1: yes
A2: yes
A3: yes
A4: no
A5: no
A6: yes
r1: no
r2: no
t1: yes
f1 to f6 (FM): yes but NOT when TIMBRE is turned up above 0
H1: yes (it sounds like it varies a little with TIMBRE turned up, but no visible phase difference)
H2 to C3: yes but NOT when TIMBRE is turned up above 0

Based on the FM waves, this makes me think that a lot of the waves use multiple oscillators, and that the one that creates the fundamental pitch resets, but that the ones creating the harmonics don't. In my opinion they should (like a DX7 does).

There is still the problem of the inconsistencies that become obvious on all waves with the introduction of the bend envelope. I'm going to do some more tests to try to figure out what this is about.

[gzifcak]

OK first I thought maybe the inconsistency had to do with a lower bandwidth for the bend envelope (called the control rate in the Nord Modular and Reaktor) than for the audio path, but I did a test in Reaktor and even bringing the control rate down to 100hz doesn't create inconsistency, just a duller pop. According to the manual the Nord Modular has a control rate of 24khz, so I doubt bandwidth is the culprit.

I've seen rumors that the Drum engine is built from the G2, does anyone have one they could try to replicate a simple patch on and see if they have similar problems?

[gzifcak]

Here's a recording of the A1 wave set to the lowest pitch and triggered by a very low note at max velocity, with very short decay, maximum bend amount, and minimum bend time.

ND2bendTest.wav

(2.09 MiB) Downloaded 189 times

As you can see, the number of cycles at the beginning of the waveform varies from hit to hit.

Since we know the sine wave (A1) phase resets to zero (still visible here), the bend envelope must either vary in depth, starting at different pitches for each hit, or in time, moving from initial pitch to the ending pitch at varying speeds for each hit.

Either way, hopefully Clavia will address this, as well as the non-resetting waveforms and waveform harmonics, in an update. As I've stated before, percussive strikes in the acoustic realm (as well as in the classic drum synths) always start at a phase position of zero.

Anyone have additional insight or see errors/oversights in my logic?

[gzifcak]

The only reason I'm so concerned about these inconsistencies is that it's just about the perfect drum synth otherwise. If these problems were fixed it would be such a precise machine. I suppose it may not matter as much if you're triggering it with sticks and monitoring through a performance amp with a limited frequency range, but for electronic music styles I think we've all gotten used to accurate repetition.

[wartaler]

Wow, lot of work ! thanks!
That's quite a detailed description!

I was too quick and blunt with my experiments & conclusions, sorry all, -and Clavia!
Your list actually matches with what I hear, I never noticed it with just sine, I was just assuming it was my ears failing to hear it, but apparently no problem with sine. Especially the ring mod sounds r1 and r2 I find quite inconsistent. You could be right about multiple osc sounds. The possibly inconsistent pitch bend behavior is interesting, personally I find inconsistencies here more subtle -by ear.. I use my ND2 mostly live, so for me it's perhaps slightly less crucial, but I do get your point wanting it to bemore consistent.

No idea if multiple osc and pitch bend related sound inconsistencies are deliberate, hardware limitations or OS 'bugs'

Perhaps report it, you can refer to your posts in this thread, it contains all details by now

Thanks,
Albert

[gzifcak]

I would like to report these issues, do you know where I can do that?

[wartaler]

here: http://www.nordkeyboards.com/support

[gzifcak]

Submitted, we'll see what happens.