[G2 1.40] Constant Module

[gurk]

There seems to be a bug in the constant module.

The constant's value jumps from 62 to 64 in bipolar mode.

When the constant is set to 64, the module's output isn't exactly 64.

If you use a constant module to tune an oscillator set to -64 back to 0 semitones, and mix it with an oscillator set at 0 semitones instead of getting a 2 x amplitude signal it produces a slow beating effect which suggests that output is not exactly 64.

Here is a patch which demonstrates this.

g

---

constant module bug.pch2
Description:	patch to demonstrate constant module error	Download (listen)
Filename:	constant module bug.pch2
Filesize:	1.29 KB
Downloaded:	1096 Time(s)

---

[mosc]

Yes, this looks like a bug to me too. Funny nobody ever noticed such a problem before...

[jamos]

I've seen that too, but it makes.. some sense.

Since the G2's opprams are stored as 7-bit values, they only have 128 different values. This could be 0 to 127, or -64 to +63. Apparently they hacked around this limitation by replacing +63 with +64.

[varice]

[gurk]

[sebber]

It's got to do with the 127 MIDI numbers, it's Clavia's solution for it. I remember having read about this. If I'm right t's been in the old NM workshop on Clavia's website, there's a workshop about how the NM deals with numbers in general.

[sebber]

Relation between knob positions and units

Any knob can be set to one of 128 positions. This exactly matches the MIDI specification, where e.g. a CC signal, or a note number with its velocity value, can have values from 0 up to 127. As MIDI CC’s are used to control the position of knobs on synths, how does the knobsetting or a CC value, if it’s assigned to a knob, relate to the effect that setting has in a module? Do some experimenting with knobs with different control curves until you fully understand what’s happening.

Note that some knobs that control a value between –64 and 64 miss the value of 63. Test this with a Constant module. If the Uni button is depressed the value of 63 is missing and if the Uni button is pressed the value of 63,5 is missing. This is as there are only 128 possible positions, between –64 and 64 are 127 integer values, adding –64 and 64 to this collection of values would mean 129 knobpositions. But MIDI has only 128 possible knobpositions and so have the module's knobs. Rescaling the –64 up to 64 range to 128 equal values would mean that all values except the –64 and 64 would become fractional numbers and 0 would disappear from the collection. As the value of 0 and 64 are so important (total silence or unity gain), Clavia chose to simply skip the 63 value and use 64 in it’s place. This also means that a knobsetting of 127 must actually be considered a setting of 128 instead of 127 if that knob has a linear behaviour!

To summarise:

It’s not the physical position of the knobs that defines what happens if we scale down an ON state of 64 units but the underlying system that couples the knob position with the arithmetic value to be used in the calculation.
It is important to only use knobs with linear behaviour if we want to have a predictable result. That's why there’s that Lin button on the ControlMixer module. The knobsetting 64 (and not 63) on the control mixer will exactly halve the signal, but the knobsetting 127 will mean unity gain, so although the arithmetic is linear the relation between the knob’s physical position and the underlying arithmetic is slightly irregular, compensating for a small deficiency in the MIDI specification.

http://www.clavia.se/nordmodular/Modularzone/LogicIntro.html

[mosc]

Sigh - the low res nobs are the weakest link in the Nord Modulars, IMHO. It just takes a minute playing an Alesis ION to see the value of higer resolution knobs.

[ian-s]

[gurk]

Ok this explains the jump from 62 to 64. (good OLD 8bit)

What about the movement in the sound as if the two synced oscillators are not perfectly in tune when you mix them together ? (as demonstrated in my patch)

In my opinion (which has changed a bit since reading Varice's post) there must be something going wrong with pitch modulation input when it receives a value of +64, hence the beating effect. Surely this is not part of the design?

I've also found that the beating effect in my patch has a different speed on G2 hardware than when run on the G2 demo software, this suggests to me that they are being tuned differently, this shouldn't be right, should it?

If you try my patch and change the octave shift on both demo and hardware, with the patch attached at the top of the page, you should hear a differing modulation speed between hardware and software.

cheers,
Graeme

[mosc]

[dasz]