[sdiy] Waveform mixing - normalization?

rsdio at audiobanshee.com rsdio at audiobanshee.com
Sun May 1 09:18:55 CEST 2016

Normalization does not guarantee that the apparent volume will be the same. Greater harmonic content (square) is perceived as louder than lesser harmonic content (triangle). Besides the sheer simplicity of a circuit that does not correct for levels, it makes sense that the waveforms are not normalized because that would not make them the same level.

Usually, the 10 V rails makes it easy to allow plenty of headroom. Oscillator outputs are often capacitively coupled, which will remove any DC offset, although I'm not sure whether that's common when combining the individual waveform outputs of a single oscillator as opposed to combining multiple oscillators with different frequencies. Although distortion is possible, I think it's expected that the sound designer will use the controls in the mixer section to reduce (or increase) distortion. If the filter can accept 10 Vpp on input, then distortion is unlikely, but some 10 V filters will distort before the input signal reaches 10 Vpp.

Most oscillators generate normalized outputs anyway, don't they? e.g. If they use a 10 V supply then the waveform is 10 Vpp.

My go-to synth is the Pro-One (or Matrix-12). I use the ability to turn individual waveforms on and all, but frankly I never know what to expect when multiple waveforms are enabled. I tend to favor the individual waveforms alone, without combining them, although I do appreciate the option to enable multiple waveforms for additional wave shapes.


On Apr 30, 2016, at 11:23 PM, Mattias Rickardsson <mr at analogue.org> wrote:
> I've never seen any normalization either.
> Can't make up my mind about free oscillator levels & mixing, it seems like the advantage is often lost. In a way I'd prefer normalization and a filter overdrive control at a later stage.
> To be honest I've never really missed waveform mixing in a one-waveform-selection-synth either. It feels like a relic from the dawn of synthesis, an additive timbre shaping method that is rarely very useful or interesting compared to waveform modulation, audio-rate modulation and subtractive shaping. Do you guys like and use waveform mixing?
> Den 26 apr. 2016 2:30 em skrev "Tom Wiltshire" <tom at electricdruid.net>:
>> Depends on the synth. I've never seen any analog hardware that implements compression at the mixing stage, but there are various schools of thought as regards "correct" levels. Some assume that you never want distortion in the signal path and are designed for sufficient headroom with all sources maxed. Others have decided that a bit of "drive" is a positive, and allow you to set the signal level high enough to push following filter stages into distortion. Op-amp distortion in the mixer is generally avoided at all costs (at least, I've never seen/heard an example).
>> On 26 Apr 2016, at 01:38, neil harper <metadata at gmx.com> wrote:
>> > I have a question about synths that allow various waveforms to be mixed together (like the minibrute), instead of a single waveform being chosen (like the minimoog) for each oscillator.
>> >
>> > Do any implement any normalization/compression to keep the resultant waveform at a consistent level?
>> >
>> > I checked the minibrute schematic and do not see this, it just goes into a simple op-amp mixer.
>> >
>> > Do any synths implement normalization/compression, or is the purpose to allow the user to overload the subsequent stages by maxing out all the waveforms in the mixer?

More information about the Synth-diy mailing list