Pan Law

[ColinP]

For mixing wet/dry in some effects, I think its a good compromise when you don't want individual wet/dry knobs.

I would argue that it's a poor choice, as is having two knobs for what should be a simple one-dimensional function. Let's look at the gain curve (in green) that results from using CA's algorithm...

WeirdWetDry.png (40.31 KiB) Viewed 1273 times

So rather than the control giving 50% wet + 50% dry at mid-point we end up with a signal that has twice the amplitude of either wet or dry (assuming that wet and dry have more or less the same amplitude and there's no significant phase cancellation).

So we get an undesirable 6 dB increase in volume at mid-point. I don't see how this is useful when we just want to blend wet and dry.

In contrast a linear crossfade results in a gain curve that's flat all the way across the control's travel.

When panning we don't generally use linear curves because the mix isn't happening via perfect summation, rather it's occurring in a complex physical space where our ears, speaker angles and room acoustics have an impact - so there are various compromises with regards to things like the hole in the middle and mono (although mono reproduction is obviously far less relevant in modern times). But in a crossfader we are effectively dealing with an ideal mono mix so linear curves make the most sense,

Although one could also argue for exponential crossfading - which is what we would get when doing a crossfade on a mixing console where the faders have exponential responses. But CA's curves are effectively crudely logarithmic in shape rather than exponential.

[borkman]

This discussion is why I put options in Full Thickness Tear - the one knob for two thing, sine law, and linear.

Opinions are all over the place on this topic. As an example, quickly scan through https://www.kvraudio.com/forum/viewtopic.php?t=426306, and you'll see the opinions: why aren't there always two knobs, why would you ever need two knobs, here's an equation or two, why not add an extra button that does a thing, how about just avoiding the whole issue by doing this and that, I never use mix knobs, thank you so much NI for the 100%:100% approach in Replika, etc. Granted this is about plugins, not modules, (and it's KVR) but I believe the arguments still hold.

It also depends on what you are doing. Are you setting a flanger by ear? Are you automating the knob (or modulating the value)? Are you doing something else? Are you more mathematically focused or more focused on adding a bit of this and that with a twiddle of a knob? It takes all kinds to make this wonderful, diverse world.

In the end, your argument is solid, from your perspective. I'm pretty sure opinions vary quite a bit on this though.

https://forum.qubitelectronix.com/t/mix ... lus/502/24 I just saw this before I hit Submit. I didn't read it all, but it's in the modular synth world, has lots of nice graphs, and has some strong opinions - one person returning the module because of the way the mix knob worked.

[utdgrant]

It also depends on what you are doing. Are you setting a flanger by ear? Are you automating the knob (or modulating the value)? Are you doing something else? Are you more mathematically focused or more focused on adding a bit of this and that with a twiddle of a knob? It takes all kinds to make this wonderful, diverse world.

I definitely see where you're coming from. Hell, I even sympathise with some of the 'fader nerds' in that final thread!

I certainly fall into the "more mathematically focused" camp.

I personally feel that it's a good rule of thumb in the VM world to stay within the +/-5V range at the output of each module. There are valid arguments against being so uptight about it (particularly by MRB) but I like to 'gain stage' things to avoid confusion in my own head.

The Osc Mix control in Bully Beef is a pure linear control by deliberate choice. As a result you're guaranteed that the output waveform will never exceed those +/- 5V bounds. In fact, the output waveform almost always has a peak and trough right on those values.

In the case of flangers and phasers, an exact 1:1 relationship between wet and dry signal has a special significance. This is because you will only get total destructive interference at the notch frequencies when the amplitudes of the in-phase and out-of-phase signals are exactly equal. (Apologies for trying to teach my granny to suck eggs). Once again, if you're trying to keep things within +/-5V in the VM world, a linear mix guarantees that the CONstructive interference peaks will never exceed those bounds. A 100%:100% 'transition-style' curve mid-point will lead to a signal which is +/-10V (+6dB). Constant-power curves are great for uncorrelated signals, but flangers and phasers can only generate their desired effects due to the fact that the signals ARE correlated. (Again, apologies for stating the obvious).

[ColinP]

Thanks Borkman for those links. You are of course correct that everyone has their own perfectly valid take on these things. Things like workflow are intensely personal and it's understandable that people have strong opinions.

Grant and I probably are at one end of the spectrum, I guess you could call it the mathematical end.

This is all good healthy discussion and I welcome the fact that it's opened up beyond pan law as there are several related concepts that come under the "VCA pairs umbrella".

One thing that seems to dominate the discussions linked by Borkman is the difference between insert and send usage. Much of the criticism of linear dry/wet crossfading is answered by handling some things as sends. Particularly effects such as reverb because we think of reverb as adding something to a constant dry signal rather than mixing between dry/wet. So the effective gain should increase as we add reverb (which the CA curves deliver providing we never go past 50%).

The beauty of a modular setup is that we can patch whatever approach we want. So even if we want to have reverb as an insert effect we can set the reverb module to 100% wet and feed its output to one channel of a mixer and the dry signal to another channel thereby giving us the two knob solution - a sort of send/return entirely inside an insert (although propagation delay compensation might sometimes be desirable on the dry side of such configuratiosns).

The problem with such complicated setups is however that things become complicated. Hiding such complexity is one of the main goals behind my current project so its possible for people to construct a module that has controls that do exactly what they want without having to look for the right knob(s) in a cluster of modules with a bird's nest of wiring.

I also spent a litle time as a semi-pro sound engineer in my youth and have a fascination for mixing desks so want to be able to construct my own specialized ones that can have bizarre configurations. For full flexibility and automation control you can construct these bit by bit but to save time I have a mono/stereo mixer strip object that contains a fader, PPM, mute and solo buttons and a pan/balance knob. But these can be just a small part of a larger mixer strip that contains whatever sends, inserts, bus routing, EQ, compression etc one wants. One can end up with some complex wiring to configure groups and busses but all that is hidden off-screen.

More on-topic, I'm wondering whether there is a flexible curve module in the VM library that gives us nice control over transfer characteristics so we can build VCA pairs with whatever curves we want?

[ColinP]

I thought about starting another thread but this is kind of related so here goes...

Should solo override mute?

This seems to be another one of those things that people argue about but I think it should. As far as I remember all the physical desks I've used work this way.

So no matter what when a channel's solo button is pressed then it's on.

Code: Select all

if soloed
	on = true
else if muted
	on = false
else if any other channel is soloed
	on = false
else
	on = true

or in more compact form using Java precedence...

Code: Select all

on = soloed || ! ( muted || anyOtherChannelSoloed )

CA can't make their mind up about this as in the six input mixers solo overrides mute but in their console mixer mute overrides solo.

[utdgrant]

Code: Select all

if soloed
	on = true
else if muted
	on = false
else if any other channel is soloed
	on = false
else
	on = true

Definitely this logic! Every time you implement it differently a kitten dies.

[borkman]

Agreed. I just checked my DAW and it behaves that way as well. They even label the toggle 'Mute / solo' (though its really a mute toggle and solo functionality is a menu option).

[utdgrant]

The new-look Linear Crossfade has just been approved by CA. The addition of the Inverted Mix Output allows it to be used as a linear panner, too. It can be quite fun to get the two input channels bouncing about the stereo field in antiphase when you modulate the balance with an LFO.

FrontPanelV1.1.jpg (30.11 KiB) Viewed 1155 times

And I finally got round to writing a User Guide.