How much gain should be applied to EFX Sends

    Oct 15 2012 | 2:55 pm
    I built a mixer to handle a bunch of external inputs as well as VSTs. I have two EFX available for each channel.
    Recently, after some great advice from people here, I corrected a big problem with my faders which were doing linear rather than log style fades. That stuff works much better now but I have found that when I use post-fade mode to feed the EFX channels, I have to set my EFX Send values very high to make them work reasonably well. I'm wondering if I should have a different kind of gain system for the effect send stuff.
    Attached are the basic ChannelStrip and EffectSend patchers and if anyone has time to see if I should be doing it differently, I'd really appreciate it.

    • Oct 15 2012 | 7:15 pm
      i cant look at max 6 patches now, but an effect send as well as an AUX or any other kind of output
      should usually have the same level as the inputs and should usually allow the "full" output
      just as every other channel, for example 1.0 in digital, or +12 db or whatever.
      how can your be not loud enough? did you lower the volume somewhere? :)
    • Oct 15 2012 | 8:59 pm
      What I'm finding is that if I use 'post-fader' mode, then unless my fader is very high, the input from the fader output to the EFX send is very quiet and so I have to have the EFX send very high (usually higher than 0dB) to get anything useful out of it. I don't really know too much about the mixer process so I haven't figured out whether I'm supposed to be adding some gain to the EFX or not.
    • Oct 15 2012 | 9:30 pm
      Your patches are missing the bpatcher abstractions, so I can't tell. One minor bit of advice would be to do something like this for your db scale, since right now when your fader is at 0, it's not fading all the way out. (you're applying -70 dB of gain, but things will still make it through.
      I recommend something like:
    • Oct 15 2012 | 9:35 pm
      Also, I have an object in my tools called "PM.MatrixVolumePan" which will turn a matrix~ into a mixer, with muting, panning, trim, etc. It doesn't do sends right now, but that would be easy enough to add. It also does all of the dsp in one matrix~...
    • Oct 16 2012 | 3:24 pm
      Thanks, Peter -- I thought the second patch I posted was actually the contents of the bpatcher used in the first one.
      I'll study your solution and also take a look at your MatrixVolumePan. I've never worked with matrix~ and indeed my MSP knowledge is very sketchy....just enough to have built the simple mixer that I needed.
      Appreciate the suggestions.
    • Oct 16 2012 | 3:30 pm
      You're absolutely right about the slight audio still making it through so the suggestion to detect -70. is a great idea....I'll modify my patcher immediately.
      However, not sure how this applies to the Send stuff. If I have my fader half way up (physically), the dB is -32 and although I can hear an instrument quietly, that also means that only -32dB of audio is getting through to the 'Send' if the latter is at 0dB. So I seem to have to put the Send way up (3 to 6dB) to get anything useful when the underlying audio is quiet. In my limited experience with physical mixers, it has not been like that.
    • Oct 16 2012 | 3:58 pm
      Wouldn't it be simpler using a [gain~] object?
    • Oct 16 2012 | 4:09 pm
      The [gain~] is hard coded for sliders. I needed to do this for dials as well as other places where there are no GUI parts.
      Also trying to separate as much as feasible the view pieces from the model pieces.
    • Oct 16 2012 | 4:35 pm
      ok good that you added that the problem is on the theoretical layer :) the problem is clearer now.
      effect sends are usually "prefader", so you just have to create a second branch of the input to the fader.
      some analog desks also offer post fader sends, and many of them have sends where the user can switch between "pre" and "post".
      as you can see, it is easier than you think.
    • Oct 16 2012 | 4:45 pm
      @dhjdhjdhj: It sounds like the problem that you're having is one of taper. You can have scale use a logarithmic mapping to give you more values up at the top, so use something like:
      scale 0. 127. -70. 0. 0.5 @classic 0
      The 0.5 means x^1/2 (which is sort, though you could use other log scalings > 0 and < 1). @classic 0 tells scale to not use the wonky IRCAM way of doing it and just use normal exponents.
    • Oct 16 2012 | 7:28 pm
      @Roman --- I already have both prefader and postfader support in my implementation, the issue is just with levels. I had a chat today with a friend who runs a recording studio and he mentioned that ProTools faders have gain up to 12dB so I might try playing with the max gain.
      @Peter I'm going to try that scaling trick right now and see what happens.
    • Oct 16 2012 | 7:50 pm
      You can also accomplish something similar using the "function" object as a lookup table. (esp. with curve mode on!)
    • Oct 16 2012 | 9:43 pm
      "If I have my fader half way up (physically), the dB is -32 and although I can hear an instrument quietly, that also means that only -32dB of audio is getting through to the 'Send' if the latter is at 0dB."
      this doesnt sound anormal to me, it is what you would exspect, except when you confused pre and post. :)
      n * 0.2 * 1 == n * 0.2
    • Oct 16 2012 | 9:56 pm
      I guess I said that badly --- what I meant by that was that when I have the fader at a volume that is appropriate for the sound in question, there isn't enough signal going into the effects section such that I can hear the desired effect along with the original sound unless the EFX Send is at a very high value. That had led me to wonder whether the Send section should have a higher multipler on its gain.
      The second part of this was that fading the volume in real time didn't feel "natural" to me. I know that's qualitative. If I have a knob or a foot pedal attached to one of the faders, the amount of change I was getting relative to the adjusts I was making felt wrong.
      Having said this, Peter's modification of the [scale] object has improved this significantly.....I'm still experimenting.
    • Oct 17 2012 | 2:05 am
      You might try a different exponent for pre and post. Pre-fader doesn't need as much of a logarithmic bent, whereas post-fader (presuming fader is
      Alternatively, you might look for a compromise value.
    • Oct 17 2012 | 4:13 am
      I'm not as concerned with prefade right now as I don't really use them much. But I made the changes you suggested for scaling the faders and the difference is terrific, new way feels much more intuitively right.
    • Oct 17 2012 | 6:00 am
      In digital mixers, the master fader for aux sends goes to 0dBFS at max, but the aux send for each channel can go up to + 6/10 dB. So sometimes you have to cheat by raising the gain on your channel to send enough gain to your FX, then apply an attenuation in the eq section for example. But the logarithmic scale sure will give you more range in your settings with the high values.
    • Oct 17 2012 | 11:30 am
      A recording engineering friend of mine told me yesterday evening that in Pro Tools, the sends go up to 12dB and that extra gain seems like it might come in very handy. I'm going to try that today. Combined with the tapering that I added yesterday, my guess is that my system is going to be much more effective.
      The advice here has been great.