I’m wondering what conventions are followed when people design a mixer. In particular I’m curious about the signal levels at the input(s) and output(s) of the circuit.
Imagine you have a simple mixer and input a euro rack signal which is 5 Vpp (5 volt peak to peak). If you now input that signal into another channel of that mixer, then if they are simply added together (so not amplified) you end up with a signal of 10 Vpp. Add some more inputs to the mixer ( I own a mixer with 28 inputs, so being able to process a large number of signals is not exceptional ) and you’ll end up with a signal that will be clipped because its sum would exceed the power supply’s range.
Does anyone know how this is solved in real life mixers? Are signals dampened and then added together? Or is it left to the user of the mixer to not input signals that are close in amplitude to the max amplitude the mixer can output?
[Edit] It is more conventional to add output level attenuation to a module or leave it up to the next module to adapt to the input level? Both are possible, but I wonder whether from a design point of view the one is advantageous over the other.
I don’t know how pro gear does it, but all DIY mixers I have seen just don’t care, if the sum of your attenuated inputs is too big it will simply be clamped to the power rails.
“Normally” input and output levels match up, so you don’t need attenuation. But you still may want attenuation in some places to get the effect you’re looking (hearing?) for. In this case use an attenuation module at those places you want them…
Any component in the signal path will add it’s own distortion/noise/whatever to the signal, so having less is better. And modules will be smaller and cheaper if not adding rarely used attenuators, so you can get more of them
My 2 cents anyway …
Not a sound engineer, but isn’t the usual approach to use pad/gain controls to sort out the input levels, and once that’s done you use the faders to balance the mix (0 dB tends to be near the top of the fader range…)
This is what I thought was what is being used, but that would imply that you need to attenuate all inputs once you add another one (or one starts to produce audio) and also adapt the output level to keep a ‘constant’ level. But that is not how I perceive my mixer to work.
No you would not want that, so this boils down to that all leveling needs to be done before hand. But the hardware needs to supply the means to do that, otherwise overflows / clipping will occur / can not be prevented, I guess.
All the synth mixer modules I’ve seen have attenuators on the inputs, before any gain stage, and usually also on the output(s). E.g. the AI Synthesis mixer:
OK, I’ve seen this in some other schematics as well. From the choice of components in this schematic one can derive that the max amplification equals to 1, so if there is no attenuation then 3 input voltages of 5Vpp would exceed the power supply voltage (I’m assuming “worst case” input voltages here).
I started thinking about levels when I saw that the Barton wave shaper has an output stage that adds up to 7 logic levels (changes of the levels as there is a capacitor in there, but in absolute terms that will be about the same voltage) from a CD4040 together and outputs it via a buffer. That would be 7 x supply voltage worst case which is clipped to 1 x supply voltage.
Yes. But of course if you have only one 10Vpp input, or three inputs but two are dialed down to zero, then you want the capability of getting a 10Vpp output: You want to be able to have unity gain. Hence the output level control (on the first gain stage) — you turn that up when there’s only one input, turn it down when there are enough inputs to cause clipping.
I am currently designing a module and I’m asking myself whether I should leave the attenuation up to the signal source (the preceding module) or attenuate it in my module and conversely whether I should allow for attenuation of my module’s output signal of leave it up to the following module to solve that.
design considerations …
I do not want to end up with an idiosyncratic solution, that’s why I’m looking for design rules / best practices etc.
It seems to me most of the time synth modules allow for attenuation of inputs (if that) but not outputs. Mixers tend to be an exception, with both input and output controls, and this discussion has clarified in my mind the reason for that: the output control is to allow the user to prevent clipping within the module. It’s not even really an output control, it’s in the feedback of the mixing stage.
Look at the schematic above. Signals are mixed at IC1A, which is configured as an inverting stage so the inputs stay away from the rails. There is a pot in the feedback allowing the gain to be adjusted between 0 and 1, so the output level can be kept below 10Vpp.
Added: Maybe I spoke too soon: Looking at a bunch of mixer module schematics, the pot in the first stage feedback isn’t at all universal practice:
Ah, I was thinking of some feedback from the module / mixer output to the module / mixer input, while you were talking about the feedback of the op-amp.
I’ve seen some wave shapers that can clip internally because of too high gain factors or too high input voltages and especially with those one can not easily hear that they are clipping because of that (which I would call “unwanted”) because their main shaping function may be based on controlled clipping in itself.
Thx for the overview. I’ve had a look at a few matrix mixers and some of them also do not have one. There does not seem to be a uniform approach to this.
