Simple Mixer Schematics

Virtual Earth 

Circuit 2 shows a basic active mixer. It uses 2 virtual earth preamps. One for the summing node and 1 to re-invert the phase of the signal.

The summing node (The point at which all the resistors meet) enteres into the inverting input of the op-amp. A feedback resistor is connected between the output of the op-amp and the inverting input. The function off this feedback loop is essentially to limit the open-loop gain of the op-amp.

Any signal entering the inverting input of the op-amp will appear at the output but it will be upside down. That is to say 180 degrees out of phase. In other words if you put 2 volts in you'd expect -2 volts out. To achieve unity gain (that is no gain or amplification at all) the feedback resistor must be the same as the summing resistor. In this case 10K. All the summing resistors are 10K and the feedback resistor is 10K. Because the feedback resistor feeds the output signal back to the inverting input of the op-amp @ 180 degrees out of phases it cancels out any gain. It also means that the inverting input of the op-amp is held pretty close (If not exactly) at zero volts. or earth potential. Thus the term "virtual earth".

Any signal coming in through the summing resistor is like dumping it to ground via 10K. It theoretically has the same loss. However the feedback resistor of 10K gives the exact opposite in gain. So if you feed 2 volts in you will get 2 volts out only it will be upside down.

Because the summing node (The inverting input of the op-amp) is at virtually earth potential, there is little chance that this signal will bleed it's way out to any of the other inputs. Essentially speaking all the audio sources are isolated from each other.

However we're still left with the problem of the phase being wrong. If the output of the first op-amp were recombined with one of the other signals at a later stage it would cancel out rather than mix. So we have to re-invert the phase with yet another op-amp. This is a unity gain amplifier just like the first except that there is no summing node as such. (Except for the feedback resistor of course) The output of these two stages will now be the summ of all the inputs with the correct phase. Because of the inherent compensation of the feedback/op-amp/summing node, there is virtually no limit to the number of inputs you can put on this. Most modern op-amps have enough drive capability that 128 inputs would be just peanuts.

However it must be remembered that you are summing the inputs so if you had a powersupply of say +/- 15 volts, and 4 inputs of +5 volts each, The result would be 20 volts mathematically speaking. But the op-amp can only produce +15 volts so you would be clipping by 25%. Distortion occurs. Most op-amps can't swing exactly to the supply rails so clipping and distortion would be even worse. In practice however most audio signal wouldn't exceed a few hundred milivolts. A 2 volt peak to peak signal is considered to be a very high level.

Simple Mixer Schematics

Active state

Active Mixer stages that use Op-Amps are generally known as virtual earth pre-amps. These are inverting in nature. 180 degrees out of phase. IE: The signal coming out of the mixer is upside down as compared to that which is entering it. You then need to use another inverting pre-amp to recover the phase.

This would seem silly at first until you realize that virtual earth means that the inverting node of the op-amp is held virtually at ground (zero volts) potential. Any signal entering the stage via one resistor cannot find it's way back out of any other resistor. This prevents the audio from *say* one synth, polluting the audio from another. Particularly useful in a Mixer with many busses and sends.

Generally speaking the pre-amps stage does not provide any gain. IE: is 1:1 unity gain. A signal passing through a resistor with no load also presents no loss. Even with values beyond 1 meg. Although you may drop the effective current at the other end of the resistor. In this case the current loss is largely irrelevant. Especially at line-level. And is compensated by the op-amp's drive current in an active system.

It is better to have a mixer stage with no gain (or unity gain) because this will not amplify the noise. If good quality op-amps are used, they will not add significantly to the over all noise performance. So the RMS voltage coming out of the mixer should be the same as the sum of all it's inputs. If gain is necessary for a microphone or phono etc, the gain should be a special stage at the top of the chain. IE: the first preamp in the mixer channel. This is then mixed with everything else once the microphone is amplified to line level. This gain stage only adds noise to the microphone and not to the sum of the signals passing through the mixer.

It is interesting to note that resistors themselves add noise to a circuit. This is known as thermal noise. Generally speaking the rule of thumb is: The larger the value the resistor, the greater the thermal noise. This may not be significant in mixer stages at line level but where large gains are required it is desirable to use smaller value resistors. (as small as possible within reason.) Of course sometimes this cannot be achieved but is worth remembering as a rule of thumb. Metal film resistors have less thermal noise than carbon film resistors and are more temperature stable over all. So now there's two reasons to use Metal films in audio circuits.

Driving the busses

Note here that Mixers are more repetitious than complex. The circuits are relatively simple it's just that there's a lot of them. Especially in large recording consoles.

Usually these desks are seen in two halves. The input half and the output half. No matter how complex the input half may become, the output half is essentially just a virtual earth pre-amp as described in the circuits above. Often it is required to have many such busses for things like effects sends, subgroups, monitor bus and so on.

One of the beauties of the virtual earth mixer is that there is also virtually no limit to the number of additional busses as well as the main bus. One could arrange an effects send buss that derives it's signal from the same channel as the main bus. Except that each has it's own volume, pan and assignment independent of each other.

Simple Mixer Schematics

PREAMBLE 

I've been cooking audio circuits for so long now I no-longer need a recipe. A lot of the theory I have forgotten over the years because I've just gotten to know the circuits by instinct. But this should serve as something of a guide to designing mixers from scratch.

The idea: 
Most people reading this would be well aware of what a mixer is used for but I'll reiterate here. The job of an Audio mixer is to combine various audio signals into a single audio signal. It is better known in electronic terms as a summing circuit. That is to say that the output is the sum of all of the inputs. A summing note is often represented as a circle with a PLUS (+) symbol in it.

Audio is of course an AC (Alternating current) signal but if we look at the incoming signals as a frozen moment in time we can represent it as 2 or more DC voltages. This is only useful to illustrate the point.

If we had two signals to be mixed. The first was 2 volts and the second was 3, the output should be the sum of these two voltages. 2+3=5. If on the other hand the two voltages were 2 volts and -3 volts then the output would be -1 volt. We are now subtracting 3 volts from the +2 volts leaving -3. It is important to recognise that we are dealing with what is known as a bipolar signal. That is one that can be positive or negative around a zero base-line.

When you get to the stage of adding many signals together, the complexity grows. In1 + In2 + In3 + In4 + .... and so on.

Because each incoming signal has it's own load impedance it is impractical just to wire all of them together and hope for the best. Especially when the following device you are trying to mix into also represents it's own load impedance. Sometimes you may be able to get away with it because the combined impedance is quite high. However most of the time it drags the whole network down and causes one or more devices to fail or distort or what ever. Usually no damage is done but it just won't work.

What is required is a little load isolation. (See Circuit 1: Passive mixer) The trade off is that you can't use terribly high value resistors because of the losses that they may cause. Especially if the load impedance of the following device is a little low. This will give the effect of severely attenuating some or all of the signals. A practical trade off has to be reached and this is as much trial and error as anything because the conditions change with each new device added or changed.

The device used at the summing node, IE: an amplifier or tape deck should be able to provide enough gain to compensate for the combined losses through the resistors and the combined loading of the system. But the loading will change depending on the combination of devices you have hooked into it.

This approach also creates another side effect. That is that a signal flowing into the summing node via one source can pollute the audio signals of other devices. Say you had two cassette decks that you wished to mix. However you also wanted to send the audio from cassette deck #1 to an effects processor. The audio from cassette deck #2, although attenuated slightly, will find it's way back to the audio from cassette deck 1 and also go to the effects processor.