# Building a matrix audio mixer

I'd like to build a 4 in, 4 out, 9 V powered audio matrix mixer. Basically it's a mixer where each input has a knob to control the amount sent to each output, effectively forming a routing matrix, like this one.

But that one is a passive mixer, very simple but subject to massive signal loss. My goal is to make something that doesn't dampen the signal - it's ok if the signal is a bit dirty.

So far, my plan is to build something like this and replicate it 4 times. Being rather new to analog electronics, I'd like some advices.

This is based on a op-amp summing amplifier. Here are my (maybe silly) questions:

1. First, is it as simple as putting 4 of those in parallel, sharing their inputs? Or do I have to add something so one bus won't crosstalk with the others?

2. If I get it right, R1/R2 controls the gain applied to each signal in the final mix. So if I want to sum the signals, R1 = R2 right ? Is it ok to have R1 > R2 so pushing the volume knob all the way up will boost the signal?

3. If only the ratio between R1 and R2 matters, how should I choose the values R1 and R2 ? Same question goes for the potentiometer's resistance - how should I choose its value?

4. I guess there are a lot of things missing here. Any suggestion for a beginner?

• I don't like that the impedance of each input changes based on the pot position. For example if it's turned all the way to GND then it's a 10k impedance, but if it's halfway then it's a 35k impedance and will contribute only 1/3.5 as strongly to the mix as the one that's set all the way to GND. As far as your circuit is concerned, the RV is in series with R2 and that means you have varying R2. Jul 18 '21 at 21:03

## 1 Answer

1. Generally correct. More below.
2. R1 > R2 will increase gain. Correct.
3. See below.
4. You'll need a split-rail power supply.

General comments:

• You are loading a 100k pot with a 10k load to virtual ground. That will load the pot significantly and may affect the "feel" of the volume control linearity. (The pots will be logarithmic.) I would be inclined to use 10k pots and 100k for R1 and R2.
• The problem with my point above is that you will now have four 10k pots loading each input channel. The four in parallel will have a combined input impedance of 2.5k.

I would be inclined to add a unity gain buffer on each input. This should then be capable of driving four potentiometers without difficulty.

If there is any likelihood of DC bias on the output then add in a DC blocking capacitor.

OK, so how will you handle signals below 0 V? This can be done using a split rail supply (which you haven't got) or by creating a mid-rail "ground" reference.

Figure 1. A mid-rail "ground" reference. Image source: CircuitDiagram.net.

In this scheme the two 100k resistors on the bottom right have 4.5 V at their junction. This is stabilised by the 2.2 μF capacitor to ground (so that the reference doesn't jump around when you whack the strings hard!). The 4.5 V reference is then fed to the op-amps' non-inverting inputs.

Note that since the op-amp chain is now biased to 4.5 V that decoupling capacitors have to be added to each of the inputs and to the mixer output to block the DC.

• Thanks a lot for your detailed answer, it makes things clearer - and also sheds some light on some concepts I don't fully grasp yet. I'm going to iterate on my design and resubmit something later. Jul 20 '21 at 9:51
• Thanks for accepting my answer. How are you powering this? Jul 20 '21 at 9:52
• it should be powered with a 9v DC adapter, the kind you use for guitar pedals. Jul 22 '21 at 13:19
• OK, so how will you handle signals below 0 V? This can be done using a split rail supply (which you haven't got) or by creating a mid-rail "ground" reference. Jul 22 '21 at 14:46
• See the update. Thank you for accepting my answer. Jul 23 '21 at 22:56