Simple stereo audio mixer power supply

Question

I am trying to build myself a very simple audio mixer that can take 3 line-level inputs and create a single output for a set of powered PC speakers. I found a schematic online (here) which I have tried to copy as closely as possible. My main question is about the power supply. I am assuming I can use +-12V for the TL072 op-amps, but I wasn't sure what the best way to do it is, I found this handy page that describes a few methods and I thought I'd use the one that utilises the TLE2426 since it seems so easy. Can someone please advise if I have implemented this correctly? Hopefully the schematic below is of good enough quality. Again, I want this to be very simple in design, but obviously if I have done something wrong then I'd really appreciate the heads-up. Thanks!

EDIT - I forgot to say that one concern of mine was the low current capacity of the TLE2426 (maybe 20-40mA) so would this be a problem in this case?

Answer 1

Other than missing decoupling capacitors, there is nothing obviously wrong with the schematic as shown. There are no values for R8 and R11, so the intermediate stage gain is unknown. If it is more than 10-ish, offset error voltages from the preceding stages will show up as a non-zero DC voltage offset on the output signals on opamp output pins 7 and 7, reducing operating headroom asymmetrically. But with +/- 12V supplies and low gain this should not be an issue; just something to be aware of in a future project design with more gain.

The current in/out of the 2426 is the algebraic difference of all amplifier output signal currents. Since all of the stages are inverting and there are an even number of them, many of these currents will cancel out. Pay very careful attention to the 2426 datasheet for guidance on decoupling and output capacitance. If the datasheet does not disallow it, add a large cap (1000 uF or larger with a 0.1 uF ceramic in parallel) across the 24 Vdc input.

For ease in future discussions, add reference designators to the opamps and connectors.

Answer 2

Since you're AC coupling both in- and output, you simply do not need a bipolar supply. Also, 12V in both directions is about 12 times as much as you'll need to amplify line-level signals.

So, redesign time:

1. use a simple unipolar supply. 12V is much, 5V would do (with different opamps than the TL072), but you seem to already have a 12V suppy. So, fine! Let's call that 12 V (or whatever) VCC_raw. The other supply rail (i.e. your outer conductor in your barrel connector), connect to ground (GND).
2. Add a bit of power supply filtering. Depending on your opamp model and maker, the opamps you use will have a good power supply noise rejection, but you will still have stray effects that you don't want to have. So, add at least a ferrite bead in series followed by a couple hundred µF of capacitor to supply ground to add noise immunity. The thus filtered voltage rail will be called VCC. (You could also add a LDO between VCC_raw and VCC, modern LDOs have good noise rejection, but I don't know whether that'll pay in your use case at all)
3. Build a virtual ground in the middle between VCC and GND: High-ohmic (e.g. 1 MΩ) voltage divider, feeding a capacitor to GND (e.g 100 nF). Add an opamp (positive supply: VCC, negative supply: GND) in a voltage follower configuration. Keep traces short. Call its output VGND. You could use a TLE2426 instead of the voltage divider+capacitor+opamp voltage follower, but I really don't think you'd win measurably much – you care not at all that VGND is exactly half power supply voltage, but only about it being stable. In any case, make sure that all your ICs have sufficient decoupling caps from their positive to negative supply, placed close to their supply pins.
4. all your TL072 non-inverting inputs: connect them to VGND instead of GND.
5. same for GND in your output adjustment potentiometer.
6. If you want to stick with the TL072: The TL074 is the same opamp, but 4 in a single case instead of only 2.
7. Personally: get a nicer opamp, the TL072 is low-noise, yes, but has tremendous need for headroom, you need at least 10 V between positive and negative supply rail. You get the same (actually, better) noise performance from an AD8668, but could work with a 5V supply.
8. add a bit of low-pass filtering to your opamps: adding a cutoff at say 40 kHz doesn't hurt anything audible (even for dogs), but you gain immunity to EMI at the cost of maybe one capacitor per amplifier.