Hi All, Geof from littleBits here.
The i36-mix module is definitely unusual, but we had a few unusual things going on that we had to address. For example, all the amps are non-inverting, which is bad manners, but our hand was forced. And handling both DC and AC signals simultaneously. There's no question that there are different ways to address these things, but here's why we did what we did, opamp by opamp, moving left to right across the schematic:
U3-U5 - input buffers. These are required because system rules for our modules are that 1meg input impedance is preferred. Putting pots directly on the input would violate that system rule.
The pots. That was the simple part.
Passive AC/DC path splitting. Ok, so most audio systems have split +- rails, but we only have 0 to 5V. Also, for littleBits, every module must "work" with any other module. And Many/most of our modules operate with DC voltages ranging from 0 to 5V. So we need a DC path to comply with system rules. Then, because it's a mixer, we have to choose: do we scale both inputs at unity (so it CLIPS when both inputs are at 50%) or do we scale both inputs at 50%, so with both inputs at 100%, the output is 100%? We chose the latter.
The AC path through C1 had to be sensible for the types of inputs we were using for, which were our synth modules mainly. Those modules generaly produce large scale if not full scale signals. Basically, we tuned this by ear, and there's about 3dB of gain at signal frequencies. That's hard to see because the impedances of the passive AC/DC splitting interact. Why did we do that? It kinda seems like the weirdest part of the circuit. In this case, simply to save the cost of more opamps.
So U2 is the AC path non-inverting recovery stage. With a gain of 4 (33k/10k +1) to compensate for lasses in the passive AC/DC split, we end up around +3dB, which again, sounded best.
U4 is a non-inverting DC recovery stage, referenced to ground for DC accuracy. Importantly, this stage must be non-inverting because of the single supply and the input polarities. That means U2 must also be non-inverting.
U1 is the output driver. Again, has to be non-inverting because of U4. Required to meet system spec of low impedance output of the module.
So I know it's a bit unusual, but it solves the "we need to pass both AC and DC signals through a single supply system" problem without adding more op amps. We are under fairly substantial cost pressures for our designs, so we had to select the minimum number of opamps that complied with system rules.
I'm not on the forum continuously, but hit me with your questions and suggestions, and I'll answer as best I can.
Thanks!
Geof