I've done a little bit of homework about the master opamp selection. It's very unlikely that picking another opamp will provide a big reduction in noise. The RC4558 specifies a worst-case input noise of 8nV/rt(Hz), which is not easy to beat with cheap parts. The N5532 is a bit better with 6nV/rt(Hz) at 1kHz, but its worse case input bias current of 800nA is a bit sucky (only 500nA for the RC4558).
You must step outside the usual suspects for these kinds of circuits to get better performance - there are some options if you have money:
- LT1007 (3.8nV/rt(Hz) input noise, 1.7V/us slew rate (minimum, equal to the typical slew rate of the RC4558), much better input offset voltage and input bias current specs) - buy the "A" grade part. Whoops! This doesn't come in a stand dual opamp package!
- LT1677 - a rail-to-rail input and output version of the LT1007. If you're neurotic about headroom, this opamp will give you plenty.
- LT1028A (1.1nV/rt(Hz) input noise (!!!), 11V/us slew rate.
That all being said, shitty electrolytic caps are noisier than the cheap RC4558. I'll take some before/after measurements of the master output noise on my Yocto when the high-quality caps I've ordered arrive. From there, I may do some opamp swapping and see what the results are. I'll be sure to share the results here.