>> Quote: output noise
> What about input noise?
AKG 414 is unity gain. Output noise IS input noise. (Yes, it is 2-stage and the second stage could add noise.... but the 2nd stage works at much lower impedance with BJT inputs so I doubt it adds enough to matter.)
> the FET parts are a little more noisy on the voltage side.
Not for source impedances over 1 Meg.
> I would bet that the discrete method is actually noisier right?
That's a VERY GENERAL assertion. Sure, I have a totally discrete Radio Shed phono preamp that is noisier than a 741 chip. And a discrete-input mike amp that is lower noise than any chip made 10 years ago.
In general: the parts can be the same. But they usually are not. Multi-mask chip processing is costly, so input devices are small-area, and you don't want to reject entire chips for a little excess noise. Each chip has high development costs, a limited market, high per-unit cost. Single devices are cheaper per acre, are often produced by the millions, price may be so low you can buy a bucket-full and sort-out the best ones.
Then again, the intrinsic matching on a chip may allow some clever tricks......
And there are a few high-volume/profit applications where a low-low-noise chip is worth the development and marketing costs.
For large-diaphragm audio-band condensers, head-amp noise is not a Big Problem. TL072 may be at the upper end of what we should accept (yet I have great recordings made with preamps of 6dB higher noise level), and strains bad in low impedance.... but golly, it is 30 years old and costs 30 cents each. Yeah, you can probably find something with lower input noise and a more serious output stage. Maybe even something which will drive term-75 loads (150 ohms) at -86dB corruption. And maybe for under $4, which is nothing compared to the mechanical costs of capsule and a nice housing.