The AP can do that, since its DAC and ADC are in the same box. If you use a professional audio workstation, you can usually slave the system for playback to the digital output of the ADC. And, some devices like standalone ADCs and DACs can be slaved to some form of word clock so that the ADC and DAC use the same clock. There will be a little bit of 'squish' from the PLLs involved, but that should be very very minor, especially with quality equipment. A lot of pro-sumer digital audio stuff can't be slaved easily or at all, and I can see how that drift will basically ruin averaged synchronous sampling. In your example, you'd need to record analog but slave the ADC to the digital DAC output to get synchrony. Many workstations can do that, but not all.
Still, using power averaging and a wide FFT, you can dig >40dB into the overall noise floor pretty easily, and that's a great help with modern amplifiers and circuits that have far less nonlinearity than noise.
Thanks for the Python / FFTW tip - there are certainly cases where synchronous processing is very useful, and this might be a better way to approach it than some of the more direct methods, like quadrature sampling and direct averaging. It's a good excuse to get up to speed with Python too!
Still, using power averaging and a wide FFT, you can dig >40dB into the overall noise floor pretty easily, and that's a great help with modern amplifiers and circuits that have far less nonlinearity than noise.
Thanks for the Python / FFTW tip - there are certainly cases where synchronous processing is very useful, and this might be a better way to approach it than some of the more direct methods, like quadrature sampling and direct averaging. It's a good excuse to get up to speed with Python too!