Diff(erent) input stages, simple is not better

[nelsonvandal]

My views on different differential input stages in folded cascode headphone amps, both AD797 and AD829 types. Current through input stage is 2 - 4 mA. All amps are tuned for a good square response. Voltage supply 10 or 18 V depending on battery or wall-power operation. In spite of great difference in THD there's very little or no audible difference with increased supply.

You probably know I'm not an engineer, so these are of course an amateurish view, but I do as good as I can with LTSpice, RMAA and an oscilloscope.

1. JFET (SK170): Good tonality but warmer than neutral. Lack of definition and dynamics. Fuzzy soundstage. Typical JFET graininess.

2. SK170 cascoded with SK246: leaner and probably brighter than neutral, increased clarity with better details and soundstage, less but obvious JFET sound.

3. CFP with SK170 and SA1016: better clarity than 1 but not as good as 2. Soundstage is also better in 2. Tonality closer to 1. Grainy.

4. SK170/SA1016 CFP and SK246 cascode: to my ears much better than the three above. Clean non-fatiguing sound with very little JFET character.

5. Bipolar (SC2362): tonality is OK, so is dynamics and soundstage. Instead of JFET graininess there's a metallic "solid state" sound. With SC2240 sibilant and even more "metallic". With BC550 the tonality is close to SC2362 but with less clarity.

6. SC2362 cascoded with SK246: better clarity and brighter but not really an enjoyable sound.

7. SC2362/SA1016 CFP: much more enjoyable than 5 and 6. Tonality warmer than neutral but it doesn't lack details or airiness.

[AndrewT]

that must have taken a fair bit of time with all those different prototypes.
Thanks for all the effort & report.

[Bonsai]

How did you arrive at these conclusions? Can you share what you used?