dlaloum wrote:Hmmm - would THD measurements at 15kHz identify and seperate a 0.7mil conical/spherical from a 0.4mil eliptical?
And if so what would the expected difference be?
Well, THD at 15kHz is inaudible anyway, not to mention tricky to measure well. In any event, one doesn't have to go to hf to measure difference, in principle. Pinch effect relates to groove angle, not curvature, so is level related. And should show up in principle even at 1kHz at high enough levels.
Some time back, I included a pinch effect distortion calculator in groovy.xls. It lets one explore pinch effect, amongst other things, and it's here :
IN PRACTICE, however, other factors than stylus profile tend to dominate THD performance. For example, Paul Miller publishes distortion measurements based on 0dB level (ref 5cm/s) at 1kHz. In these tests, the DL103R is better than the 2M Red by far, and the 2M Black by a little. AT OC9/III and AT440MLa are better than all, but not that much better than the DL103R.
Such things, and my own listening tests, led me to conclude that popular opinion of the audible effects of distortion arising from spherical profile styli is often greatly exaggerated. In normal programme material. As Bran K suggests, I also concluded that much marketing hype might be responsible.
What's near impossible to do is try a substitution where just the stylus profile changes. Not the cantilever, suspension, azimuth, rotation/shank alignment, VTA etc etc. That makes it near impossible to be conclusive about measuring, or listening to, effects of stylus profile in isolation, as I see it.
Sphericals have a couple of advantages too. Easier to make to close tolerance of geometry and polish. Easier to mount on the cantilever, eliminating zenith alignment contribution to overall tolerance. Lower friction, improved trackability and surface flicker noise. Can be run at higher VTF, better trackability. Base clearance can be better, lower surface noise. But some of these advantages might equally apply to 'quasi sphericals', or 'bi-radials' as described on this thread, of course.