Hi Pavel,
These results are very much inline with the amplifier testing done by Stereophile using simulated speaker loads so it's not that new. What you could do to examine this in more detail is to copy Stereophile's simulated load in XSim, but maybe replace the last inductor with an ideal speaker. Then add a resistance in series with the amplifier.
Not sure if it can measure this low, but you could attempt to measure cable capacitance and inductance with DATS. I'm sure there are better tools for this, or you can find the published specs or measurements online. Some reviewers include this data for cables they measure, especially some Wireworld reviews.
Best,
Erik
Last edited:
Apparently after I started typing a reply, see it now. Better than nominal of course since deviation scales with impedance.
Any thought on the LF distortion mechanism? The 60 and 120 Hz bumps originally suggested differential chassis potentials but measurements made with a high impedance balanced input arguably handle that. Plus that curious distortion reduction above 200 Hz...
Hi Erik,
though I have not mentioned it, I have used a simulated speaker RLC load for about 12 years, however, I have found quite big differences between real speakers and simulated loads.
Though measuring/simulating of frequency response is in a good conformance between the two approaches, for non-linear distortion is RLC dummy load useless. R, L and C are linear, unless you use L with iron core. BUT, impedance of the real speaker is nonlinear. It is a function of both frequency and amplitude. Z = Z(f,V). The circuit elements of real speaker are non-linear with cone excursion and frequency. Thus, you get distortion of speaker current even if drive voltage has no distortion. This will never happen with linear RLC dummy load. So, RLC dummy load made from discrete elements is a kind of approximation, but it does not describe the reality well. You need a real speaker to make an investigation.
Pavel
though I have not mentioned it, I have used a simulated speaker RLC load for about 12 years, however, I have found quite big differences between real speakers and simulated loads.
Though measuring/simulating of frequency response is in a good conformance between the two approaches, for non-linear distortion is RLC dummy load useless. R, L and C are linear, unless you use L with iron core. BUT, impedance of the real speaker is nonlinear. It is a function of both frequency and amplitude. Z = Z(f,V). The circuit elements of real speaker are non-linear with cone excursion and frequency. Thus, you get distortion of speaker current even if drive voltage has no distortion. This will never happen with linear RLC dummy load. So, RLC dummy load made from discrete elements is a kind of approximation, but it does not describe the reality well. You need a real speaker to make an investigation.
Pavel
Nonlinear speaker impedance, like explained here
Current drive of speakers and speaker distortion
Lumped elements of speaker model are nonlinear with amplitude (excursion) and frequency (voice coil inductance, e.g.).
Thus speaker current is nonlinear with linear voltage drive. Nonlinear current * cable resistance results in the distortion shown, measured at the end of the cable loaded with the speaker.
Yes, the amplifier very close to speaker terminals would eliminate cable resistance and inductance influence, but also with active speakers (active crossover) it is easier to define operating conditions of individual drivers, regardless if we have SW crossover or link level crossover with opamps.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.