Amp/Preamp Asylum Looking for a new Amp or Preamp? If you're after tubes, post over here.

---

[ Follow Ups ] [ Post Followup ] Thread: [ Display  All  Email ] [ Amp/Preamp Asylum ]

---

[ Alert Moderator ]

---

RE: The undistorted truth

Posted by 13th Duke of Wymbourne (A) on March 18, 2025 at 16:58:25

In Reply to: RE: The undistorted truth posted by morricab on March 17, 2025 at 08:49:38:

Good references.
Let's deal with the fluff from Benchmark first: that cross-over distortion is insidious because it creates intermodulation. This is a red herring as all non-linearities create intermodulation just as they create harmonics. Studying the generation of harmonics from a single-tone gives you a view of how non-linear an amplifier is without the added complexity of all the other tones. Whether we should be more concerned about the audible impact of IMD is another, maybe larger, topic.

I think the crux of this whole matter is the definition of class-B and you'll probably say "Doh! Of course it is". I've been biased (sic) by Self's definition of optimally biased class B. If you look at his book or website you'll see that 'optimal' means the least disturbance to the output stage small signal gain as you go through the hand-over region. In his book I don't see him explicitly reference the equivalent, presumably small, class-A current for that condition, maybe it can be extracted on more thorough review. In his 'blameless' design the output stage small signal gain doesn't vary much and, therefore, I think you can model that as part of the overall amplifier linearity and lump all the distortion together as something 'weakly non-linear' where I would expect relative distortion to fall faster than the input level falls, and vice versa. I found a Benchmark app note where they compare their AHB2 to a 'convention class AB design'. They imply that the conventional design will suffer cross-over distortion and will be worse at low output levels but then they completely fail to prove it in the measurements of both as relative distortion is flat with output power at low power levels. I suggest it will improve, Benchmark suggest it gets worse but it doesn't change so we have some distortion cancellation (technical joke). The one takeaway from that Benchmark paper is they can readily identify between the two amps in ABX tests at 10mW output. The AHB2 THD is 35dB better (but we can't say how much is cross-over related from THD). So, boos to Benchmark for using a bit of FUD in their material but hoorays for promoting ABX.

Moving on to the Rod Elliot piece, nothing to discuss there as all his writings are well reasoned and backed-up with analysis. In his analysis the output devices fully switch off which brings me back to the crux, if the definition of class B must be that the small signal gain of the output stage drops considerably/goes to zero then, for sure, the discontinuity will generate a whole lot of distortion and lower/eliminate the loop gain to correct it. A situation like that does not fit a 'weakly non-linear' model or even a strongly non-linear regime (e.g. 1dB gain compression), it has crossed the event horizon to oblivion.

In summary, we should assume that anything labelled 'class-B' with low measured distortion, like the Topping, must follow Self's optimal class-B bias (or similar). Anything truly class-B must have a significant drop in the small signal gain of the output stage at cross-over compared to the equivalent class A bias and we can call that morricab-B. Or, maybe, moribund-B because why would you do that?

Edits: 03/18/25

This post is made possible by the generous support of people like you and our sponsors:

Sonic Craft

---

[ Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2025, Audio Asylum, all rights reserved ]

---

[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]

---