I started replying my previous post before you posted this answer. This fits with what @jhstewart9 was saying about speaker distortion masking small amp distortions.
Hi @waltube ,
@jhstewart9 with "LHS" wants to focus on the Left Hand Side, he refers to the -40 to 0 dB peaks at low frequencies (below 1kHz down to DC).
Could it be linked to measurements and not to real PSU noise?
@jhstewart9 with "LHS" wants to focus on the Left Hand Side, he refers to the -40 to 0 dB peaks at low frequencies (below 1kHz down to DC).
Could it be linked to measurements and not to real PSU noise?
With all the respect in the world, you asked an ambiguous question.
Without context.
And then you offer some low level annoying dissing who asks a legitimate question about context.
Context is important to every question.
My 2 cents ymmv etc
I would worry most that our attempts to draw broad general conclusions from small personal unblinded data sets might somehow be misconstrued as real information, then amplified by the magical power of modern echo chamber discussion groups. Folk who get too tired of insisting on some validation eventually get tired enough to just leave. But we see it in all areas of discourse, from nutzy techie folk like us to politics, where people will believe all kinds of foolishness. There's an ancient Chinese curse "May you live in interesting times."
All good fortune,
Chris
All good fortune,
Chris
I think this quote from your post is one of the best things I’ve read in a while (and with a back hand reference to France’s “banana specifications” of the 1980’s:
That will then be like asking people how curved they prefer their banana.
My 2 cents, ymmv
So long, and thanks for all the fish! 😛
You didn't get the joke .... or you didn't read the "most important" guide. 😉
You didn't get the joke .... or you didn't read the "most important" guide. 😉
At last one of a few sensible persons speaks up, in this herd of sheep 🤣 .
Es gibt anderes Forum. Kannst du vieleicht einen freundlichen und gut empfehlen😉?
Last edited:
Hi Schirm,
42 is the answer to the question of life the universe and everything. I recommend to you Douglas Adam’s’ “The Hitchikers guide to the Galaxy” for a short and highly entertaining exploration of this question. Before it was a book, it was a radio series broadcast by the BBC. You can probably find somewhere on the interweb that has it available to listen to.
Apologies for the lack of context. It was meant as a joke and I figured people of a certain age might get a giggle out of it.
It may be a bit silly and absurdist for you and @revintage in your earnest seriousness however.
My 2 cents, ymmv etc
@revintage
I passed through your same questions and I got similar replies. There's no "bad people", "sheeps" or anything like that.
It's people trying to explain you that getting the classical answer "-20dB monotonic from harmonic to harmonic" can be a target, but each amp will react differently on each speaker, so you will never get what you'd expect. The other important answer you got is that at the distortions you showed, speaker distortion will mask everything.
I passed through your same questions and I got similar replies. There's no "bad people", "sheeps" or anything like that.
It's people trying to explain you that getting the classical answer "-20dB monotonic from harmonic to harmonic" can be a target, but each amp will react differently on each speaker, so you will never get what you'd expect. The other important answer you got is that at the distortions you showed, speaker distortion will mask everything.
In great oversimplification terms, even order distortion is the happy, good sounding distortion, and odd order is the harsh sounding stuff. There are two evils here but on sounds good to the human ear. Single ended tube amps are loved for their even ordered distortion spectrum. Solid state is mostly odd ordered distortion. I’d rather have slightly less power and as a result less odd order distortion. Again, greatly oversimplified but true in general.
Later Douglas Adams books expanded on the The Answer* (42). An even larger computer was built to try to determine The Question to which the answer was 42, and involved an elaborate analog computer running over billions of years (the Earth), which unfortunately was destroyed, just as it was about to finish its run, by the Vogons to build a hyperspacial bypass.
Not to spoil the machinations, but The Question was ulimately determined to be "What's six times nine?" Seems appropriate for this thread.
All good fortune,
Chris
Not to spoil the machinations, but The Question was ulimately determined to be "What's six times nine?" Seems appropriate for this thread.
All good fortune,
Chris
My take:
HD should be sufficiently low, but #1 the circuit needs to be well understood.
Unweighted THD is a bit notorious for being both moderately useful as a "broken? / not broken?" indicator for various circuit ideas, but also seems to give false positives. Hence, strange and conflicting anecdotes abound: 0.1% sounding really good, then 0.05% sounds bad again. Then incremental improvements seem pretty consistent through 0.01%, until 0.001% fades into "nothingness". Not because it's some holy grail, but there's merely a lack of meaningful improvement, and usually the speaker gets blamed as the bottleneck. Meanwhile, that 0.1% still sounds better, despite having a similar bottleneck so people start rolling their eyes and making up stories about "pleasing distortion" and subjectivists adding sweetness to their distortion boxes.
Meanwhile, I've noticed there is some technical truth to masking making certain things sound better. And it's purely physical, not psychological. If you have IMD, or specifically some amplitude modulation, and you add a steady tone that matches the centre frequency of the 'problem' sound, that will reduce the dB level of the unwanted modulation. Draw it on graph paper yourself and you will see that it's deceptively simple and not magic!
E.g. Sketch a sine wave with strong modulation so the tremolo peaks vary from ±1 to ±10. That's 20dB of modulation. Then add a steady ±5 sine on top. By adding HD (which could be a tiny amount) you've now stabilized the sound and the ratio of modulation is now only 15:6 or 8dB.
What happens if a SE amplifier is optimized for energy efficiency by making it push-pull? The harmonics cancel out... Some of the time. This is a likely case of "the better it measures, the worse it gets". One naive approach would be to get output transistor (or tube) matching as close as possible. But the likely outcome is that the parameters momentarily change, so the harmonics oscillate between near-perfect cancellation and not-so-perfect cancellation. Highly optimised and balanced LTPs? Same problem. Then people say, oh we tune them by machine, not by hand. Human ears aren't sensitive enough to select the lowest distortion... But, they don't pause to think, do they? Maybe human ears are indeed the ultimate arbiter to decide the best ratio of orthogonal parameters like steady HD vs amplitude modulation?
Nelson Pass wrote about IMD getting more complex as more gain stages are stacked in series. This is one thing I'm keeping in mind for myself. It's awfully tempting to go from 2 up to 3 stages, with the simulator promising small fractions of a % improvements in HD, when it may not be desireable. Same with LTP gain stages vs their emitter-feedback poorer cousins, with lower gain, higher distortion, but also less to mess up.
HD should be sufficiently low, but #1 the circuit needs to be well understood.
Unweighted THD is a bit notorious for being both moderately useful as a "broken? / not broken?" indicator for various circuit ideas, but also seems to give false positives. Hence, strange and conflicting anecdotes abound: 0.1% sounding really good, then 0.05% sounds bad again. Then incremental improvements seem pretty consistent through 0.01%, until 0.001% fades into "nothingness". Not because it's some holy grail, but there's merely a lack of meaningful improvement, and usually the speaker gets blamed as the bottleneck. Meanwhile, that 0.1% still sounds better, despite having a similar bottleneck so people start rolling their eyes and making up stories about "pleasing distortion" and subjectivists adding sweetness to their distortion boxes.
Meanwhile, I've noticed there is some technical truth to masking making certain things sound better. And it's purely physical, not psychological. If you have IMD, or specifically some amplitude modulation, and you add a steady tone that matches the centre frequency of the 'problem' sound, that will reduce the dB level of the unwanted modulation. Draw it on graph paper yourself and you will see that it's deceptively simple and not magic!
E.g. Sketch a sine wave with strong modulation so the tremolo peaks vary from ±1 to ±10. That's 20dB of modulation. Then add a steady ±5 sine on top. By adding HD (which could be a tiny amount) you've now stabilized the sound and the ratio of modulation is now only 15:6 or 8dB.
What happens if a SE amplifier is optimized for energy efficiency by making it push-pull? The harmonics cancel out... Some of the time. This is a likely case of "the better it measures, the worse it gets". One naive approach would be to get output transistor (or tube) matching as close as possible. But the likely outcome is that the parameters momentarily change, so the harmonics oscillate between near-perfect cancellation and not-so-perfect cancellation. Highly optimised and balanced LTPs? Same problem. Then people say, oh we tune them by machine, not by hand. Human ears aren't sensitive enough to select the lowest distortion... But, they don't pause to think, do they? Maybe human ears are indeed the ultimate arbiter to decide the best ratio of orthogonal parameters like steady HD vs amplitude modulation?
Nelson Pass wrote about IMD getting more complex as more gain stages are stacked in series. This is one thing I'm keeping in mind for myself. It's awfully tempting to go from 2 up to 3 stages, with the simulator promising small fractions of a % improvements in HD, when it may not be desireable. Same with LTP gain stages vs their emitter-feedback poorer cousins, with lower gain, higher distortion, but also less to mess up.
The funny things is that some questions comes from people that like to play with virtual world and not on desk testing the circuit and see what happen modifyin live some parameters.
The people who test the real circuit normally posts the results with some technical consideration.
It is evident that a single shot of FFT ( virtual) can't say anything; and is not possible to predict anything about sound
In the real world!
Walter
The people who test the real circuit normally posts the results with some technical consideration.
It is evident that a single shot of FFT ( virtual) can't say anything; and is not possible to predict anything about sound
In the real world!
Walter