^ I think you may have hit the nail on the head as to why some people have a problem with it
Not quite.
As you have seen stated by Bruno - "If you test a magnetic core with a sinewave the distortion looks a little like soft clipping, perfectly benign. But what came out of tests on iron parts in loudspeakers was that hysteresis has a long term memory so you can get intermodulation between things that happen now and things that happened 10 minutes ago. With music this distortion sounds like half correlated noise."
Maybe a different test signal or different approach is needed for magnetic cores & JN could pitch in here?
But the general principle is that not all conditions that are met in audio are necessarily covered by the standard sinewave testing & FFT analysis
No, it is not awesome. The thesis is quite poor and useless with current designs.
Read the full interview (it's worth reading), rather than carefully selected out-takes....
Good advice
Scott, if you read the interview you will see this about feedback - I've quoted the whole exchange so as not to be accused of "carefully selected out-takes"
Executive analysis - loop gain of 75DB all the way upto to 20KHz means that high amounts of feedback can operate right up to 20KHz & not run out before this.
As he says "I suspect most basic class AB amps have a loop gain of 75 dB, but only up to 100Hz or so, then it starts dropping."
"For a linear amp design with a standard 1st order compensation to hit 75 dB at 20 kHz it’d have to have a Gain Bandwidth Product of 110MHz. Two-pole compensation is the highest I’ve ever seen in a class AB amplifier and even then 76 dB at 20 kHz is pretty astronomical."
So, yes, Scott feedback reduces IMD but it seems of limited effectiveness in a lot of normal amplifiers
My new Halo was hooked up by the time he came around, but the halo is now in for diagnosis/repair so the yammie is back in play.
Only issue I’ve got now is my replacement tweeters (although same design) won’t take the abuse like the original burhoes.
Thanks to pavels 20/22khz test tone one was relegated to ‘Harley parts’ box
So unless I find a original replacement it might not be repeatable.
Maybe we need to take a better look at our amp design roots?
May be he is talking about feedback? Tubes/fet, when zero feedback, distortion is only H2. When you increase fb the high orders also increased but then fall again with lots of feedback. BJT otoh, high orders are high even without fb (I have never liked class A bjt amps).
Psychoacoustics to audio is like quantum theory to Physics (if you know what i mean).
The threshold is not the point, imo. Psychoacoustics will talk about masking, correlated versus uncorrelated noise/distortion.
It seems to me that I and Bruno have a little similarity in design approach? There is issues with high order distortions. Choosing tube/mosfet and zero feedback designs will eliminate this issue but there are other issues with H2.
So, feedback design is chosen. In order to avoid the issues with high order distortions, the feedback is maximal. This is possible with using high bandwidth transistors. When we maximize feedback and open up the amp to ultrasonic, we need to know how to deal with it, especially understanding stability and compensation. Bruno talks about advanced loop control. Basically this is to master stability art because that is the consequence you're willing to take when you approach amp design this way.
And how exactly did you draw this conclusion from what Bruno (correctly, BTW) said?
Sound Separation
Have you read reviewers talk about sound separation? Usually using words such as "you can follow each instrument individually". When I read Srajan's review about Bruno's class D amp I didn't see this aspect mentioned. I suspect, class D is not good in term of this quality aspect. I'm happy because I found this quality aspect is even more important than 'transparency' and other aspects. Currently I'm investigating this 'separation' aspect. Once I know how to maximize this quality aspect, it will be a killer amplifier.
Have you read reviewers talk about sound separation? Usually using words such as "you can follow each instrument individually". When I read Srajan's review about Bruno's class D amp I didn't see this aspect mentioned. I suspect, class D is not good in term of this quality aspect. I'm happy because I found this quality aspect is even more important than 'transparency' and other aspects. Currently I'm investigating this 'separation' aspect. Once I know how to maximize this quality aspect, it will be a killer amplifier.
Fourier transforms of steps and impulses. ( and you can multiply the step and the sines to give sines that only last as long as the step, and then multiply by an exponential to give a decay, etc.).
TheFourierTransform.com - Step Function and the Signum Function
And then you can look at the list of transforms on the next page.
Learn the math before you poo poo it.
I dont understand how people who use, or believe FFTs are real, and can acuratly measure THD, and IMD, dont think fourier transforms can be applied to music.
Dunno, how about photosynthesis as an analogy? We don't know quite how it works but it does.
Thresholds are still a large part of that though, unless I'm missing your point?
The second part is more to the point (new ways of measuring), imo. Threshold is about scalars, magnitudes, something irrelevant when you compare correlated versus uncorrelated noise/distortions, for example.
Last edited:
Quite the contrary, the Fourier transform assumes infinite sine waves. Window functions are used to make the Fourier transform work (to a good approximation) for finite-length samples.
No, it doesn't. One needs to assume periodicity only if wants a Fourier series representation, expressed as an infinite sum of sines. For aperiodic signals, the Fourier transform provides a representation in which the signal is, in the general case, an infinite sum of infinitesimal sines. Otherwise said, for an aperiodic signal, the Fourier coefficients are integrals rather than closed form expressions. A Fourier inverse transform exists and can be used to re-build the original aperiodic signal from its Fourier transform.
The reason why Fourier series coefficients are commonly used in engineering (from audio to optical spectra analysis) is that switching to the Fourier transform doesn't provide any extra insight, for the common electrical/optical time invariant circuits.
If you think audio, specifically, would benefit from such a different (aperiodic signals Fourier transform) approach, then this would be one of those extraordinary claims that needs extraordinary proof before engaging in further work. Until such extraordinary proof is provided, we tacitly extend our signals to periodicity and use our spectrum analyzers to determine circuits performance.
Last edited:
@benb,
You might find it more intuitive to start with something basic: discrete fourier transform tutorial - Google Search
You might find it more intuitive to start with something basic: discrete fourier transform tutorial - Google Search
Human communication is context based. I considered the context given by the modified PCB inside of the DAC and its "EMC modifications" when answering syn08's question about the "smallish ICs" .
I responded with:
you might note the "Could be" at the beginning, which could (at least I hoped so) in a sufficient way indicate the speculative nature of my statement.
So, I don't know and I didn't pretend to know.
IOW, what you and syn08 were trying was classical FUD although you might have to reread the (classic) definition.
This is a pretty good summary of it, I think.
http://complextoreal.com/wp-content/uploads/2012/12/fft4.pdf
Last edited:
It is still useful to remember that measurement with sine-waves (you know single-tone or pure-tone ) in the usual way (means using a distortion analyzer or voltmeter) is considered as so-called "steady state" measurement.
@ gpauk,
Besides, that it is anecdotal information, the argument about the already sufficient measured approximation to perfection is used at least for four decades, usually in combination with "that now ......." while the former generation (back then already near enough to perfection) wasn't .
Although it will be most probably true someday (maybe indeed already today) the reasoning as such is questionable.
) in the usual way (means using a distortion analyzer or voltmeter) is considered as so-called "steady state" measurement. @ gpauk,
Mmmmerrrrilll? Pedantic? Shirley not....
Bruno also says he thinks that the Purifi amps are pretty close to perfect.
Looking at the way they measure, it's hard to disagree on any sane basis.
Certainly close enough that any discussion of what needs fixed moves to speakers.... Which is hardly rocket science...
Besides, that it is anecdotal information, the argument about the already sufficient measured approximation to perfection is used at least for four decades, usually in combination with "that now ......." while the former generation (back then already near enough to perfection) wasn't .
Although it will be most probably true someday (maybe indeed already today) the reasoning as such is questionable.
Human communication is context based. I considered the context given by the modified PCB inside of the DAC and its "EMC modifications" when answering syn08's question about the "smallish ICs" .
I responded with:
you might note the "Could be" at the beginning, which could (at least I hoped so) in a sufficient way indicate the speculative nature of my statement.
So, I don't know and I didn't pretend to know.
IOW, what you and syn08 were trying was classical FUD although you might have to reread the (classic) definition.
Otherwise said, you accept the premise that shielding ICs does something audible for the DAC, then (while explicitely admitting you have no idea about the topic) try to extend this to smaller ICs.
Once again, if this is not FUD, then nothing else is. Of course, combined with an ad hominem directed to, and intended to discredit, those that do have an idea about the topic.
It's ugly and yet another proof of the disgusting intellectual dishonesty you are practicing on public fora. That, because I can't imagine you really believe in what you are preaching. You are not stupid enough for this.
It is still useful to remember that measurement with sine-waves (you know single-tone or pure-tone
More FUD. A "steady state measurement" is by definition a measurement taken on a DUT, when the DUT doesn't change in any way during the measurement. We call such a DUT a "time invariant system" and if somebody claims that audio electronics is not time invariant, it better come with some extraordinary proof before asking to be taken seriously. The most common examples of time variant systems (and for which a local time invariant approximation is not good enough) are the planet Earth thermodynamics and aircrafts in flight.
.- Status
- Not open for further replies.