Yeah! Graham is wonderfull, but as everybody, has a small defect!
He do not like to talk about the new developments that is happening inside his laboratories.
I receive some secret fax, from my spyes, that something very interesting, that sound very, very, very good, are beeing produced there under top secret.
Someone told me, also, that there are more than 6 foot tall guards surrounding his house 24 hours a day, telephone disconnected because people trying to have informs to publish.
The codename is Pearl, a revoluctionary design, to make us astonished.
I tried hard to have informs, but could not enter his Mail, i was blocked not to insist.
I expect you may be more lucky than I.
A genius is working guys.... i am sure, do not expect some different CCS, ore some twisted long tail, or double something.
I am tired to ask him, he may already sold the design to some foreign Government or some University.
And do you think i am Joking, no, i am not!
regards,
Carlos
He do not like to talk about the new developments that is happening inside his laboratories.
I receive some secret fax, from my spyes, that something very interesting, that sound very, very, very good, are beeing produced there under top secret.
Someone told me, also, that there are more than 6 foot tall guards surrounding his house 24 hours a day, telephone disconnected because people trying to have informs to publish.
The codename is Pearl, a revoluctionary design, to make us astonished.
I tried hard to have informs, but could not enter his Mail, i was blocked not to insist.
I expect you may be more lucky than I.
A genius is working guys.... i am sure, do not expect some different CCS, ore some twisted long tail, or double something.
I am tired to ask him, he may already sold the design to some foreign Government or some University.
And do you think i am Joking, no, i am not!
regards,
Carlos
Hi Lumanauw,
The voltage at the VAS is not the same as that at input or output. It carries additional high frequency components necessary to compensate for output stage bias commutation and is not linear when loudspeaker current flow is high, especially when that current is out of phase with input signal voltage.
Last year I here suggested that it is input to output first cycle distortion mechanisms that need to be observed and not continuous sinewave.
I am very well aware that suddenly starting sinewaves give rise to components that are discontinuous and not audio, but it does reveal the amplifier errors that arise due to inherent slowness to correct back EMF modified output voltage.
Hi Carlos,
You have just sent me an info page about class-D. (Thanks) You could post it here.
Are we all wasting our time, and as Carlos suggests, making circuits that are at least 30years out of date ?
As with a valve amplifier the voltage discrimination arises *after* the (switching filter) output inductance. Obviously the discrimination takes a finite time period, but it could now be faster than the NFB response of most pre-1990 hi-fi amplifiers, and it cannot be influenced by loudspeaker back EMF.
Cheers ........... Graham.
The voltage at the VAS is not the same as that at input or output. It carries additional high frequency components necessary to compensate for output stage bias commutation and is not linear when loudspeaker current flow is high, especially when that current is out of phase with input signal voltage.
Last year I here suggested that it is input to output first cycle distortion mechanisms that need to be observed and not continuous sinewave.
I am very well aware that suddenly starting sinewaves give rise to components that are discontinuous and not audio, but it does reveal the amplifier errors that arise due to inherent slowness to correct back EMF modified output voltage.
Hi Carlos,
You have just sent me an info page about class-D. (Thanks) You could post it here.
Are we all wasting our time, and as Carlos suggests, making circuits that are at least 30years out of date ?
As with a valve amplifier the voltage discrimination arises *after* the (switching filter) output inductance. Obviously the discrimination takes a finite time period, but it could now be faster than the NFB response of most pre-1990 hi-fi amplifiers, and it cannot be influenced by loudspeaker back EMF.
Cheers ........... Graham.
who am i to discuss with honorable Graham Maynard.
I have to hear and try to understand.
And use some words to make some noise...but cannot discuss really with you...have no condition!
I am watching people running to those small class D things, not hi fidelity, but highly accepted by common people.
If you think in top hi fidelity, of course analogue can reproduce the sound waves with perfection, and digital cannot, as sample never enougth, the waves re-constructed after digital to analogic may be a little bit out of correct shape...i am think about that too.
But i cannot close my eyes Graham.... those "invaders" are everywhere...entering TV Sets, portable recorder, Telephone..... Home Systems and everywhere we need some audio amplification.
If we try to see the market tendences, it is clear, that everybody will jump to class D as it is commercial, good to produce, to sale, small, ligthweigth, good power, and quality that common mortal are accepting without any discuss or doubt...so...it is good to sale, to produce, to make money, as will be accepted.
Of course, in the middle to the top level conference of people alike you, those "things" will never enter, as too much "markeatable", not serious enougth to enter some Graham and top Enginners meeting...i can realise that, and i think you are not wrong.
But for the ones, that want to make money with those things..have to run fast in that direction, as this is beeing shown by Market.
regards,
Carlos
I have to hear and try to understand.
And use some words to make some noise...but cannot discuss really with you...have no condition!
I am watching people running to those small class D things, not hi fidelity, but highly accepted by common people.
If you think in top hi fidelity, of course analogue can reproduce the sound waves with perfection, and digital cannot, as sample never enougth, the waves re-constructed after digital to analogic may be a little bit out of correct shape...i am think about that too.
But i cannot close my eyes Graham.... those "invaders" are everywhere...entering TV Sets, portable recorder, Telephone..... Home Systems and everywhere we need some audio amplification.
If we try to see the market tendences, it is clear, that everybody will jump to class D as it is commercial, good to produce, to sale, small, ligthweigth, good power, and quality that common mortal are accepting without any discuss or doubt...so...it is good to sale, to produce, to make money, as will be accepted.
Of course, in the middle to the top level conference of people alike you, those "things" will never enter, as too much "markeatable", not serious enougth to enter some Graham and top Enginners meeting...i can realise that, and i think you are not wrong.
But for the ones, that want to make money with those things..have to run fast in that direction, as this is beeing shown by Market.
regards,
Carlos
Hi, GM,
This back EMF is more complicated than I think.
After experimenting and find this back EMF is important, I have a simple question. What's the cure? The case is in ordinary feedback amp, with 3 stages.
I have tought of possibilities. Very fast amp. ClassA output stage. Special cct design. Or something else? I don't know the cure.
This back EMF is more complicated than I think.
After experimenting and find this back EMF is important, I have a simple question. What's the cure? The case is in ordinary feedback amp, with 3 stages.
I have tought of possibilities. Very fast amp. ClassA output stage. Special cct design. Or something else? I don't know the cure.
Hi David Lumanauw,
I have followed your 'diyAudio' enquiries and information seeking with interest, and I must admit that I have been sitting back for some time to see if any new answers might appear.
You ask. Very fast ? Class-A ? It is the fast and unfortunately big plus hot running class-As that are the very amplifiers that are less upset by back EMF !
We have new designs by Susan and Pavel but so far these are not of higher power. MikeW is working on a 10 output device design, and Nelson is working on a JLH type of design. I myself have a novel design based on the JLH which our enthusiastic Carlos built in a day and had working with devices he had to hand, though he is now studying class-D.
My personal rules for 'AUDIO' as opposed to audio frequency amplification are;-
Low propagation delay; i.e. low path capacitance and fast transistors.
No signal/NFB delaying C.dom with good open loop frequency and phase responses prior to NFB; for stability this means a maximum of two stages, as with the JLH class-A.
NFB to minimise forward distortion, and to damp a dynamic loudspeaker.
A very low NFB loop generated output impedance to protect the amplifier itself from phase shifted loudspeaker back EMF.
Constant class-A operation so that there is no break of gainful conduction for either output half no matter what the loudspeaker reaction, and thus no possibilty of reverse driven bias commutation prior to class-AB overload or high power operation.
To improve overall efficiency, as with a Quad Current Dumper, have a class-AB or class-B output arrangement running simultaneously, or even both together with the class-A. All conduction increases and switch-offs to be staggered so that there is a smooth transconduction characteristic, unlike parallel devices where all switch and glitch together.
All of these arrangements are in my new circuit, as per attached photos of my Class-AABB amplifier. It too has 10 output devices and is currently quiescing at 80W. So far it has peaked about 150W levels. Design is for at least 200W - 4 ohms, and it has a really authorative immediacy on cleanly amplified transients.
Note that this is only an open prototype arrangement so that I can try different mods and updates, the wood is 2ft/60cms long.
Yes there are two 2SC3241driver transistors on the circuit board, these are in a signal driven active dynamic bias arrangement which increases/decreases output current on a per-half basis to maintain input signal voltage waveform at output irrespectively of momentarily asymmetric loudspeaker load demands up to 15A (or more).
I will reveal the circuit shortly when I am completely happy with progress at intended maximum power. That's if my loudspeakers and ears last out !
Cheers ........ Graham.
Cheers ........... Graham.
I have followed your 'diyAudio' enquiries and information seeking with interest, and I must admit that I have been sitting back for some time to see if any new answers might appear.
You ask. Very fast ? Class-A ? It is the fast and unfortunately big plus hot running class-As that are the very amplifiers that are less upset by back EMF !
We have new designs by Susan and Pavel but so far these are not of higher power. MikeW is working on a 10 output device design, and Nelson is working on a JLH type of design. I myself have a novel design based on the JLH which our enthusiastic Carlos built in a day and had working with devices he had to hand, though he is now studying class-D.
My personal rules for 'AUDIO' as opposed to audio frequency amplification are;-
Low propagation delay; i.e. low path capacitance and fast transistors.
No signal/NFB delaying C.dom with good open loop frequency and phase responses prior to NFB; for stability this means a maximum of two stages, as with the JLH class-A.
NFB to minimise forward distortion, and to damp a dynamic loudspeaker.
A very low NFB loop generated output impedance to protect the amplifier itself from phase shifted loudspeaker back EMF.
Constant class-A operation so that there is no break of gainful conduction for either output half no matter what the loudspeaker reaction, and thus no possibilty of reverse driven bias commutation prior to class-AB overload or high power operation.
To improve overall efficiency, as with a Quad Current Dumper, have a class-AB or class-B output arrangement running simultaneously, or even both together with the class-A. All conduction increases and switch-offs to be staggered so that there is a smooth transconduction characteristic, unlike parallel devices where all switch and glitch together.
All of these arrangements are in my new circuit, as per attached photos of my Class-AABB amplifier. It too has 10 output devices and is currently quiescing at 80W. So far it has peaked about 150W levels. Design is for at least 200W - 4 ohms, and it has a really authorative immediacy on cleanly amplified transients.
Note that this is only an open prototype arrangement so that I can try different mods and updates, the wood is 2ft/60cms long.
Yes there are two 2SC3241driver transistors on the circuit board, these are in a signal driven active dynamic bias arrangement which increases/decreases output current on a per-half basis to maintain input signal voltage waveform at output irrespectively of momentarily asymmetric loudspeaker load demands up to 15A (or more).
I will reveal the circuit shortly when I am completely happy with progress at intended maximum power. That's if my loudspeakers and ears last out !
Cheers ........ Graham.
Cheers ........... Graham.
Generally sensible guys say that 0.1% is possible to hear,
but not less than this.
It should be noted that loudspeakers are the weak link in todays systems.
They produce fairly much distortion.
Then we have the room factor - sounds will bounce at floor and walls.
Klippel has designed a very nice blind listening test for one's own distortion audibility threshold. Get some good headphones, find a quiet time alone (a night when there are no ambient sounds) and time to concentrate. You have to listen hard and carefully. But the median threshold is -21dB. A staggeringly high value of distortion and probably the basis of why many things are spec'd at 10%THD. However, there are those of us with a more acute sense of perception for distortion and it makes a world of difference to have an amp with lower distortion. I tested at -45dB, still not as high as some folks with golden ears who can detect -54dB.
Listening Test
Take the test and you might be surprised at the results (one way or another). It uses several types of distortion and the test is repeated in different orders to make sure that the results are not random guess /luck based.
Within distortion types, the even order harmonics are much less noticeable, and euphonic even - so it's hard to put a single number on distortion. Always better to see an FFT showing where the distortion lies. I would rather hear an amp with 0.05% THD and most of it on 2nd order and a bit less on 3rd order and nothing else than an amp with 0.005% THD with dominant 3rd order and 5th order and less 2nd or 4th order.
As an example of what I think sounds great to my ears is an amp with say 0.006% THD and dominant 2nd order with a bit of 3rd order harmonic distortion. Here is the output of the Aksa Lender preamp driving a 7kohm load at 40Vpp (enough for making 25w rms into 8ohms using a 0dB gain power buffer amp like an F4, MoFo, etc). It's also important that amps have as little 50/60hz mains hum as possible and you can see here it is flat like a battery.
Listening Test
Take the test and you might be surprised at the results (one way or another). It uses several types of distortion and the test is repeated in different orders to make sure that the results are not random guess /luck based.
Within distortion types, the even order harmonics are much less noticeable, and euphonic even - so it's hard to put a single number on distortion. Always better to see an FFT showing where the distortion lies. I would rather hear an amp with 0.05% THD and most of it on 2nd order and a bit less on 3rd order and nothing else than an amp with 0.005% THD with dominant 3rd order and 5th order and less 2nd or 4th order.
An externally hosted image should be here but it was not working when we last tested it.
As an example of what I think sounds great to my ears is an amp with say 0.006% THD and dominant 2nd order with a bit of 3rd order harmonic distortion. Here is the output of the Aksa Lender preamp driving a 7kohm load at 40Vpp (enough for making 25w rms into 8ohms using a 0dB gain power buffer amp like an F4, MoFo, etc). It's also important that amps have as little 50/60hz mains hum as possible and you can see here it is flat like a battery.
Last edited:
Wow, you restarted a 13-year-old thread!
For what it's worth, I think the distortion and other inaccuracies of the whole signal chain should ideally be inaudible, which means that the distortion of one single amplifier in the chain should not only be inaudible, but should be way below the threshold of audibility.
The problem with many tests is that they lack scientific rigour. You will find tests on YouTube which demonstrate that you can't hear phase shift for example. However with sound editing software you can easily prove that phase is audible by changing the phase of harmonics on a sine wave. Views that anything beyond the audible bandwidth don't matter, is another common perspective. However DAC's that have switchable filters that operate beyond the audible bandwidth have clearly audible effects.
My experience is that dynamic changes are at least four times as audible as steady state behaviour. This might mean that -45dB could be -57dB and -54dB is -66db. When I modified a valve amp to reduce its distortion at low powers to 0.006% it sounded the same as a MOSFET with similar distortion. Another two amps at 0.12% and 0.08% respectively, don't sound the same.
My experience is that dynamic changes are at least four times as audible as steady state behaviour. This might mean that -45dB could be -57dB and -54dB is -66db. When I modified a valve amp to reduce its distortion at low powers to 0.006% it sounded the same as a MOSFET with similar distortion. Another two amps at 0.12% and 0.08% respectively, don't sound the same.
Do you know of any scientifically rigourous tests that show that and if so, could you please provide literature references? The only blind listening test I know of that showed that ultrasonics might have an impact is a Japanese experiment with electroencephalograms from the 1990's:
Tsutomu Oohashi, Emi Nishina, Norie Kawai, Yoshitaka Fuwamoto and Hiroshi Imai, "High-frequency sound above the audible range affects brain electric activity and sound perception", Audio Engineering Society preprint 3207, presented at the 91st Convention, October 1991
Japanese gamelan players listened to a recording of Balinese gamelan music low-pass filtered at 26 kHz with or without a super tweeter that only reproduced the sounds above 26 kHz. Their brain waves were monitored with EEG equipment and were different when the super tweeter was on. It wasn't clear to me from the article whether the test was double- or only single-blind (which would render it non-rigourous).
By the way, recordings that have been peak sample normalized, like many commercial recordings, can drive digital interpolation filters without headroom into hard clipping (so-called intersample overshoots). This issue is probably less severe with smooth than with steep filters. That could be an alternative explanation for the perceived differences.
Strange to wake up a thread after 13 years. That's some deep grave digging. 
John Curl mentioned 13 years ago that the harmonic structure matters in hearing distortion. There are tests going back about 90 years that support his claims. There were some tests done in Japan in the 70s (I think) that linked certain harmonic distortion spectra with the sound of amps. Alas, I've not been able to find the original work.
It would be nice to see a rigorous study about the audibility of the structure of harmonic distortion. E.G., are odd harmonic really worse? Do lower harmonics mask upper harmonics? Is the perception of HD the same across the audio band? Etc.
John Curl mentioned 13 years ago that the harmonic structure matters in hearing distortion. There are tests going back about 90 years that support his claims. There were some tests done in Japan in the 70s (I think) that linked certain harmonic distortion spectra with the sound of amps. Alas, I've not been able to find the original work.
It would be nice to see a rigorous study about the audibility of the structure of harmonic distortion. E.G., are odd harmonic really worse? Do lower harmonics mask upper harmonics? Is the perception of HD the same across the audio band? Etc.
I think these days it is not worth getting too focussed on scientific tests. Most of what you need to know can be determined with a decent sound card and editing software. In the past the people who have conducted tests have not acknowledged that dynamic behaviour is an important part of what we hear. Constructing a test to measure this is near impossible. Also scientific tests must be reproduceable so that anyone, anywhere can conduct them.
I posted some ago a bandwidth limit test which is simply to listen to a 1KHz square wave, bandwidth limited and wide band. I found that the wide band signal could be detected by a change of pressure on the ear drum, compared with the bandwidth limited version (headphones). Of course, you now have the interpretation - is this important?
Last edited:
Then there is also phase information that provides stereo imaging. We may hear amplitude only to 20kHz but psychoacoustically, we can detect phase shifts as small as several uSec. Thus, audio amplifiers with flat phase up to 300kHz will have better stereo imaging than amps that don’t. The pinna has asymmetric folds of order 5mm and the purpose is to introduce minute phase shifts so that even with one ear, we can spatially locate sound sources and know that it is located above, below, front or back. There was also a study conducted that measured the ears ability to perceive phase shifts of this order.
Last edited:
As David Blackmer points out it is odd pieces of data, not necessarily what we know and measure as distortion presently. YouTube
My understanding is distortion in equipment is partially a artifact that arises from incorrectly presenting time information.
Bandwidth of audio equipment also has to escape the Nyquist / Shannon mentality
of limits that apply to digital mediums, https://www.earthworksaudio.com/wp-content/uploads/2012/07/The-world-beyond-20kHz.pdf
My understanding is distortion in equipment is partially a artifact that arises from incorrectly presenting time information.
Bandwidth of audio equipment also has to escape the Nyquist / Shannon mentality
of limits that apply to digital mediums, https://www.earthworksaudio.com/wp-content/uploads/2012/07/The-world-beyond-20kHz.pdf
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.