Quite reassuring ;-D - thanks!!
When someone has a speaker system that when this amp is connected and outputting 0,2mW (milli watts) can detect 0,005% THD+N, I will eat my old hat.
If the speaker system produces 100dB/W/m the sound pressure at 3 meter, receiving 0,2mW, will be: 56,5dB - a very low level - like a normal conversation. Now, the distortion components being 0,005% would lie a further -86 dB down.... i.e. completely ridiculously low absolute level and about 45 dB under any noise floor of a very quiet room. And surely way below any human hearing capability - even for the golden 1 ppm of the population. Even Mk4 could not hear this 😉
The same speaker system will require 150W to produce the level (@3m) of an orchestra crescendo at 3 row center which hits about 115dB.
//
When someone has a speaker system that when this amp is connected and outputting 0,2mW (milli watts) can detect 0,005% THD+N, I will eat my old hat.
If the speaker system produces 100dB/W/m the sound pressure at 3 meter, receiving 0,2mW, will be: 56,5dB - a very low level - like a normal conversation. Now, the distortion components being 0,005% would lie a further -86 dB down.... i.e. completely ridiculously low absolute level and about 45 dB under any noise floor of a very quiet room. And surely way below any human hearing capability - even for the golden 1 ppm of the population. Even Mk4 could not hear this 😉
The same speaker system will require 150W to produce the level (@3m) of an orchestra crescendo at 3 row center which hits about 115dB.
//
But please show it also not only for 1kHz but also for higher frequencies and what could be revealed is perhaps not so much THD but rather perhaps N - this would adress the "mushy hash" comment above... and why not multitone...?
//
//
Nick Sukhov said:
yes, and my small addition - hearing only tweeters (woofer and mid replace with power resistors).
yes, and my small addition - hearing only tweeters (woofer and mid replace with power resistors).
yes, really all who compare this way, stopped using D. Even without AP555.
AP555 couldn't measure anything in my case, because its threshold -120 dB and mine -138. I use RAW with special 2T notch bridges.
https://www.patreon.com/posts/zachem-azh-3-2t-102965745
https://www.patreon.com/posts/138-db-v-umzch-100439556
https://www.patreon.com/posts/154-db-parshivka-100869890
Wants FLAC 24bit\96kHz - please https://www.patreon.com/posts/naskolko-pink-i-105175659
https://www.patreon.com/posts/105278609
Unfortunately, that's exactly how I feel, dear Nick.
There is a language barrier, I hope we can overcome it.
Thanks for the ABBA concept, it's quite interesting and a perennial favorite anyway.
kind regards,
HBt.
That graph shows me typical class D amps imperfection ,I consider such class D amp that have 0,1% THD&N characteristic (which is nothing special IMHO) , regardless that in range between 1W and up to full power THD&N is much lower , also I will be interested to see what is the real level of output RF noise spectrum at full output power near to clipping or even at clipping point .
Attachments
Cool - but isn't it time to leave the 1 tone analysis.. or at least complement it with multitone?
//
Dear @Nick Sukhov
I have now watched your introductory video again - and believe me, for me as a German this is (now) a real challenge, listening, watching and reading the autotext is enormously exhausting, especially if you are deeply involved in the matter yourself and try to understand what you as an EE want to tell other EEs or DIYers.
It would be nice if you would release your schematic here, then each of us can play with our favorite simulation tool or implement real test setups.
Please don't get me wrong, but as much as I like the pictures of the finished product and they make me curious, the chosen, knowing, apodictic, monotonous tone of voice in connection with MC12 is very tiring.
Nevertheless, I like your projects - and I don't want to initiate any ranking battles, euphoric discussions and comparisons with competing developments are generally unnecessary, misunderstandings are simply programmed here.
I assume that your new 2 x 150 watt audio amplifier will work flawlessly - and will find a lot of friends and approval.
The concept is very exciting.
#
What I would be very happy about would be:
if you would invest your knowledge and commitment in an open forum project, I have in mind a completely discrete solution of the ancient Sandman concept, in the modification of the Matsushita engineers.
In other words, a 25 to 50 watt audio amplifier with the Technics Class AA (VC-4) seal of approval.
with your guidence
regards,
HBt.
I have now watched your introductory video again - and believe me, for me as a German this is (now) a real challenge, listening, watching and reading the autotext is enormously exhausting, especially if you are deeply involved in the matter yourself and try to understand what you as an EE want to tell other EEs or DIYers.
It would be nice if you would release your schematic here, then each of us can play with our favorite simulation tool or implement real test setups.
Please don't get me wrong, but as much as I like the pictures of the finished product and they make me curious, the chosen, knowing, apodictic, monotonous tone of voice in connection with MC12 is very tiring.
Nevertheless, I like your projects - and I don't want to initiate any ranking battles, euphoric discussions and comparisons with competing developments are generally unnecessary, misunderstandings are simply programmed here.
I assume that your new 2 x 150 watt audio amplifier will work flawlessly - and will find a lot of friends and approval.
The concept is very exciting.
#
What I would be very happy about would be:
if you would invest your knowledge and commitment in an open forum project, I have in mind a completely discrete solution of the ancient Sandman concept, in the modification of the Matsushita engineers.
In other words, a 25 to 50 watt audio amplifier with the Technics Class AA (VC-4) seal of approval.
with your guidence
regards,
HBt.
What about at 10kHz, and what is the switching frequency rejection.
That said at 1kHz things are very easy, see below for 200W in red and 1W RMS in green on 8R.
Attachments
Nick's patented ABBA is but a new name for Yamaha's Linear Transfer Bias from late 1970s. The basic idea is to run several pairs of output devices at different quiescent currents. Here is an implementation example (Yamaha M-2):
Nick uses modern transistors, and some resistors are different.
BTW there is no such thing as Gm doubling. There is a nice, if a bit math-heavy, note on this topic by Marshall Leach (attached). TL;DR, the "doubling" idea is based on an incorrect assumption that the emitter current in each transistor in an output pair is independent from the other transistor.
Did you notice that 0.1% THD+N is for 10 uW output level?
Essentially, it’s just noise at extremely low level. Not even Benchmark AHB2 or Topping B200/B100 are close.
I know and use English descriptions and subtitles where I can. But where they are not there, you can use translate.google.com
for example: https://photos.app.goo.gl/FrS5qSZyQygGXVpr5
A universal test for any amplifier is the direct subtraction of its output signal from a linear (LTI) distortionless copy of itself, the residual being pure distortion/error signal
The linear "copy" is realized by measuring a precision impulse response of the amp (at speaker terminals) and using that as a convolution kernel for the input signal. This is then compared to a recording of the real signal (again, at the speaker terminals). The source signal and level used for obtaining the impulse response must be well-chosen, and some preprocessing might be beneficial, like averaging.
Sort of a null test and it works with any signal, giving easily quantifiable and repeatable results. With sample-synced record-while-playback and good tools like REW and DeltaWave the resolution than can be achieved is breathtaking.
Further, we can also us the output of a second, totally different amplifier (say, any best-effort non class-D) and use that as the reference. The trick for subtraction is that each amp is fed with the source signal convolved with the impulse response of the other amp.
The alleged Class-D hash would be easily exposed. That's the kind of proof I would expect to be presented by the supporters of the universal(?) hash characteristic (even for the best of today's offerings? I mean, really?).
Once we have exposed the hash residual we can even add that to the playback of the best-effort reference amplifiers and see if it translates... it should, and that would be a valuable additional data point what's going on with that dubious hash claim.
OK, I will soon add German subtitles to my videos. Sorry, I do not know German ...