Oliver and Locanthi gave us the correct version of Class AB in the last century. However, many have succumbed to ear-damaging Class D in the new century. This is while it's been several years since the improved versions of the old good Class AB were introduced. One of those are DoubleCross by Cordell/Hegglun that is at least order of THD better. Another one is my patented Class ABBA that is 30 times better than plain vanilla Class AB. The hardware costs for those improvements are negligible -- for Class ABBA it is just a pair of additional low power resistors. The detailed schematic description, PCB gerbers, BOM, and datasheets for my amplifier are all here:
https://www.patreon.com/posts/umzch-vv-xxi-v-81828557
The screenshots of AB/DoubleCross/ABBA simulation in easy-to-compare side-by-side layout are attached. The simulation was made using Microcap 12 that is probably the best simulator for audio circuitry. Endless thanks to its creator, Andy Thompson.
My master class with online analysis of all three circuits is available here (with multilingual subtitles):
Photos of printed circuit boards and assembled amplifier: https://www.diyaudio.com/community/...d-state-pics-here.96192/page-382#post-7799636
All Microcap *.cir files used for analysis can be found here:
https://www.patreon.com/posts/cravnenie-av-po-112445977
The copy of a free Microcap 12 install with all required libraries can be also found here:
https://www.patreon.com/posts/podstrakhoval-12-84096514
I hope we will overcome the Class D dominance together, won't we?
https://www.patreon.com/posts/umzch-vv-xxi-v-81828557
The screenshots of AB/DoubleCross/ABBA simulation in easy-to-compare side-by-side layout are attached. The simulation was made using Microcap 12 that is probably the best simulator for audio circuitry. Endless thanks to its creator, Andy Thompson.
My master class with online analysis of all three circuits is available here (with multilingual subtitles):
Photos of printed circuit boards and assembled amplifier: https://www.diyaudio.com/community/...d-state-pics-here.96192/page-382#post-7799636
All Microcap *.cir files used for analysis can be found here:
https://www.patreon.com/posts/cravnenie-av-po-112445977
The copy of a free Microcap 12 install with all required libraries can be also found here:
https://www.patreon.com/posts/podstrakhoval-12-84096514
I hope we will overcome the Class D dominance together, won't we?
Attachments
Jan, I ask a simple question. What is THD of the Purifi output stage before it is covered by NFB loop, especially taking into account the hysteresis of the low-pass filter. See https://www.patreon.com/posts/chem-class-d-ab-98110747
And I do not think Purify or even Eigentakt\Nilai will be able to overcome the dead zone of a pair of powerful output switch MOSFETs like DoubleCross or ABBA overcome Gm doubling. Is any class D amp has such a THD - https://photos.app.goo.gl/CVgavQRGdafz1X2b9 https://www.patreon.com/posts/59383957 ?
And I do not think Purify or even Eigentakt\Nilai will be able to overcome the dead zone of a pair of powerful output switch MOSFETs like DoubleCross or ABBA overcome Gm doubling. Is any class D amp has such a THD - https://photos.app.goo.gl/CVgavQRGdafz1X2b9 https://www.patreon.com/posts/59383957 ?
Last edited:
THD at the fifth decimal place or better simply has no effect on the final sound quality. First, our ears are not that sensitive to THD, and second, that current with a miserable THD passes through our speakers with an incomparably higher THD than anything else. That's where the story ends for me (and many others).
Chasing ultra-low distortion of any kind is more about measuring techniques than listening to music. I can't stand these various reviews where they talk about ultra-low THD figures (DACs mostly) and based on that conclude which device is better.
Chasing ultra-low distortion of any kind is more about measuring techniques than listening to music. I can't stand these various reviews where they talk about ultra-low THD figures (DACs mostly) and based on that conclude which device is better.
NIXIE62 , the thing is that our ears and speakers reduce their distortion when the sound level is reduced, while most amplifiers, on the contrary, increase distortion when the level is reduced due to the Class AB step or Class D dead zone and hysteresis. So here I agree with Nelson Pass FirstWatt, - "The FIRST watt is the most important watt."
In my case, it is the first 10W. 🤣
I listened to a lot of amplifiers, sometimes AB class sounds better than A class, there is something else much more important. I don't know exactly what, but i think it has nothing to do with THD. Not to mention outstanding tube amplifiers with tons of THD and miserable damping factor.
I listened to a lot of amplifiers, sometimes AB class sounds better than A class, there is something else much more important. I don't know exactly what, but i think it has nothing to do with THD. Not to mention outstanding tube amplifiers with tons of THD and miserable damping factor.
Nick, the Purifi doesn't have a wingspan, it is straight. There's no xover or Gm doubling like in (A)B.
We measured THD at 200W of something like 0.00003%.
I don't know the open loop distortion and frankly, I don't care. I only care about performance of the complete amp.
I appreciate your design, it is very good and well executed, my compliments!
But it is an illusion to think you can battle class D with it.
If there is an audible difference between a top of the line class D and an A(B) it is because the class D is much cleaner and therefor for some sounds 'clinical' because they are used to colored sound.
Jan
So to end the discussion maybe we should do a focus and put the microscope on the Purifi in the 0,01W-1W region... and not only talk about 1/5/200W performance...?
Can you do this please?
//
Almost exactly the same words they used to promote that dreadful early digital.
That headache you have is cause your ears are unused to the cleanliness 🙂
Yes, there also people that swear that 1948 shellac records are sounding the absolute best.
Many construct their own comfort zone and stay away from anything outside.
They never discover how much nicer it can be!
It's probably a basic human thing for security and comfort.
But let us not hijack Nick's thread....
Jan
Many construct their own comfort zone and stay away from anything outside.
They never discover how much nicer it can be!
It's probably a basic human thing for security and comfort.
But let us not hijack Nick's thread....
Jan
yes, and my small addition - hearing only tweeters (woofer and mid replace with power resistors). - Anyone immediately qiuckly will distinguish the mush of class D sound.
PS. But Jan is right - my thread is AB/DoubleCross/ABBA comparison not D ;-)
PPS to Jan - about your "1948" - compare this vinyl with digital master (attached), wich is better? - https://www.patreon.com/posts/naskolko-pink-i-105175659
Last edited:
CFA is the worst of all possible, and your screenshot is of amp with NFB loop. My thread is about output stage without feedback loop. Add 78
dB of loop gain @ 20 kHz to my 0.0015% and then compare.
Last edited:
That is an interesting way to compare, never thought about that. Good idea.
Come on, how can anyone judge which sounds 'better' looking at those graphs?- about your "1948" - compare this vinyl with digital master (attached), wich is better? - https://www.patreon.com/posts/naskolko-pink-i-105175659
Maybe you can mention a preference, for instance if you like a boosted mid/range range. But better?
Jan
The results are impressive, but I don't expect the bias current to be accurately maintained with conventional transistors. The -2mV/C temperature coefficient is hard to work around. The distortion cancellation will not be effective once the bias has drifted.Nick Sukhov said:
BTW, the noise measurements in class D amplifier reviews tell me all that I need to know. 😉
Ed
I fear the cancellation will become worse and worse at higher frequencies due to switching distortion - alas this is the regime where GNFB starts to tail off and can't help much.
It would be interesting to see measurements of physical output stages, as real devices often confound expectations gleaned from simulation. I have a suspicion that poorly matched output devices could provide some cancellation and perform better than well-matched devices (at least for cross-over artifacts - large-signal distortion less so?)
It would be interesting to see measurements of physical output stages, as real devices often confound expectations gleaned from simulation. I have a suspicion that poorly matched output devices could provide some cancellation and perform better than well-matched devices (at least for cross-over artifacts - large-signal distortion less so?)
First you claimed that class D amplifier is bad. Maybe because you never find good class D amplifier or cannot design good class D amplifier. Then Jan Didden prove your opinion wrong with measured the OUTPUT of his class D amplifier.
An amplifier judged by the output compare to the input, not the topology. No global feedback feedback or with global feedback it does not matter.
Congratulation on your patent Nick.
Reviewing Data and Videos posted.
My first introduction to Microcap
Thanks for the data archive.
Reviewing Data and Videos posted.
My first introduction to Microcap
Thanks for the data archive.
Back to the "THD wars". yeah !
CFA is not "bad"(below 1). Being so fast that it can actually correct the Xover "glitch". My design stays 5-8 PPM 20-70mA Re.
VFA's need to optimally biased @ 80mA. Deviation from this really kills the specs.
Class D can be just as good , I built a Hypex NC500 kit. It performed and sounded the same as the Wolverine.
It's 2 PPM was indistinguishable from the <1PPM of the Wolverine EF3 class AB.
What was nice about the class D was not needing big " boat anchor " heatsinks and power supplies ! SMALLER.
The Wolverine was cheaper - those Hypex's cost @ 400$ each.
Class D will win (commercially) , OEM's can manufacture them for 1/4 the cost , 1/4 the weight and still charge $2000.
The Purify and the Wolverine are tied @ .5PPM , so Class D is tied for performance. AB is easier to repair - that's why I like it.
The best solution is to use a class D for your bass , and a choice 60-80W AB for >80hz. Less Bulk - cheap.
But who cares ?? I sure can't hear the difference.... it's just subjective and psychological , at best.
OS
CFA is not "bad"(below 1). Being so fast that it can actually correct the Xover "glitch". My design stays 5-8 PPM 20-70mA Re.
VFA's need to optimally biased @ 80mA. Deviation from this really kills the specs.
Class D can be just as good , I built a Hypex NC500 kit. It performed and sounded the same as the Wolverine.
It's 2 PPM was indistinguishable from the <1PPM of the Wolverine EF3 class AB.
What was nice about the class D was not needing big " boat anchor " heatsinks and power supplies ! SMALLER.
The Wolverine was cheaper - those Hypex's cost @ 400$ each.
Class D will win (commercially) , OEM's can manufacture them for 1/4 the cost , 1/4 the weight and still charge $2000.
The Purify and the Wolverine are tied @ .5PPM , so Class D is tied for performance. AB is easier to repair - that's why I like it.
Quite easy to even beat it - even PPB at 1khz (below 2). The only drawback is bulk !
The best solution is to use a class D for your bass , and a choice 60-80W AB for >80hz. Less Bulk - cheap.
But who cares ?? I sure can't hear the difference.... it's just subjective and psychological , at best.
OS
