The similarities are existing indeed, unfortunately this kind of principle seems to be invented again and again and most people (including Yamaha with their later models) make the same mistakes again and again:
- The Bias current is choosen to be rather low (Very tempting indeed to do so), which means, that the power supply lines are polluted with high amounts of overtones , not much less than equally biased Class AB. This leads into the land of layout critical circuitry and a significant rise of distortion at lower impedances (At 16 Ohms these circuits simulate very well). Unfortunately the drawbacks of choosing low bias currents were never really well addressed in such designs (Probably only on the Sony TA-N902, which has a very careful layout and rather high bias).
- Some topologies like the Nelson Pass approach or Technics New Class A push the switching distortion problem from the output devices into other parts like diodes. This should then correctly called "Pseudo Class A" or "Class CA".
- In most approaches the input impedance is difficult to drive at high frequencies, because capcitive or dropping local loop gain effects relativate the advantages of the theoretically almost non existing crossover distortion, so finally the problem is again pushed into preceiding stages or into the need of stellar amounts of global feedback, which is known to be difficult to achieve with low order compensation schemes ...
The first one of these points Nelson Pass criticized about the so-called "Copy Cats" of his dynamic bias circuit like Technics, Yamaha, Pioneer, Denon etc. more or less for good reason. Ironically his first dynamic bias creation, the nowadays almost legendary 800A doesn't idle at very high bias as well (200W idle advertized, << 40W in real life) and it isn't by far free from the above described effects. Finally Peter Blomley probably had the right to call Nelson Pass a "Copy Cat", as he calls the "Japanese". IMO the so-called "Japanese" rather copied Blomley and Schwarz, than that they copied Pass ..
Neither Masao Noro of Yamaha, nor Nelson Pass of Threshold were the inventors of dynamic biasing (Peter Blomley was it most likely, with all the potential flaws of his first circuit). Unfortunately no one really implemented it completely right to date, except Yamaha in the MX-10000, which amongst other things idles the HCA stage at 7A. Unfortunately this is only possible because it is implemented at the expense of a separate "small" power amplifier stage, which runs at +-4V !
I don't say, that this kind of circuitry is impossible to realize, but if you don't go the way of high bias currents like Yamaha did in the MX-10000, it seems rather difficult to achieve the "text book" performance in a real amplifier with real world loads,a real world layout and real world components ...
This is certainly not suited for DIY ...
- The Bias current is choosen to be rather low (Very tempting indeed to do so), which means, that the power supply lines are polluted with high amounts of overtones , not much less than equally biased Class AB. This leads into the land of layout critical circuitry and a significant rise of distortion at lower impedances (At 16 Ohms these circuits simulate very well). Unfortunately the drawbacks of choosing low bias currents were never really well addressed in such designs (Probably only on the Sony TA-N902, which has a very careful layout and rather high bias).
- Some topologies like the Nelson Pass approach or Technics New Class A push the switching distortion problem from the output devices into other parts like diodes. This should then correctly called "Pseudo Class A" or "Class CA".
- In most approaches the input impedance is difficult to drive at high frequencies, because capcitive or dropping local loop gain effects relativate the advantages of the theoretically almost non existing crossover distortion, so finally the problem is again pushed into preceiding stages or into the need of stellar amounts of global feedback, which is known to be difficult to achieve with low order compensation schemes ...
The first one of these points Nelson Pass criticized about the so-called "Copy Cats" of his dynamic bias circuit like Technics, Yamaha, Pioneer, Denon etc. more or less for good reason. Ironically his first dynamic bias creation, the nowadays almost legendary 800A doesn't idle at very high bias as well (200W idle advertized, << 40W in real life) and it isn't by far free from the above described effects. Finally Peter Blomley probably had the right to call Nelson Pass a "Copy Cat", as he calls the "Japanese". IMO the so-called "Japanese" rather copied Blomley and Schwarz, than that they copied Pass ..
Neither Masao Noro of Yamaha, nor Nelson Pass of Threshold were the inventors of dynamic biasing (Peter Blomley was it most likely, with all the potential flaws of his first circuit). Unfortunately no one really implemented it completely right to date, except Yamaha in the MX-10000, which amongst other things idles the HCA stage at 7A. Unfortunately this is only possible because it is implemented at the expense of a separate "small" power amplifier stage, which runs at +-4V !
I don't say, that this kind of circuitry is impossible to realize, but if you don't go the way of high bias currents like Yamaha did in the MX-10000, it seems rather difficult to achieve the "text book" performance in a real amplifier with real world loads,a real world layout and real world components ...
This is certainly not suited for DIY ...
Conclusion for me is in the meantime, that all low bias audio amps (below 100mA for the output stage) will always be different in sonic performance in comparison to pure Class-A/single ended, no matter, which topologies and error corrections are in use.
Basiclly the same how to read by the third paragraph of
https://passlabs.com/articles/single-ended-class-a
Also the thread
http://www.diyaudio.com/forums/soli...ologies-100ma-idle-sound-closest-class-4.html
provides here no other findings.
Additional I note by several listening tests, that there is no justification for a very effort-designed front-end topology used for a low bias output buffer.
The topology by post #13 about
http://www.diyaudio.com/forums/solid-state/99967-anybody-have-scan-483b-motorola-2.html
provides best possible audible sonic quality in Class-AB. More complex circuits provides more excellent measured results (especially much lower values of THD), but from the audible view nobody hears distinct advantages (mainly through masking effects from the complex THD spectrum, generated in the output), if the quality is definitely right for the power supply of both versions.
But one question is still of interest for me regarded bi/multiamping systems resp. active loudspeaker systems:
below which frequency the benefits of Class A are no longer audible (I know exactly, below 300 Hz nobody can say, If there is pure Class-A or Class-AB power amp with only 30mA output bias is in use) ??
Basiclly the same how to read by the third paragraph of
https://passlabs.com/articles/single-ended-class-a
Also the thread
http://www.diyaudio.com/forums/soli...ologies-100ma-idle-sound-closest-class-4.html
provides here no other findings.
Additional I note by several listening tests, that there is no justification for a very effort-designed front-end topology used for a low bias output buffer.
The topology by post #13 about
http://www.diyaudio.com/forums/solid-state/99967-anybody-have-scan-483b-motorola-2.html
provides best possible audible sonic quality in Class-AB. More complex circuits provides more excellent measured results (especially much lower values of THD), but from the audible view nobody hears distinct advantages (mainly through masking effects from the complex THD spectrum, generated in the output), if the quality is definitely right for the power supply of both versions.
But one question is still of interest for me regarded bi/multiamping systems resp. active loudspeaker systems:
below which frequency the benefits of Class A are no longer audible (I know exactly, below 300 Hz nobody can say, If there is pure Class-A or Class-AB power amp with only 30mA output bias is in use) ??
Last edited:
Is the distortion caused from what the input line of amplifier catch the radiation of supply line ?
If so , all of conventional classAB amplifier have same, or worse the problem?
The HCA OPS of MX-10000 is feedforward OPS,
then it output only distortion negating signal voltage and sink output of main amp's signal current ,isn't it?
Is is need to be classA? to suppress the error of ERCO?
How do you think?
If its HCA OPS is conventional ClassAB and share the supply with main OPS ,
the supply current will not be much polluted, because it is BTL.
...The Yamaha MX-10000 was the only amp, where true HCA was implemented, the other so called HCA amps (MX-1000, 2000 etc.) were more or less engineered by marketing department ...
I can't completely agree with some of these statements about the these amps, the comments are not entirely conclusive. I've owned mint specimens of each of them to determine for myself. The 1000 is very nice, its somewhat warm and can swing large voltages and has just shy of 45A available for each channel. The MX-2000 is anything but watered down from its 10000 brethren. Rather, its a smaller serving for those who seek appropriate portions. I've used mine quite a bit with the Quad ESL-57s and in the past a pair of Apogee Duettas and it doesn't seem to care what load was paired to it. It's a fantastically clean sounding piece of kit, quality construction and you can hear a pin drop. I had some friends over one night and we went until about 5AM, it just sounded that good. It draws 150 Watts from the outlet sitting idle and dynamically increases for a bit after you reach a certain output level for longer durations than the musical passages demand, I'd compare it McIntosh's meter hold function that rises then slowly falls back to resting, this must be the HCA bias. It will also provide 0.0002% THD at 10 Watts output Power into 8 Ohms, and under 0.005% THD at 130W into 8 Ohms at 1kHz. Most of my listening was in the 5 Watt range. It will actually provide a little over 190 Watts into 8 Ohms in practice with very low distortion. The MX-1000U sits at 115W receptacle draw when idling. What would become known and then forgotten as the HCA bias scheme design, was first conceived in the 1970's by Hiroyasu Yamaguchi and after a long wait the office awarded him the US patent US4803441 in 1983. it was a different method of biasing, and it was revised by Masao Noro in 1987, Japan patent JP62214707 and the US patent US7024028 on his behalf for Nippon Gakki, Yamaha. I posted that info elsewhere about 6 years ago. I'm not sure how the same few people could claim someone copied Nelson, because Pass Labs was not formed until 1991, so there is a 20 year gap there, a myth laid out as little more than supposition. The MX-10000 and MX-2000 are very unique sounding amplifiers, as is the MX-1000. There are actually several known versions of the MX-2000, the differences I'm aware of are in the shielded input board and capacitor choices used on each amp board. The 2000 I purchased a couple years ago vary somewhat from the schematics linked earlier in this thread, as do the components including Black Gate capacitors inside the amp compared to other owners I've spoken with. The PC5002M is also quite a bit different compared to the MX-10000, to put it lightly. In fact, the PC5002M is the pro version of the 101M, not the MX-10000.
As far as how exactly this type of dynamic bias works, I can't describe it because Spice has been unable to model it properly. All we got was distortion, yet the actual 10000 and 2000 amps make almost none.
Last edited:
Any news?
check out also this threads:
http://www.diyaudio.com/forums/chip...ed-topologies-yamaha-mx-10000-quad-405-a.html
http://www.diyaudio.com/forums/pass...s-ab-topology-about-hca-why-not-realized.html
Last edited:
Perhaps I am not following the whole thread. Threshold was granted a patent
on the Class A bias scheme in 1976. Are you referring to a more recent patent?
Interesting that you answered part of the post ~1/2 year later. Will search USPTO.Perhaps I am not following the whole thread. Threshold was granted a patent
on the Class A bias scheme in 1976. Are you referring to a more recent patent?
Threshold Corporation is the Assignee for a digital patent application. Four patents showed up for Threshold Corporation [Assignee] with Nelson S. Pass Inventor.
Last edited:
Member
Joined 2009
Paid Member
Hello Bigun. The patent referenced in SONY's is US 4,274,059; which was granted to JVC, and may have been named Super A in its amps. I could not find both until now; thanks to you. I continue to be interested in Dynamic Class A ever since those older days.
Best regards.
Member
Joined 2009
Paid Member
I haven't seen much on this forum about people using this output stage. If I'm not mistaken, Carlos (DX Destroyer) has used it and says it works. However, I'm feeling skeptical that it makes a real difference to the sound - it seems that Class AB optimally biassed within a global feedback loop sounds darn fine. Question is whether this Super A biassing provides enough benefit for use of the output stage open loop as a Class A + B output stage ? Is there an optimal bias ?
The middle schematic of the SONY shows its output stage operating stand-alone; perhaps like you suggested above.
Attachments
I don't understand this term: "switch your amp to mosfets"
Means this "replace POWER-BjT's by POWER-MOSFET's in the power output stages" ?
