One question that has popped up is why parallel the devices, someone hinted at the noise being like 1000 times, where lower current density in BJTs results in more noise, you can now remove the devices you want and study the effect. The OLG is higher than LM3886. If we all build this expect different results as the design is also exploiting the math we discussed in the feedback thread (run the sim several times and note the changing THD as the simulator also recognizes this some will get 0.000000% while others will get to upto 0.000005%). Feel free to modify and study effects.
As mentioned in our earlier discussion, for noise there should be one and only one main voltage amplification stage in a three stage amplifier composed of LTP,VAS and Output stage, either the LTP or the VAS. But thats just the beginning. The math shows that when breaking this law, meaning we have a small storm in the LTP then we need to stir up a category 5 in the VAS, this has the precedence of the success of Morpheus type architectures,blameless architectures over many years from stk series, lm3886 and many opamps.
For amplifiers such as Bryston and Musical Fidelity A1 it is the LTP that serves as the main voltage amplification stage. For SYMEF it is the VAS thats the main voltage amplification stage. Thats why these amplifiers seem noisier to owners of SYMEF, they call SYMEF the dark quiet amplifier, but everything has its cost
For amplifiers such as Bryston and Musical Fidelity A1 it is the LTP that serves as the main voltage amplification stage. For SYMEF it is the VAS thats the main voltage amplification stage. Thats why these amplifiers seem noisier to owners of SYMEF, they call SYMEF the dark quiet amplifier, but everything has its cost
Suddenly with above diagram the potential and future of class D becomes apparent, the last improvement frontier is in the integrator or error amplifier
You're crushing Q7 practically into saturation. Unless I'm mistaken, all the Vcb you have is set by the base current of Q7 and R28. How about adding a bit of resistances from the anode of D7 to the base of Q16? I'd ensure that Vcb(Q7) > 200 mV across temperature.
Tom
Tom
Post #22 gives a good idea on why some manufacturers transitioned from VFA to CFA apart from the usual marketing stuff, while all they had to do was degenerate the LTP or understand the feedback math and follow the rules such as Morpheus.
Last edited:
Two more remarks:
Why use high Vce transistors for Q1 and Q2? You could use low voltage transistors in this place that are easier to find with better parameters.
It may be a worthwhile consideration to connect D9 and C8 to the tail node instead of ground. This will keep the Vce of Q1 and Q2 more constant, thus lowering distortion a bit.
Why use high Vce transistors for Q1 and Q2? You could use low voltage transistors in this place that are easier to find with better parameters.
It may be a worthwhile consideration to connect D9 and C8 to the tail node instead of ground. This will keep the Vce of Q1 and Q2 more constant, thus lowering distortion a bit.
H
HAYK
I wonder what Minek thinks about this amp. When he designs and builds regularly extremely low distortions of 0.000005%.
Post in thread 'Unusual amp from 1987' https://www.diyaudio.com/community/threads/unusual-amp-from-1987.357369/post-7492293
Post in thread 'Unusual amp from 1987' https://www.diyaudio.com/community/threads/unusual-amp-from-1987.357369/post-7492293
Hayk, well what is finally implemented in that thread isn't quite exactly the same as the original idea
Will prepare PCB for this one, it will be as is not multiple compensation
Gerbers and Schematic here https://www.diyaudio.com/community/threads/morpheus-ultra-low-thd.406485/post-7532043 feedback welcome
Heres another link to the files https://www.pcbway.com/project/shareproject/Morpheus_Audio_Power_Amplifier_73d650c9.html
Post #22 is very important in having a global view of all audio.
Those familiar with some of Jeff Rowlands work, whether its the LM3886 or the other ultra low THD designs will realize that Orchestra is what Jeff and other audio engineers have been looking for all this time. In many of Jeff designs he aims for the single voltage amplification stage, the orchestra is the embodiment of all one voltage amplification stage approaches. The no gain or low gain LTP also allows one to comfortably use SMPS. What makes LM3886 special is that like the Morpheus its exploiting math to sound good, but oh boy does Morpheus go all out.
Those who own the right Jeff Rowland amplifier and a SYMEF get the gist of what am talking about. You can hear the silence
Those familiar with some of Jeff Rowlands work, whether its the LM3886 or the other ultra low THD designs will realize that Orchestra is what Jeff and other audio engineers have been looking for all this time. In many of Jeff designs he aims for the single voltage amplification stage, the orchestra is the embodiment of all one voltage amplification stage approaches. The no gain or low gain LTP also allows one to comfortably use SMPS. What makes LM3886 special is that like the Morpheus its exploiting math to sound good, but oh boy does Morpheus go all out.
Those who own the right Jeff Rowland amplifier and a SYMEF get the gist of what am talking about. You can hear the silence
Phase margins are one way of expressing stability, although some feel that a reactive load that can dip lower than rated may need more safety thus the quoted 45degrees to 60degrees safety margin, 20degrees can suffice in most cases, for the LM3886 its about 26degrees at 4ohm and 35degrees at 8ohm
The 2SD2560 with hfe 30000 ft 70Mhz, cob 120pf needs a little slow starting. Look out for the next video where Dr Stephen slow starts the output stage with some base resistors on the drivers