Here's the orchestra audio power amplifier.
https://www.pcbway.com/project/shareproject/Orchestra_audio_power_amplifier_20a20a77.html
https://www.pcbway.com/project/shareproject/Orchestra_audio_power_amplifier_20a20a77.html
Morpheus and Orchestra are essentially the same, yet very different depending on how you look at them
Did you deliberately use images to make content inaccessible and make it awkward to comment point-by-point? Text as text please, not as image.
The Blameless low R1/R2 values are to keep voltage noise low, no other reason. Any capacitor can handle low currents, its large currents that cause problems, or non-linearity (voltage coefficient).
The Blameless low R1/R2 values are to keep voltage noise low, no other reason. Any capacitor can handle low currents, its large currents that cause problems, or non-linearity (voltage coefficient).
Hi Mark the images are not for making content inaccessible unless translation software is in use. Thanks for pointing out the issue though. Will accompany images with text.
We did have a lengthy capacitor discussion on another thread. Lets keep the second point open to further analysis
We did have a lengthy capacitor discussion on another thread. Lets keep the second point open to further analysis
Looking at feedback math we can see that Morpheus and Orchestra easily take on amplifiers like the Denon PMA 1500AE series and are at a greater advantage
Both the Morpheus https://www.diyaudio.com/community/threads/morpheus-ultra-low-thd.406485/ and the SYMEF v 2024 https://www.diyaudio.com/community/threads/symef-amplifier.198500/#post-2745105 also sport 0.000x% THD
I am sure that you could have made the circuit even more complex than this. You should consider that less is more, have a look at Nelson's designs. Simulating things are also not the ultimate since the simulator cannot change and step everything "randomly" The more complex the design, obviously the layout besides more complex the worst lumped effects will be had. Anyway good luck into punting this to the community, especially regarding 0.000 x THD, that is not the way calculating compounded noise or distortion.
Yes , less IS more.
There are two forks in the road -
- very linear symmetrical designs that need very little OLG.
-less linear "blameless" type designs where you fight stability to get "gobs and gobs" of gain to correct the imperfect circuit.
Lately , I want for my amps to be the former. Single simple miller can also get 5 PPM. Self said to linearize as much as possible
and apply feedback to correct what is left over.
My best symmetrical design can do .0005% with its hawksford-CFA and one cap. Its "core" is VERY simple- 8 devices only need 10db NFB @
20Khz . "Blameless" Wolverine" has multiple compensations ,current mirror "helpers" and cascodes everywhere ! 45db @ 20K only
gets a few PPM improvement.
The "law of returns" definitely applies to amp design.
Edit - Curl "Parasound" and Pass amps are just about all symmetrical designs with low NFB.
OS
Less is usually less, and in amplifier linearity its demonstrably the case. Every single-transistor amplifier is awful, for instance.
So I completely call-out this bogus assertion
If only there were a perfect simple circuit, but real active devices are not linear at all - simple circuits using active devices can get to about -80dB distortion and that's it (and that's using heavy local feedback such as CFP). Feedback uses gain to confer the linearity of passive components onto the whole circuit, demonstrably more is more and the magic happens (ppm linearity from a circuit using lowly BJTs and FETs). Unconditional stability is readily achievable. If you actually read about the design of the Blameless amp you'd see that every part of it is optimized for linearity locally as well as adding global feedback around it. Its always the output stage that is the biggest source of non-linearity (and not just cross-over), for which there is no other practical way to linearize than a feedback loop or loops around it.
So I completely call-out this bogus assertion
Enlighten me. Then why is a non-beta enhanced simple "blameless" barely able to do .05% 20K ?
P3A is one of these amps @ 10db/20K.
Same 10db on a leach type amp = .001%.
I suppose I am just bored with the endless reiterations of the "blameless" constantly rehashed on this forum.
What are they searching for ? Mo' PPM ?
OS
P3A is one of these amps @ 10db/20K.
Same 10db on a leach type amp = .001%.
yes , one of the most common topics on DIYA is "why does my P3A oscillate"
I suppose I am just bored with the endless reiterations of the "blameless" constantly rehashed on this forum.
What are they searching for ? Mo' PPM ?
OS
My two cents:
The class A stages in an amplifier can be designed to have arbitrarily low distortion by choosing the amount of deviation around the operating point. These stages have large emitter resistors and drive resistive loads.
Unfortunately, the final stage must operate in class AB for efficiency. A class AB emitter follower has 0.05-0.5% THD. The lower number applies to many devices run in parallel. That quickly becomes expensive.
A well-designed amplifier's open-loop THD is dominated by the output stage. Closed-loop, the output stage's distortion is divided by the amount of global feedback.
For example, my complementary amplifier has 0.25% THD (open-loop) / 15 (24dB feedback) = 0.017% THD closed-loop into 4 ohms. My amplifier's feedback remains constant up to 20KHz.
Distortion will vary widely with amplitude (worst-case is a few volts peak), bias current (an optimal bias at idle may be 2-4x higher under load), load impedance (8->4 ohms roughly doubles the distortion), and for most amplifiers, frequency.
The THD numbers quoted on this board are usually optimistic because they are not measured at the worst-case amplitude, far-from-optimal bias, a low-impedance load, 20KHz, and mismatched devices.
Ed
The class A stages in an amplifier can be designed to have arbitrarily low distortion by choosing the amount of deviation around the operating point. These stages have large emitter resistors and drive resistive loads.
Unfortunately, the final stage must operate in class AB for efficiency. A class AB emitter follower has 0.05-0.5% THD. The lower number applies to many devices run in parallel. That quickly becomes expensive.
A well-designed amplifier's open-loop THD is dominated by the output stage. Closed-loop, the output stage's distortion is divided by the amount of global feedback.
For example, my complementary amplifier has 0.25% THD (open-loop) / 15 (24dB feedback) = 0.017% THD closed-loop into 4 ohms. My amplifier's feedback remains constant up to 20KHz.
Distortion will vary widely with amplitude (worst-case is a few volts peak), bias current (an optimal bias at idle may be 2-4x higher under load), load impedance (8->4 ohms roughly doubles the distortion), and for most amplifiers, frequency.
The THD numbers quoted on this board are usually optimistic because they are not measured at the worst-case amplitude, far-from-optimal bias, a low-impedance load, 20KHz, and mismatched devices.
Ed
What is non-beta enhanced exactly?
Standard Blameless measure about 0.005% at 20kHz into 8R, read the book!
I don't know the P3A, but from what I can glean from a few searches its not a Blameless design at all, and the original VAS transistor is a BD140 which has no capacitance spec(!)
Last edited:
For most people, so long as it works acceptably few care about the technology or design in use, whether simple or complex, in reality some chase resolution hoping that the higher the resolution the more likely nothing is wrong, but even THD is just a snap shot of a time and state, whatever happens beyond that window is what determines which amplifier will be switched on next time, luckily most people have poor resolution