5 ms·

I was under impression that class A amplifiers generally have the best amplifier characteristics (linearity, THD), and class D was mainly used in applications p

by fest 6y ago

I was under impression that class A amplifiers generally have the best amplifier characteristics (linearity, THD), and class D was mainly used in applications preferring efficiency/size/price over audio quality. Apparently class D has gotten lot better for the majority of applications.

asimpletune 6y ago
Yes, I believe the main thing that’s changed over the years is the switching speed. IIRC that’s due to geranium transistors?
- pjc50 6y ago
  Geranium is the plant, germanium is the transistor. But everything is silicon, and has improved for the usual Moore's law reasons. A class D amp is a closed-loop control system. While it can be done entirely in analogue, having a microprocessor in there lets the designer algorithmically compensate for nonlinearity of the output.
  - analog31 6y ago
    Most commercial switchmode amps are fully analog. The large power amps with built in DSP tend to process the audio signal before it reaches the power amp circuit. Artifacts such as harmonics can be driven down way below any credible threshold of audibility, with conventional techniques. Especially in big amps, the power amp already has some rather heavy responsibilities, just to be stable when driving odd loads, and generally not blowing up. This tends to discourage putting a lot of weird stuff inside the feedback loop.
jeffbee 6y ago
Class A architecture's main advantage is it is very easy to achieve high objective performance, while trading away heat and power (and consequently size and mass). You can easily get 3rd harmonic below -80dB and you can eliminate even harmonics entirely. All you have to do it spend a lot on heat sinks and commit to a large power bill. The problem is going to be reaching really stellar objective performance with the Class A design, because all those space age (not in a good way) power transistors are very slow and hard to drive, and that limits your error correction techniques. That's where these fancy designs are better. Basically they're using lightning-fast transistors and a DSP to bridge a power supply directly into a speaker. Objectively these designs have 3 orders of magnitude less total distortion, unmeasurable 2nd and 3rd harmonics, inaudible (-120dB or better) intermodulation products. The space age designs have simply no chance.
- willis936 6y ago
  By slow do you mean high capacitance between base and emitter? I’m under the impression that even high power BJTs will be good out to 100 kHz. FETs are even better I believe. Adding more components to the signal path degrades the signal slightly at best and introduces additional non-linearities at worst.
  - jeffbee 6y ago
    I think signal path degradation is thinking that can only be applied to linear amplifiers, and Class D amps are anything but linear. They literally short the power supply into the load with nanosecond timing. The transfer function through all that lossy, distorting stuff in the output path is pre-computed and equalized out by the DSP. If I had been more clear about what I was trying to say, I wouldn't have said slow and fast. What I was trying to emphasize was that the linear power amps are optimized around a linear ideal of a transistor that never existed. By contrast, the Class D amplifier exploits the characteristics of the transistors we actually make today: high dv/dt, high di/dt, Rds(on) close to zero. There are fifty companies out there trying to outdo each other on these stats and the closer they get to infinity/zero, the better the output of these new amps becomes. There are fewer people out there any more trying to make transistors more linear in the linear region.
    - amluto 6y ago
      You can always just use a big power op-amp. For some reason, audiophiles object to feedback, but any decent op-amp should give excellent audio performance. Power consumption and heat dissipation will be pretty bad, though. So you should go with class D.
      - hatsunearu 6y ago
        I tried designing a modest amplifier using op amps for the majority of the frontend. It's basically impossible unless you add a lot of stuff to buffer the opamp (which is why I wanted to use an op amp instead of discrete stuff)... big power op amps that have decent distortion exists, but not at their rated power. You get like 0.1% thd at typical levels needed to drive an output stage. And also thermal distortion will absolutely be an issue at that point.
hatsunearu 6y ago
It's actually class B that has the best performance. A lot of people say class B is bad because crossover distortion, but what they are actually referring to is class C (the classic picture where a bit of the waveform is conducted by the top one and a bit of the waveform is conducted by the bottom one, with some flat spots in between) Class AB is bad because near the crossover, both top and bottom devices conduct and that actually causes an assload of crossover distortion. Properly biased class B precisely hands over conduction from one side to the other, making crossover distortion practically negligible. Class A is theoretically the best but you can get good performance, and perhaps even better performance than A with a properly designed class B amplifier.
- hakfoo 6y ago
  I'd always heard the problem with class AB was less that both devices were on, but that when it's switching off one side near the zero point, that action added noise. There were a lot of branded tricks in the dying moments of "stereo as a status symbol" (early 1980s) which tried to mitigate this. (Technics New Class A, JVC Super-A, etc.)
  - hatsunearu 6y ago
    think of it this way: if you apply a sinusoid into a power storage, when the device (or the top/bottom device on its own) conducts 100% of the time, it is class A. When it conducts <50% of the time, it is class C. When it conducts >50% of the time (i.e. near the middle, both devices conduct), it is class AB. When it exactly conducts 50% of the time, it is class B. Turns out at the crossover point, when it's is class AB, you get something called gm doubling where the large signal gain goes up because both devices are conducting and therefore providing gain. When you're in class C (incorrectly called class B by noobs), no device conduct near the middle so you get crossover distortion in the opposite direction. In class B, you theoretically get near-perfect handoff which makes the crossover distortion minimal (at least compared to class A)
- SomeoneFromCA 6y ago
  This is the terminology Douglas Self uses. No one apart of him call class B class C.

by fest 6y ago

asimpletune 6y ago

Yes, I believe the main thing that’s changed over the years is the switching speed. IIRC that’s due to geranium transistors?

pjc50 6y ago
Geranium is the plant, germanium is the transistor. But everything is silicon, and has improved for the usual Moore's law reasons. A class D amp is a closed-loop control system. While it can be done entirely in analogue, having a microprocessor in there lets the designer algorithmically compensate for nonlinearity of the output.
- analog31 6y ago
  Most commercial switchmode amps are fully analog. The large power amps with built in DSP tend to process the audio signal before it reaches the power amp circuit. Artifacts such as harmonics can be driven down way below any credible threshold of audibility, with conventional techniques. Especially in big amps, the power amp already has some rather heavy responsibilities, just to be stable when driving odd loads, and generally not blowing up. This tends to discourage putting a lot of weird stuff inside the feedback loop.

jeffbee 6y ago

Class A architecture's main advantage is it is very easy to achieve high objective performance, while trading away heat and power (and consequently size and mass). You can easily get 3rd harmonic below -80dB and you can eliminate even harmonics entirely. All you have to do it spend a lot on heat sinks and commit to a large power bill. The problem is going to be reaching really stellar objective performance with the Class A design, because all those space age (not in a good way) power transistors are very slow and hard to drive, and that limits your error correction techniques. That's where these fancy designs are better. Basically they're using lightning-fast transistors and a DSP to bridge a power supply directly into a speaker. Objectively these designs have 3 orders of magnitude less total distortion, unmeasurable 2nd and 3rd harmonics, inaudible (-120dB or better) intermodulation products. The space age designs have simply no chance.

willis936 6y ago
By slow do you mean high capacitance between base and emitter? I’m under the impression that even high power BJTs will be good out to 100 kHz. FETs are even better I believe. Adding more components to the signal path degrades the signal slightly at best and introduces additional non-linearities at worst.
- jeffbee 6y ago
  I think signal path degradation is thinking that can only be applied to linear amplifiers, and Class D amps are anything but linear. They literally short the power supply into the load with nanosecond timing. The transfer function through all that lossy, distorting stuff in the output path is pre-computed and equalized out by the DSP. If I had been more clear about what I was trying to say, I wouldn't have said slow and fast. What I was trying to emphasize was that the linear power amps are optimized around a linear ideal of a transistor that never existed. By contrast, the Class D amplifier exploits the characteristics of the transistors we actually make today: high dv/dt, high di/dt, Rds(on) close to zero. There are fifty companies out there trying to outdo each other on these stats and the closer they get to infinity/zero, the better the output of these new amps becomes. There are fewer people out there any more trying to make transistors more linear in the linear region.
  - amluto 6y ago
    You can always just use a big power op-amp. For some reason, audiophiles object to feedback, but any decent op-amp should give excellent audio performance. Power consumption and heat dissipation will be pretty bad, though. So you should go with class D.
    - hatsunearu 6y ago
      I tried designing a modest amplifier using op amps for the majority of the frontend. It's basically impossible unless you add a lot of stuff to buffer the opamp (which is why I wanted to use an op amp instead of discrete stuff)... big power op amps that have decent distortion exists, but not at their rated power. You get like 0.1% thd at typical levels needed to drive an output stage. And also thermal distortion will absolutely be an issue at that point.

hatsunearu 6y ago

It's actually class B that has the best performance. A lot of people say class B is bad because crossover distortion, but what they are actually referring to is class C (the classic picture where a bit of the waveform is conducted by the top one and a bit of the waveform is conducted by the bottom one, with some flat spots in between) Class AB is bad because near the crossover, both top and bottom devices conduct and that actually causes an assload of crossover distortion. Properly biased class B precisely hands over conduction from one side to the other, making crossover distortion practically negligible. Class A is theoretically the best but you can get good performance, and perhaps even better performance than A with a properly designed class B amplifier.

hakfoo 6y ago
I'd always heard the problem with class AB was less that both devices were on, but that when it's switching off one side near the zero point, that action added noise. There were a lot of branded tricks in the dying moments of "stereo as a status symbol" (early 1980s) which tried to mitigate this. (Technics New Class A, JVC Super-A, etc.)
- hatsunearu 6y ago
  think of it this way: if you apply a sinusoid into a power storage, when the device (or the top/bottom device on its own) conducts 100% of the time, it is class A. When it conducts <50% of the time, it is class C. When it conducts >50% of the time (i.e. near the middle, both devices conduct), it is class AB. When it exactly conducts 50% of the time, it is class B. Turns out at the crossover point, when it's is class AB, you get something called gm doubling where the large signal gain goes up because both devices are conducting and therefore providing gain. When you're in class C (incorrectly called class B by noobs), no device conduct near the middle so you get crossover distortion in the opposite direction. In class B, you theoretically get near-perfect handoff which makes the crossover distortion minimal (at least compared to class A)
SomeoneFromCA 6y ago
This is the terminology Douglas Self uses. No one apart of him call class B class C.