Class A or Class D solid-state amplifiers (modern designs)

What have you been able to glean about the long term reliability of class D amps, what goes wrong, and what is the result? 

Electrolytic caps last longer if they have a charge on them and they are not subjected to heat. WRT the other parts used, surface mount has been around for several decades and reliability seems to be at or better then the level of reliability of thru-hole parts. I expect class D amps will hold up very well.

@petaluman As you might expect I have read Nelson's article before.

He addresses the issue of poles in the amplifier caused by adding more stages of gain and how that adversely affects the use of feedback. I addressed that issue in my first post to this thread.

A class D amp only has one stage of gain FWIW. Nelson does not address the solutions class D offers to the issues described in his article.

That analogy doesn't hold water, if you see what I did there.

I havenot heard that many that are point-to-point wired amplifiers using the best parts quality with customer made transformers and chokes, Audio Note non-magnetic resistors, V-Caps, copper plates, etc.  I have heard most of the typical tube amps most of you are familiar with but not many Class A tube amplifiers. 

@bigkidz 

😉 We've been building point to point wired class A triode amps for 48 years. Custom built chokes and transformers, V-Caps, custom copper wire made to our specs for hookup; we were the first to use Caddock resistors in the audio industry (1978) and garnered many awards and reviews in the high end press over the decades. Totally FWIW department...

Objectivism ruled and high negative feedback gave vanishingly low THD and IM numbers.  However, the resulting sound quality got worse, not better.  TIM distortion was discovered, feedback was dialed back, and the importance of listening was once again proved.

This statement isn't entirely correct! TIM was simply caused by part of the amplifier being outside the feedback loop. With regards to 'high feedback' there were none in the 1970s that had enough Gain Bandwidth Product to support that feedback at high frequencies- so at higher frequencies the feedback fell off and distortion increased- resulting in the 'harsh and bright' solid state sound that has plagued solid state amps since their inception.

his statement was to me that he has heard amps that compete - so since I cannot get a hold of his amps - I am asking him to provide me with his experience based upon his response to me.

To date, we have not heard a class D amplifier that competes in sound quality.  We are open to hear something that comes close but not yet to our ears.

"I have. FWIW class D amps vary at least as much in sound as tube amps do."

The amp that convinced me that we needed to being working with class D was a Cherry amp. I've not heard the Orchard but it seems to get very nice comments. A couple of customers have the AGD amplifier and are very happy with them.

The bottom line is you can't say if you've heard one you've heard them all! I've heard some that I had to really ask if the manufacturer was serious (that was a long time ago).

To date, we have not heard a class D amplifier that competes in sound quality.  We are open to hear something that comes close but not yet to our ears.

I have. FWIW class D amps vary at least as much in sound as tube amps do.

Let’s compare and contrast. Which one is technically better?

Here's what you're up against: distortion. Sure, class A does not have crossover distortion. Neither does class D.  The main thing you're dealing with is IMD and harmonic distortion.

Class A amps make both. They are class A to try to minimize that by being in the most linear portion of the operating curve of the output device (whether tube or transistor). But they will make distortion and unless the circuit is designed to be zero feedback there will have to be loop negative feedback to get the distortion into acceptable bounds.

That's where you get into trouble. Back in the 1970s and 80s, the semiconductors needed to actually be able to run the kind of feedback you need didn't exist. We had to wait until the 1990s for that.

If you're thinking that an entire generation of audiophiles grew up with amps that sound bright and harsh you are correct.

The issue is something called Gain Bandwidth Product. You need a lot lot of it! GBP supports the operation of feedback, and to have it support feedback at 10KHz you need more than was possible before sometime in the 1990s. What happens if there isn't enough GBP is the feedback value decreases at higher frequencies- resulting in higher distortion.

This is why the 1KHz harmonic distortion figure tells you little about the amp. What is more important is distortion vs frequency- it should not rise across the audio band, otherwise the amp will have higher distortion at higher frequencies, and this will translate directly into 'harsh and bright' since our ears convert distortion into tonality.

In addition to the semiconductors needed you also need the design. And the will to create it. That took until the 2000s...

As a result, us older audiophiles have been hearing 'harsh and bright' in solid state designs for decades on at this point. That is literally the reason why there are still tube amps around!

If you've been digging deep, you may have realized this isn't a class A vs class AB thing- both classes of operation have can have this issue.

Class D offers a way around this problem.

In a regular amp, if you try to put a lot of feedback in the design you run a risk of the amp being unstable and prone to oscillation unless the feedback loop is very carefully designed, and maybe even then. The reason is there are frequency poles in every amplifier design and they cause phase shift at high frequencies (often outside the audio band). At some frequency the phase shift is so severe that the feedback becomes positive rather than negative- and so the amp oscillates.

In technical/engineering terms this condition is described as 'the phase margin of the amplifier being exceeded'.

In a class D amp you can put a lot of feedback in the design, with the expectation that it will do exactly this. The oscillation is then used as the switching frequency, killing two birds with one stone.

At this point you can have enough GBP that the distortion remains constant at all frequencies, and the distortion normally caused by the application of feedback can also be cleaned up by the simple fact of so much feedback available. You really need to have 35dB or more of feedback to really allow feedback to work right- less than that and it contributes distortion of its own.

This is very difficult to do in a class A or AB amp, although there are a few examples.

That is why some class D designs can easily keep up with class A designs (and FWIW, I've been working with class A amplifier designs for nearly 50 years, if that means anything...).