Class D is just Dandy!


I thought it was time we had a pro- Class D thread. There's plenty of threads about comparisons, or detractors of Class D.

That's fine, you don't have to like Class D amps, and if you don't please go participate on one of those threads.

For those of us who are very happy and excited about having musical, capable amps that we can afford to keep on 24/7 and don't require large spaces to put them in, this thread is for you.

Please share your experiences with class D amps!
erik_squires

Showing 50 responses by erik_squires

Personally, I think it's a horrible thing to price out a product based on component costs.

I think everyone would be horrified if we did this consistently across audio.

There's more to a high frequency design than the cost of the transistors. PCB layout must be done meticulously, and test failures common, so R&D matters.

Still, having said that, Class-D amplifiers are quite good already. Better than some Class A amps, worse than some Class A/B.

If there are actually audible benefits to GaN based amplifiers I look forward to hearing them.
Whether a designer takes advantage of this fact is another matter but generally speaking GaNFET amps should sound better as they have less distortion on that account.

@atmasphere

As you are well aware, I’m kind of touching on an issue of specmanship that that affects us with all amps. Can we hear "better" when you go from 0.01% distortion to 0.001%? I know distortion can be reduced, which technically means more fidelity to the input signal. What I don’t know is that this alone is audibility, let alone desirable.

Like Tube amps, or Pass (who makes it clear this is his goal) the distortion profile may have more to do with euphonic capability than merely lower distortion.
but replace my simple little single ended tube amp?


First, @wolf_garcia , I applaud you for being willing to expose your audio quirks without fear. I wish I had that kind of personal courage. ;-)

But that’s just it. If you love the LP sound, only the LP sound will be good enough. If you have a particular niche of amp, no other amp will sound as good. It’s not proof Class D is behind the best solid state amps. If you have a love for 30 watt Class A, same thing. It's a bit like comparing all restaurants to mom or dad's home cooking.  Kind of unfair to rate them that way, unless of course your mom (or dad) was an absolutely terrible cook.

As I always say, buy and enjoy what you like, even if it’s’ some quirky SET or full-range with a giganormous bass horn. That’s fine.

But I think it’s unfair to diss Class D because it doesn’t sound like a SET. Among large solid state amps, I don’t know anyone who can listen and say "Oh, that’s a Class D amp, I can tell because....."

Best,

E
Really, after having had this same exact discussion for a year or two we need to rehash it a millionth time?  


Yes..... but this time it's a lot quieter somehow.
Hy Roxy,

I never did. My experience in the pro world was in motion picture auditoriums. At the time my employer bought Hafler amplifier modules, attached their own supplies, and their own surround processing. We probably kept Hafler in business a lot longer than they would have been otherwise. I understand they have come back though.

My day job is esoteric related to IT.

My current set-up uses custom loudspeakers driven by ICEPower 250ASP modules. I did a lot of listening and comparing them to Parasound Halo A23 amplifiers, which while small, are fabulous performers. I rate the two as equivalent in sound quality.

Best,

E
I should say that in the mid-80’s I did get to listen to early Class D amplifier from one of our vendors. It was not then up to snuff, especially not with metal dome compression drivers. :)

Best,

E
Greg,

Good to hear. TBH I've only heard B&W speakers with Krell amps, and that combination never made me reach for my wallet.

Interesting you find the ICEPower modules working so well for you. :)

Best,


E
Hi @roxy54

You know, I probably heard the same Class A amps @whitestix heard in the Bay Area and I have to concur. I re-iterate, blindfolded and level matched I don’t think I could tell them apart from the Parasound A23’s. Both are excellent, modern sounding amps via my reference speakers.

Negatives?
The ICEPower 250 ASP modules are in no way euphonic. They lack the utter calmness and dark backgrounds of the very best Ayre or Arcam systems. They lack the juiciness if not excessive boldness of the best big Conrad Johnson tube amps.

Despite all of this, they are very very engaging and when the music and sources are right just as compelling to listen to as the best amplifiers I have heard.

Unlike the worst SS amps, they also don’t sound thin, compressed, underpowered or grainy at all. I just listened to Trombone Shorty’s latest tracks via Jazz FM 91 and man, freaking awesome playback. Absolutely nothing was missing nor called attention to itself in terms of the amplifier.

Best,


E
Hi randy-11,

If you mean the multi-channel amps with Hypex/nCore modules, YES THEY DO! 

I would just be sure to listen first. I have heard a couple of modern NAD gear and I didn't like it. This is a brand I really encourage you to listen / try out before committing.

Best,


E
Guys,

You know I was just thinking about something. If the current crop of digital amps had appeared in the 1980's I think they would have pretty much ended the solid state / linear amplifier market.

The SS amps at the time were, by and large, full of compromises, and consumers were hungry for all things digital.

This did not happen however, the digital amps of the 1990's weren't nearly as good as they are now, and solid state amplifiers also continued to advance to where they are today.

Still fun to think about. :)

Best,


E
Interesting info on the Technics SE-R1 amplifier from @georgehifi

One thing I thought was particularly interesting (and a lot of work) was the Load Adaptive Phase Calibration.

All digital amps must go through a low pass filter at the output. This may cause some variability in the high end response based on many things. Modern digital amps with easy to drive speakers should not suffer this very much (especially if they use feedback at the output, shhhhh!)

The SE-R1 takes on the approach of measuring and directly compensating for this at the input instead of using feedback.

This is a LOT of number crunching to put into an amp. Cool ideas for sure.
Interesting, @georgehifi re garding transistor manufacturing.

In 1976 Dr. Leach introduced his paper, Build a low TIM amplifier which transformed, if not codified, what would be linear amplifier design for the next 3 decades.

Also 80's were flirting with MOSFETs that promised to behave more like tubes. High input impedance, more linearity, blah blah blah. and some worked out, some worked out some of the time. :) I think James Bonjiorno (spelling?) and Sumo in general were among the early names that were associated with this. But we digress.

Well, there is always some technology around the corner, some new way of winding a transformer that promises never-before heard of improvements. 99% of the time, these changes don't pass the test of time, and are forgotten by their own inventors, or fail to make enough of an impression to be adopted by others. There are also a number of amplifiers who end up being door stops once the manufacturer of the super-special magic transistor stops making them.  << sigh >>

Remember Halcro with their super low distortion, universal accolades? It went under, came back... and now who knows.

I guess my point is, we can always look forward to new things, but we can't let that stop us from enjoying what we can get now. :)

Best,

E

@georgehifi

Very interesting! :)

Hexfets..... I seem to remember ... was it Toshiba that invested heavily in them for a while?

Best,

E
@georgehifi

Your link just reminded me of something. One thing I really like the Japanese manufacturers for is power meters. Yamaha (professional), Hitachi, Technics, early Onkyo. Gorgeous analog meters in the 1980’s. I’m very happy to see some of this coming back, even if it is in gear I can’t afford. :)
I really really like the idea of putting a tube pre in front of these amps, especially with the high-input impedance varieties I've read about.

I often suggested tube pre/ss amp before, so this idea appeals to me a great deal.

If I wasn't 50/50 HT and music I would consider it more seriously for myself.

Best,

E
Todd,

Remind me please your preamp?

I was a big fan of the old-school CJ preamps. I had thought about getting a modern Cary instead, but man, the press they get about service quality.

I've not listened to CJ pre's since they came out with the ART line.

Best,

E
You know, one digital-ish amp technology we haven't talked about is Yamaha's EEEngine. It is a hybrid approach, which reminds me of Carver's Magnetic Field and NAD's voltage switching technology.

It uses a digital amplifier to provide the + and - voltage rails, but a linear amplifier at the output.

http://www.yamahaproaudio.com/global/en/training_support/selftraining/technology/eeengine.jsp

I've never knowingly heard them, but if anyone is using one let me know. :)

Best,

E
Todd,

Have you heard the Mytek DAC/preamp combo? They're pretty good if you have liminted number of sources. :)
@chrshanl37

I think that's an excellent idea. :) Some modules provide supplementary 12V power, so with a little tinkering you may indeed be able to revive them without very much work! :)

You may also want to think about replacing any lights with modern LED equivalents. May help with the power supply situation.

Best,


E
It's really interesting to me what great participation one gets when you start the thread off right. :) Thanks to everyone for chiming in.

Best,

E
Great info @noble100

I don't usually agree with the theme that "Class D is about to get as good as Class A" as I think it has been there for a while.

Linear amplifier distortion can be reduced by feedback, something digital amps may use in buckets.

Where I think the new Technics amps using this transistor will be most interesting is in amp/speaker matching, as the average Class D module is not the current delivering brute that say a giant Boulder or Krell is across the frequency spectrum. Technics has an interesting way of handling it. I handle it by making easy to drive speakers. :)


Best,

E
Todd,

If you like tubes, stay with tubes. :) No matter how good the Mytek is, it won’t be a tube pre. :)

Best,

E
Welcome @whitestix  !

I am glad you are finding it informative, I'm learning a lot more about the marketplace, and happy to find so many sharing my appreciation for the latest amps!

Best,

E
It may not be, but it works totally differently.

In a digital amp the feedback affects the width of the "on" period.

First, I’m not sure that anyone has proven in a linear amplifier why moderate feedback is a bad thing. Second, in class D it is a couple of stages away from the output signal, so whatever arguments you might have made for a linear amp’s feedback I don’t know that they would work at all.

As I understand it, Technics is using more of a feed-forward loop. They analyze the errors with the output stage, and then digitally correct it ahead of time, before the actual power amplifier stage. Talk about a huge amount of processing and signal meddling though. :)

The Technics approach is VERY similar to digital EQ and phase correction if you were to limit the correction to purely electrical, it is identical in theory.

If you go that far, might as well include the acoustic output of the speakers as well and finish the job. :)



Best,

E
Hi George,

Feed forward in a linear amplifier is very different. :)

In this case, I’m using it more as an analogy or principle rather than a specific circuit design.

While most amplifiers have feedback at the output stage, the Technics does the correction mathematically at the input, pushing the error compensation from the front instead of the back.

For this reason, I think of this as a feed-forward idea, being the opposite of feed-backwards.

Best,

E
Hifidream :

Interesting, very similar to the Yamaha EEEngine I wrote about a little while ago, and therefore similar to the Carver amps too.

Best,

E
I wish we weren’t ragging on Pass.... but yeah, my own experience (but in 2 different systems) was that ICEPower ASP was their equal.

Which also, by extension and if I have any integrity, means the Parasound Halo amps I've heard are also very very close.

Best,

E
Also, Fremer’s position about switching amps all having these problems is utter nonsense. Even though the review is five years old, by 2012 digital amps were getting very good. This was unsubstantiated bias which is part of the reason why Stereophile subjective opinions don’t matter at all to me anymore.

http://www.stereophile.com/content/mark-levinson-no53-reference-monoblock-power-amplifier#gJgWo1W0Vq...

The ASP range which was released in 2003 is what I listen to today and I'm perfectly happy to say Fremer was wrong then and wrong now.

https://en.wikipedia.org/wiki/ICEpower

Best,

E
@georgehifi 

Guess I didn't pay enough for them to find the problems Fremer had.

Best,

E
Always fun to talk tech.... but I've been living with dreamy sounding amps for a while now. I don't need my switching frequency increased to have the best.

E
I'm afraid I haven't had time to go out and listen to a lot since the last show here in ... Long Beach? Newport Beach?

I've also gotten to a poin where I'm very happy with the sound quality at home but not the remotes. :) Seems bad for me to find a dealer to listen to gear I'm not going to buy.

Maybe I should ask manufacturers to send me gear to review instead??
At $17k a pair? No. No thank you.

I’ve heard current linear amps that were about that expensive and I wouldn’t switch to them either. To me the idea that we need super technology to finally fix the Class D problem just wont’ fly.

Blame my tin ears if you must, but this was solved a decade ago.

Best,

E
Also, the meters on the Technics SE-R1 are super cool. Wish I could buy the casework alone.
Hi @guidocorona

The general principles that a higher switching frequency with better transistors yields lower distortion and higher efficiency is not really in question.

I question the audibility of anything measurably better than the current state of the art from ICEPower, nCore or Pascal. I question the importance of 0.03% distortion vs. 0.003% vs. 0.00000001%.  Having a baking scale that measures in femto-grams does not help you bake a better cookie.

If there are audible benefits, I think the answer will be somewhere else. Such as linearity (lack of compression), noise shape or handling difficult to drive speakers (complex impedance curves), etc.

I won’t get excited at all right now over an amp with a high switching frequency, or lower distortion. Especially not at high end prices.

There may be audible differences between the major Class-D technologies, but harping on things I think were solved a decade ago I don’t think will help me find a "better" sounding amp. Maybe a "different" sounding amp though.

Best,

E


I'm just saying that I am open to the possibility that different Class D amps, like their linear counterparts, may sound different on different speakers.

But if this IS true, I won't attribute it to Class D vs. linear.

I believe nCore's do have exceptionally low output impedance across the frequency spectrum compared to ICEPower so I'm not surprised. :)

It also depends where your speaker is difficult to drive. With my speakers, there's unlikely to be much difference since they are easy to drive. ESL's however have the hardest trouble at the top end. I would not be surprised if the difference between an nCore and ICEpower module was more measurable/audible with them.

Other speakers, like say Focal's with double woofers, are hardest to drive in the bass, a place where both nCore and ICEPower excel.

Best,

E
So, let’s get into some meaningful spec-manship. Let’s compare the frequency response, and output impedance (8 / output impedance = damping factor) of ICEPower and nCore to see about what could matter with different speakers.

Here are the source docs I start with:

Ncore docs:

https://www.diyclassd.com/documenten/download/860

ICEPower :

http://www.icepower.bang-olufsen.com/files/solutions/icepower250aspdata.pdf

I’ll use a couple of examples. My own custom built speakers which have around 9 Ohms of impedance at the top octave, and ESL’s which are basically giant capacitors, and murderously low impedance at the top octave often near 0.3 Ohms.  I should point out most SS amps have coils at the output to protect from hyper-sonic oscillation. This coil acts as a low pass filter and will cause the frequency response to droop at the top octave. This is not by itself unusual or ONLY about Class D amplifiers.

Using a "normal" speaker like mine let's examine how the output impedance changes response at the 20kHz mark:

nCore: -0.003 dB
ICEPower: - 0.3 dB

0.3 dB isn’t inaudible.... but VERY subtle. Below this, in the bass where many speakers get difficult to drive, there’s only tiny differences in response. Certainly better than most tube and zero feedback SS amps. This is a small difference which fits well within the idea of "system matching" an amp to personal preferences, speakers and room tuning.

But let’s take an extreme case of an ESL which has around 0.3 Ohms in the top octave:

nCore: -0.08 dB
ICEPower: -6 dB

This is a case where the nCore could make a very significant, audible difference when playing the same speakers.

My point to this is, based on specs, both modules will perform very well with normal loads. Any differences in the top octave may be inaudible, or useful.

This is one example of where I think specs and technology can help us understand how to match an amplifier with our system better.

Best,


E
Following on, at 200 Hz the output impedance is 10x better with nCore, but the absolute figures are tiny. The difference in output between a high and low impedance speaker is less than 0.01 dB.

Most "normal" speakers get very difficult to drive in the bass, not the treble, so for these owners, I think nCore and ICEPower are equivalent and subject to personal preference.

If you have an ESL though, I would really expect the nCore to over perform.

Best,


E


Hi Tim,

I just tested it. It actually is a PDF download, so check your download files. :)

Best,

E
Hi @guidocorona

I wish I had a way to listen to an nCore pair without spending money. :) I would be better able to discuss the sound quality honestly.

What I can repeat, is that compared to Parasound Halo A23's which are heavily class A biased, I cannot hear a difference.

Best,

E
Hi @guidocorona

Thanks for the suggestion. I feel bad asking for a loaner pair with no intention of buying. :)

To everyone else, it is really the contributors on this thread, even those i don't agree with, who keep it interesting and welcoming. :)

Best,

E
One thing I'd like to see for the nCore modules which I can't find is output phase angle. If anyone can point me to a doc which has this I would appreciate it.

I notice that the ICEPower units do have some phase shift involved. It is minor compared to what your average multi-driver speaker and crossover will do, but it's significant. This is something the Technics SE-A1 compensates for in the input.
Hi Kdude,

That output coil is perfectly in line with the output coil in many SS amps.

Based on my simulations, this ends up being no more than about 10 degrees of phase shift at the top end. Far better than ICEPower. :)

Best,

E


A little confirmation bias there, George.

The phase shift in the nCore, if ONLY due to the output coil is minor, about 10 degrees, and almost none from 10kHz or below. We don’t actually know the output phase shift without a direct measurement. We can only estimate what a 10 uH coil would do.

The size of the output coil is in the same order of magnitude of most SS amplifier output coils. Maybe 2-3 times bigger.

Also please note the effective nCore 400 output impedance is not 1.25 Ohms at 20 kHz but around 0.003 Ohms. Probably due to feedback being outside the coil.

These are minor quibbles. In fact, the interesting part about the Technics amplifier is that it solves this not by a better output section but by measuring the frequency and phase differences with the actual speaker load and then compensating for it at the input. This is pure frequency and phase equalization. You could do this with any current Class D amp plus a miniDSP if you chose to.

The ICEpower amps on the other hand have much bigger phase shifts. No idea what it’s attributed to.

Of course, time marches on, and progress will be made but what we have now is really very very good.

Best,


E
@kdude66

No worries at all. :) I think that the participation of so many here who have been pleasantly surprised and are enjoying Class D amps speaks for itself.

We also went a little deep into theory without measurement. :-) We should stay away from that. My bad.

Best,

E
Hi Phusis,

Not sure why high efficiency would be a bad combo. The issue with different Class D designs, as covered, has more to do (in my mind) with impedance drops of exotics.

They are dead quiet.

When I first heard digital amplifiers it was with a pro monitors based on pioneer metal-dome compression drivers.  OMG that was painful. :)

But that was in the 80's.

Class D amps are also dead quiet, so that's not a big deal.

Still, I don't have recent experience, so I hope some one who does will chime in.

Best,

E
@h2oaudio

VERY interesting. :) Please elaborate on the amps you are using, and how the speakers are configured (bi-amped? etc.)

Thank you!


Erik
My own take on high efficiency speakers and flea tube amps:

People LIKE the sloppiness.

That is, the high impedance outputs of the amps make the electrical frequency response much more variable, and .... that's what some people like. :)

I don't think it's from "damping" or over-controlling the drivers at all. But hey, no matter. Buy what you like.

Best,

E