Advance in Class D amps?

[sir sanders zingmore]

Some big claims here

Hypex presents a decisive leap forward in class D audio performance. Ncore® is the first class D amplifier not just to nudge the best linear amplifiers, but to surpass them in every aspect relevant to sound quality

There are links to some technical stuff that I don't understand. Perhaps someone can explain it to me....

[DoggieHowser]

What about the new flavor of the month: Devialet?

Supposed to be a combination of Class A and D.

Costs 11k Euros, but it does work as a DAC, phono stage and integrated all in one. All the rave at the recent show in Germany.

[Upgrayed]

Have no doubt that class D is the future(and the present for some of us), there will be many advances and different implementations on the way.

Hopefully the prices will offer access to the masses

[Zaphod Beeblebrox]

There's no real mystery with Class D amplifiers. Switching frequency. That's it. THAT is, by a very long margin, the separator between Class D amps (good and bad) and high quality linear amps. Particularly when using ESLs. When the switching frequencies of Class D amps approach 5MHz, then they will be ready to tackle the big boys. At 10MHz, it will all be over. Linear amps will not be able to match Class D for any practical system.

AFAIK, the very best Class D amps manage around 1MHz switching frequency. There is a way to go.

[Keith_W]

Doggie Howser I must admit I am having some difficulty comprehending how it is possible to combine Class A and Class D. The way I understand it, Class D does not depend on a bias current at all - rather, it pulse width modulates a current at very high frequency. Would you happen to know how it works?

[krebetman]

Sounds similar, to my simplistic level of understanding, to the approach used in the Diavelet amp?

[DoggieHowser]

Doggie Howser I must admit I am having some difficulty comprehending how it is possible to combine Class A and Class D. The way I understand it, Class D does not depend on a bias current at all - rather, it pulse width modulates a current at very high frequency. Would you happen to know how it works?

No clue. Just reading the literature and reviews like everyone else

But I recall Quad's current dumping architecture on the 606 mentioned using two amps as well. A lower quality one which supplies juice. Loads of it. And a second one which gives it the finesse.

[Orpheus]

No clue. Just reading the literature and reviews like everyone else

But I recall Quad's current dumping architecture on the 606 mentioned using two amps as well. A lower quality one which supplies juice. Loads of it. And a second one which gives it the finesse.

The first "current dumper" was the 405. A less kind way of putting how it works was that the first one makes a mess, and the second one tries to tidy up after it.

The 405 is an incredibly tough and resilient amp (I have one driving our TV sound system, into Audio Physics Virgo III, and it is perfectly competent, but it is a bit rough. The 909, apparently, is a significant improvement.

[Grimmie]

Have no doubt that class D is the future(and the present for some of us), there will be many advances and different implementations on the way.

Hopefully the prices will offer access to the masses

It seems very likely to me that when they have the R & D completed and can produce the goods as Zaph is suggesting they must, then the cost is bound to reflect the difficulty in arriving to that state and the cost of implementing it. That almost certainly means that Hi end class D will be high priced as well.

Grimmie

[krebetman]

It seems very likely to me that when they have the R & D completed and can produce the goods as Zaph is suggesting they must, then the cost is bound to reflect the difficulty in arriving to that state and the cost of implementing it. That almost certainly means that Hi end class D will be high priced as well.

Grimmie

Maybe at first, but the trickle-down effect would surely be very rapid.

[Nigel]

I suspect if it was easy, they'd be doing it by now.

As we have heard in other threads (ie anecdotally) Class D seems fine for bass reproduction so at some freq it becomes discernibly worse than some other classes. I have no idea how one could determine that point, but it may be useful info for the modern trend of multiple amplifier/driver systems. And of course as the switching rate increases, perhaps that freq point would rise ..?? maybe ???

ZB, do you think one day Class D will be the technology of choice ?

[eras]

I don't quite follow the problem with switching frequency. Having a look at the Hypex charts, and also at the Nuforce V3 technology, they both seem to be very linear and capable of high fidelity reproduction. Where do they lose out? The Nuforce is now employing an improved capacitor board which should help to smooth things out as well, surely?

[Zaphod Beeblebrox]

I suspect if it was easy, they'd be doing it by now.

As we have heard in other threads (ie anecdotally) Class D seems fine for bass reproduction so at some freq it becomes discernibly worse than some other classes. I have no idea how one could determine that point, but it may be useful info for the modern trend of multiple amplifier/driver systems. And of course as the switching rate increases, perhaps that freq point would rise ..?? maybe ???

ZB, do you think one day Class D will be the technology of choice ?

We are already seeing Class D as the amplification of choice for most mid-fi products. Initially, it was chosen for 'high end' products, mostly due to the high cost involved (normal with new technologies). Now the dust has settled somewhat, Class D is occupying the middle ground. Prices will/have been driven lower, such that Class D will occupy the low end almost completely over the next few years.

IF Class D switching frequencies can be pushed to the 5MHz area, then it will certainly provide very serious competition for the very best linear amplifiers. Certainly, within the next few years, there will be far fewer linear amplifiers available in all price levels.

Your question is deceptively simple. The answer is not. Regardless of how good Class D becomes, there is likely to remain a hard core band of listeners that will never accept the technology. Likewise, if switching frequencies do not reach high enough, then there is likely to remain a hard core band of listeners who will not accept Class D because it is simply not good enough.

[Zaphod Beeblebrox]

I don't quite follow the problem with switching frequency. Having a look at the Hypex charts, and also at the Nuforce V3 technology, they both seem to be very linear and capable of high fidelity reproduction. Where do they lose out? The Nuforce is now employing an improved capacitor board which should help to smooth things out as well, surely?

Examine the frequency response. It is pitifully inadequate for a high end amplifier. The response should be 0dB at 20kHz. ANY deviation at this point means that serious and audible phase shift is occurring well down the frequency range. The only way to deal with this (AFAIK) is to move the switching frequency to a significantly higher point. I would also like to see the figures of any high end amplifier at 1 Ohm loads, as this is mandatory for ESL use.

[georgehifi]

Examine the frequency response. It is pitifully inadequate for a high end amplifier. The response should be 0dB at 20kHz. .

To add to this, I have had the opportunity to hear one of the many Nuforce "version" that tried to get the frequency flatter out to 20khz, (it was still down a bit) by raising the HF switching frequencies filter. It was emitting the nastiest low level HF rubbish from the tweeter that could be heard at a distance of nearly 2mts away and sounded harsh because of it.

Like Zaph said once they can get the switching frequencies much higher then they can change the filters that filter out the switching rubbish, that are at the moment are creating phase shifts well down into the audio band to get rid of the HF rubbish, and to my ears make the mids and treble sound opaque and 2 dimensional.

Once they raise the switching frequency then linear amps will have something to worry about, at the moment digital amps make great subwoofer amps.

Cheers George

Edited June 29, 2011 by georgehifi

[bhobba]

Some big claims here There are links to some technical stuff that I don't understand. Perhaps someone can explain it to me....

Sound quality as measured by what? Exceeding current designs in exactly what audible way? It is well known for example you can achieve spectacular distortion figures by huge amounts of negative feedback, but it sounds rather ordinary - like the life has been sucked out of the music. SET's measure rather poorly yet in the right set-up and on the right music can sound spectacular. Claims like this needs to be backed up by people actually listening to it.

Thanks

Bill

Edited June 29, 2011 by bhobba

[Orpheus]

To add to this, I have had the opportunity to hear one of the many Nuforce "version" that tried to get the frequency flatter out to 20khz, (it was still down a bit) by raising the HF switching frequencies filter. It was emitting the nastiest low level HF rubbish from the tweeter that could be heard at a distance of nearly 2mts away and sounded harsh because of it.

Like Zaph said once they can get the switching frequencies much higher then they can change the filters that filter out the switching rubbish, that are at the moment are creating phase shifts well down into the audio band to get rid of the HF rubbish, and to my ears make the mids and treble sound opaque and 2 dimensional.

Once they raise the switching frequency then linear amps will have something to worry about, at the moment digital amps make great subwoofer amps.

Cheers George

Interesting technical explanation of just what I heard (though I would call the mids and trebles harsh and lifeless) when I listened to Nuforce, Channel Island, and Bel Canto amps some years ago.

[georgehifi]

Interesting technical explanation of just what I heard (though I would call the mids and trebles harsh and lifeless) when I listened to Nuforce, Channel Island, and Bel Canto amps some years ago.

Yes harsh when the HF filter is not low enough so it filters all the HF rubbish out.

And opaque and 2 dimentional when the filter is low enough so it does filter most of the HF rubbish, but then becomes too low as now it cuts into the audio band and creates phase shifts. It's flawed either way.

Like Zaph said move the switching frequency up much higher (in the future when the technology becomes availible) so it can be filtered properly without effecting the audio band and causing phase shifts, maybe then they'll have a winner.

Cheers George

[dart6]

Are you guys familar with spectron D class amps. They've been going for many years and are quite popular in the states www.spectronaudio.com

[THOMO]

I do not understand why frequency responses of amplifiers need to be perfectly flat throughout the audible spectrum when speakers never are.

In another post Statman posted a response of a valve amp driving a typical speaker load.The response looked bad on a plot but then when you looked at it closely it was flat to + or - about 0.6db.

Most speakers are at best + or - 2db throughout the audible spectrum[more typically 6db] and in a real room rather than an anechoic chamber far less flat than that.

So surely any slight deviation from a flat response in an amplifier will be masked by any speakers you are using.

[Keith_W]

Can I ask a question? It is my understanding that most of those F/R graphs of amplifiers are measured against a dummy load. Speakers are not dummy loads - the impedance can vary by quite a bit. Does this mean that the real world frequency response of many amplifiers is quite different to what the measurements would suggest? I am guessing the answer would be yes :nana

[valvelover]

Can I ask a question? It is my understanding that most of those F/R graphs of amplifiers are measured against a dummy load. Speakers are not dummy loads - the impedance can vary by quite a bit. Does this mean that the real world frequency response of many amplifiers is quite different to what the measurements would suggest? I am guessing the answer would be yes :nana

Interesting question. let's take it one step further - how does this affect Class D amps in real world situations I wonder? I've never heard a class D amp.

Sound quality as measured by what? Exceeding current designs in exactly what audible way? It is well known for example you can achieve spectacular distortion figures by huge amounts of negative feedback, but it sounds rather ordinary - like the life has been sucked out of the music. SET's measure rather poorly yet in the right set-up and on the right music can sound spectacular. Claims like this needs to be backed up by people actually listening to it.

Thanks

Bill

I agree Bill. I seem to remember CDs being perfect and better than LPs. Still waiting for that promise to happen ;-) I find it amusing that a 60 year old technology can teach CDs a lesson ;-) I suspect that the same will be true of Class D amps versus other designs. I'll use an analogy - just because a woman looks nice to the eyes, doesn't mean she is nice as a person. The same applies to measurements and hi fi gear imho.

Dave

[georgehifi]

Can I ask a question? It is my understanding that most of those F/R graphs of amplifiers are measured against a dummy load. Yes usually 8ohms, Stereophile does give also a graph of real speaker loads frquency response as well when they test amps.

Speakers are not dummy loads - the impedance can vary by quite a bit. Does this mean that the real world frequency response of many amplifiers is quite different to what the measurements would suggest? I am guessing the answer would be yes :nana

In part yes, if you have a amplifier that has highish output impedance you will not get a ruler flat response when measured against a real speaker load these amps are usually tube and low feedback in design, and it can work this way a little with transistor amps as well. The more feedback the lower the output impedance the higher the dampimg factor. The real trick is design an amp to be low output impedance (high damping factor) and not to have too much fedback all at the same time.

Cheers George

[Zaphod Beeblebrox]

I do not understand why frequency responses of amplifiers need to be perfectly flat throughout the audible spectrum when speakers never are.

Not so. SOME loudspeakers do present very flat response curves, through parts of their range. Most do not. An amplifier designer (and speaker designer, CD player designer, et al) should always design for the most perfect possible performance parameters. It makes no sense to design a deliberately crippled product. It is also important to realise that the PHASE response of a system can cause serious sonic degradation. The phase shift caused by the output filters on some amps (including most Class D designs) can be quite audible to many listeners.

In another post Statman posted a response of a valve amp driving a typical speaker load.The response looked bad on a plot but then when you looked at it closely it was flat to + or - about 0.6db.

Most speakers are at best + or - 2db throughout the audible spectrum[more typically 6db] and in a real room rather than an anechoic chamber far less flat than that.

So surely any slight deviation from a flat response in an amplifier will be masked by any speakers you are using.

As my dad used to say:

"Two wrongs do not make a right." Just because a speaker system may exhibit a falling HF response (though many certainly do not), that does not suggest that it is OK for an amplifier to exhibit a failing in this area as well.

It is 2011 people. We've had the technology to build amplifiers that exhibit a perfectly flat frequency response well past 20kHz for many decades. We've had the technology to build amplifiers that exhibit a perfect phase response to past 20kHz for almost as long. When Class D amplifier can catch up to what was being built 30 years ago, they will be ready to mix it with the big boys. Not until then.

[Keith_W]

Yes Zaph it is 2011. Some people might say that big Class A/B amps are the hifi equivalent of an exhaust belching Hummer. Every time you listen to an hour of Mahler, the glaciers melt just that little bit more