Class D versus the rest

0 Members and 1 Guest are viewing this topic. Read 189249 times.

Freo-1

Re: Class D versus the rest
« Reply #260 on: 5 May 2012, 02:51 am »
Freo-1:

I run my amps in triode for the most part, but funny enough, I find some genre's of music actually sounds better in ultra, rock, the blues mostly. Could be the way they were mastered.

If I recall correctly, in class a/b, you lose the even harmonics.  I would think that the same holds true for switching amps (class d). Does this sound right, or is it possible for class d to produce all the harmonics even though they're switching amps?

Jim

That is a good question.  Here is a link to address distortion regarding Class D amps:

[url][http://www.diyaudio.com/forums/class-d/174018-thd-n-class-d-amps-hd-noise-limited.html/url]


Freo-1

Re: Class D versus the rest
« Reply #261 on: 5 May 2012, 03:02 am »
Here is a cool link to infor about Class D amps:

http://www.analog.com/library/analogDialogue/archives/40-06/class_d.pdf

OzarkTom

Re: Class D versus the rest
« Reply #262 on: 5 May 2012, 03:05 am »
I am afraid that I am eating my words about Class D amps after reading this review and buying the TBI Millenia. I have a friend that bought one too, and he is as astounded as I am.

http://www.tnt-audio.com/ampli/3_mini_amps_e.html

You can hear the singers grunt and groan and smack their lips in between their verses, but the amp still retains the magic of any Class A or tube amp that I ever heard. It never becomes sterile or SS sounding.

The battery makes it a world class amp, the switching power supply that comes with it just makes it OK. At $500, this is as great a value as the Musical Fidelity A1 Class A integrated ever was. Maybe I should gut the electronics out and put it in a DIY case. I could probably sell it for $1K to the DIY'ers.

Tube amps with output tranformers sound very veiled in comparison.  I never thought I would like a Class D amp in a million years.

Freo-1

Re: Class D versus the rest
« Reply #263 on: 5 May 2012, 03:07 am »
Freo-1:

Thanks for the clarification.  For me it is a personal choice, after hearing it in person. I buy what my ears like and it doesn't matter what class it is. With that said, my ears like tubes, that is not to say that a non-tube amp will capture my attention if it sounds good to my ears.

I plan to attend the upcoming show in Newport Beach, CA, no doubt, there will be a bevy of Class D strutting their thing.

Jim

Jim,

Anothere link:

[url][http://www.diyaudio.com/forums/class-d/176178-clipping-distortion-class-d-amplifier.html/url]

medium jim

Re: Class D versus the rest
« Reply #264 on: 5 May 2012, 03:34 am »
Freo-1

I was doing some research and came up with this 2009 disertation from the University of Berkley:

http://rfic.eecs.berkeley.edu/~niknejad/ee242/pdf/eecs242_class_EF_PAs.pdf

On Page 21 is the following statement regarding class d:

Switch capacitance limits efficiency in high frequency
applications.

Maybe this is why I felt a tightness or grain in the high frequencies of the Bel Canto setup that a friend has.   

I have a Premier (Pioneer) Head Unit in one of my vehicles that is Class D and it sounds amazing in that application and feel no fatigue even on long excursions.  However, I'm not doing critical listening while driving.

Jim

Freo-1

Re: Class D versus the rest
« Reply #265 on: 5 May 2012, 12:07 pm »
Freo-1

I was doing some research and came up with this 2009 disertation from the University of Berkley:

http://rfic.eecs.berkeley.edu/~niknejad/ee242/pdf/eecs242_class_EF_PAs.pdf

On Page 21 is the following statement regarding class d:

Switch capacitance limits efficiency in high frequency
applications.

Maybe this is why I felt a tightness or grain in the high frequencies of the Bel Canto setup that a friend has.   

I have a Premier (Pioneer) Head Unit in one of my vehicles that is Class D and it sounds amazing in that application and feel no fatigue even on long excursions.  However, I'm not doing critical listening while driving.

Jim

I too have run across that paper.   Personally, never have been a fan of the HF domain with Class D.  It's one of the main reasons I do not keep any Class D amps long term for home audio.  Mobile audio is a whole different issue.

Here is an interesting paper from National Semiconductor:

http://www.national.com/assets/en/appnotes/ClassDAmplifierFAQ.pdf

werd

Re: Class D versus the rest
« Reply #266 on: 5 May 2012, 02:10 pm »
Speaking of class D, anybody else hear of those new amps from Hypex. I think they are called ?

In my most robotic voice with arms twitching robotically up and down and upper torso robotically moving side to side.

"Must buy Ncores... Must buy Ncores !!!"

Freo-1

Re: Class D versus the rest
« Reply #267 on: 6 May 2012, 05:06 pm »
Freo-1

I was doing some research and came up with this 2009 disertation from the University of Berkley:

http://rfic.eecs.berkeley.edu/~niknejad/ee242/pdf/eecs242_class_EF_PAs.pdf

On Page 21 is the following statement regarding class d:

Switch capacitance limits efficiency in high frequency
applications.

Maybe this is why I felt a tightness or grain in the high frequencies of the Bel Canto setup that a friend has.   

I have a Premier (Pioneer) Head Unit in one of my vehicles that is Class D and it sounds amazing in that application and feel no fatigue even on long excursions.  However, I'm not doing critical listening while driving.

Jim

 
Jim,
 
The following may explain why HF with Class D is not always pristine:
 
 
"  In addition, class D amplifiers may contain elevated noise in the region above 20 kHz. Normally, a 20 kHz low-pass filter is used to limit noise measurements to the audible frequency range. However, the 3 to 6 pole low pass filter used in most audio analyzers may not be sharp enough to eliminate all of this extra noise, so the measurement accuracy may be impacted. A better filter to use is a low-pass filter conforming to the AES-17 standard. Originally intended for measuring D/A converters, this sharp low-pass filter is also ideal for measuring class D amplifiers. For Audio Precision's 2700 Series, the optional S-AES17 filter is available. All of the built-in digital filters present in APx analyzers already provide a sharp cut-off, so this issue is no longer a concern.
 
Traditional distortion measurements may not catch high frequency distortion problems, which can be present in class D amplifiers. That's because the usual THD+N measurement, using a low-pass cut-off frequency of 20 kHz, can't measure 3rd harmonics of fundamentals over about 6 kHz. The usual SMPTE IMD measurement is also insensitive to high frequency distortion, as it uses a fundamental of only 7 kHz. A twin-tone IMD measurement, using two high frequency fundamentals at 18 kHz and 20 kHz, solves this problem and reveals high frequency distortion problems. Both the 2700 Series and the APx Series include twin-tone DFD IMD testing capability.
 
Common Challenges:
- High frequency switching artifacts can cause distortion on an analyzer's inputs. A high quality external filter, such as AP's two channel AUX-0025 or eight channel AUX-0100, is essential to reduce high frequency switching noise that can affect measurement accuracy.
Out-of-band noise can affect noise measurements when using a standard low-pass filter. A sharp low-pass filter will keep significant out-of-band noise from affecting the results. The filters on the APx Series already have the necessary sharp slope. The optional S-AES17 filter is needed for the 2700 Series.

- High frequency distortion is missed with traditional THD+N and SMPTE IMD tests. The twin-tone (DFD) IMD test, available on the APx and 2700 Series, will reveal high frequency distortion missed by the other tests."

medium jim

Re: Class D versus the rest
« Reply #268 on: 6 May 2012, 05:31 pm »
Freo-1

Thanks for that.  Who knows, maybe someone will be able to figure it all out, but I would surmise, it will be called something other than Class D at that point. 

Jim

Freo-1


earflappin

Re: Class D versus the rest
« Reply #270 on: 6 May 2012, 06:53 pm »
I am afraid that I am eating my words about Class D amps after reading this review and buying the TBI Millenia. I have a friend that bought one too, and he is as astounded as I am.

http://www.tnt-audio.com/ampli/3_mini_amps_e.html

You can hear the singers grunt and groan and smack their lips in between their verses, but the amp still retains the magic of any Class A or tube amp that I ever heard. It never becomes sterile or SS sounding.

The battery makes it a world class amp, the switching power supply that comes with it just makes it OK. At $500, this is as great a value as the Musical Fidelity A1 Class A integrated ever was. Maybe I should gut the electronics out and put it in a DIY case. I could probably sell it for $1K to the DIY'ers.

Tube amps with output tranformers sound very veiled in comparison.  I never thought I would like a Class D amp in a million years.

Tom, nice post.  Your experience is similar to mine.  I was a die hard Class A guy for years - SS and tubes, including OTL's.  I tried Class D several years ago and hated them so I refused to try it again until I read about all the buzz on the ClassD SDS kit amps.  While the SDS was better than my Class A amps in a couple of areas, it had an over-caffeinated, glassy, unnatural sounding HF so I went back to my Class A's. 

Then I tried a Hephaestus Class D amp.  The Hepha was better overall than the SDS (but then it should be given its price point).  It had the tonality of excellent Class A amps and a more natural HF, but was not as resolved and dynamic as top Class A amps.  For example, it took only a few seconds to hear that the Clayton M300 Class A's were superior to the Hepha's.

Then I bought the NCORE 400's in late January.  I know many on AC are tired of the perceived hyping of the Hypex NCOREs...   :lol:  And I get that sentiment, as I have felt the same way with many other products that have been highly touted on this and other forums, only to be sold in droves a year later.  And who knows for sure if the NCOREs won't go the same way?

All I can say personally is that in my system and to my ears, I prefer the NCOREs to the: (1) last two Class A amps I have owned (Atma-Sphere M60's and Pass Labs XA30.5 and (2) SDS and Hepha Class D's I still own.  The NCOREs are more accurate, resolving and musical and they are the only Class D amp I have heard that finally gets the HF right.  Will they be everyone's cup of tea?  No. 




TRADERXFAN

  • Full Member
  • Posts: 1061
  • Trillions will vanish... it's a debt blackhole.
    • GALLERY
Re: Class D versus the rest
« Reply #271 on: 6 May 2012, 06:57 pm »
I think there is another angle to this, and that is the psycho-acoustic. If you cannot hear it, does it matter? 

I think it was Geddes, [who has done scientific research in perception, not just as a speaker designer]  saying that the typical amplifier distortion measurements don't correspond with perceived sound quality. That only certain types of distortion, and that being related to "crossover distortion" and the measured result of this distortion as a specific amp would switch from A to B on an AB.

I think he said that the thermal tracking of the components were important in managing this and so his leaning toward chip amps for this as a low cost effective solution to get high quality amplification --I am not very technically inclined, so most likely am misconstruing something here.  So please take this just as an idea to broaden the area of discussion for why differences in topology and execution exist, not that I have represented this correctly.

 So in a class D, it may not result in the same type of distortion and so perceived sound quality could be high relative to class AB for this reason. 

-Tony

Freo-1

Re: Class D versus the rest
« Reply #272 on: 6 May 2012, 07:07 pm »
Tom, nice post.  Your experience is similar to mine.  I was a die hard Class A guy for years - SS and tubes, including OTL's.  I tried Class D several years ago and hated them so I refused to try it again until I read about all the buzz on the ClassD SDS kit amps.  While the SDS was better than my Class A amps in a couple of areas, it had an over-caffeinated, glassy, unnatural sounding HF so I went back to my Class A's. 

Then I tried a Hephaestus Class D amp.  The Hepha was better overall than the SDS (but then it should be given its price point).  It had the tonality of excellent Class A amps and a more natural HF, but was not as resolved and dynamic as top Class A amps.  For example, it took only a few seconds to hear that the Clayton M300 Class A's were superior to the Hepha's.

Then I bought the NCORE 400's in late January.  I know many on AC are tired of the perceived hyping of the Hypex NCOREs...   :lol:  And I get that sentiment, as I have felt the same way with many other products that have been highly touted on this and other forums, only to be sold in droves a year later.  And who knows for sure if the NCOREs won't go the same way?

All I can say personally is that in my system and to my ears, I prefer the NCOREs to the: (1) last two Class A amps I have owned (Atma-Sphere M60's and Pass Labs XA30.5 and (2) SDS and Hepha Class D's I still own.  The NCOREs are more accurate, resolving and musical and they are the only Class D amp I have heard that finally gets the HF right.  Will they be everyone's cup of tea?  No.

 
Interesting post.  What speakers are you using? 
 
I could understand how a higher powered Class D amp would work better on some speakers than a XA 30.5, but for speakers that are moderately efficient, it is more of a challenge.  HF is where Class D to me falls short (vocals as well, but less so).
Something to consider is what kind of load is presented to the speaker.  This link was posted on another forum to discuss this aspect:
 
http://www.audiograph.se/Downloads/PowerCube_12p_brochure_complete.pdf

earflappin

Re: Class D versus the rest
« Reply #273 on: 6 May 2012, 07:19 pm »

 
Interesting post.  What speakers are you using? 
 
I could understand how a higher powered Class D amp would work better on some speakers than a XA 30.5, but for speakers that are moderately efficient, it is more of a challenge.  HF is where Class D to me falls short (vocals as well, but less so).
Something to consider is what kind of load is presented to the speaker.  This link was posted on another forum to discuss this aspect:
 
http://www.audiograph.se/Downloads/PowerCube_12p_brochure_complete.pdf

Freo-1, all of my listening comparisons in the past 2 years have been done on Geddes Abbey 95db efficient 2 way monitors that roll off at about 12db/octave below around 50 hz.  I cannot comment on how the NCORE compares to the Class A amps I referenced on other speakers (less efficient or otherwise), although I have read many posts on various forums where comparisons have been made on more demanding speakers such as the Maggies and others.

Freo-1

Re: Class D versus the rest
« Reply #274 on: 6 May 2012, 07:38 pm »
Freo-1, all of my listening comparisons in the past 2 years have been done on Geddes Abbey 95db efficient 2 way monitors that roll off at about 12db/octave below around 50 hz.  I cannot comment on how the NCORE compares to the Class A amps I referenced on other speakers (less efficient or otherwise), although I have read many posts on various forums where comparisons have been made on more demanding speakers such as the Maggies and others.

Thanks for responding.  My experience with Class D amps on Source Technologies 277SE and Legacy Signature III’s has been a mixed bag at best.  None of them sounded as good as a Threshold SA/3 or Pass Las XA 30.5 on these speakers.
 
Recently obtained a pair of Cary Signature Oak Model Ones, with SEAS Excel drivers.  The overall performance from the XA 30.5 is simply incredible, with life like HF and vocals VERY few systems can achieve.
 
So, am interested in how the latest N Core amps will sound with the Legacy as well as the Cary.  Based on how Class D amps work, it would take quite an engineering achievement to overcome the design limitations associated with the topology in the HF area.
 
I do think speaker interaction has not been talked about enough here.  This alone could explain the wide differences of opinion and results with listening tests.

cab

Re: Class D versus the rest
« Reply #275 on: 6 May 2012, 07:42 pm »
Unlike most other common class d amps such as the tripath, without post filter feedback, Hypex's UCD as well as ncore have an essentially flat frequency response vs speaker impedance down to less than 2 ohm.....

Freo-1

Re: Class D versus the rest
« Reply #276 on: 6 May 2012, 07:59 pm »
Unlike most other common class d amps such as the tripath, without post filter feedback, Hypex's UCD as well as ncore have an essentially flat frequency response vs speaker impedance down to less than 2 ohm.....

How do they measure that? 
 
Example:
 
"  In addition, class D amplifiers may contain elevated noise in the region above 20 kHz. Normally, a 20 kHz low-pass filter is used to limit noise measurements to the audible frequency range. However, the 3 to 6 pole low pass filter used in most audio analyzers may not be sharp enough to eliminate all of this extra noise, so the measurement accuracy may be impacted. A better filter to use is a low-pass filter conforming to the AES-17 standard. Originally intended for measuring D/A converters, this sharp low-pass filter is also ideal for measuring class D amplifiers. For Audio Precision's 2700 Series, the optional S-AES17 filter is available. All of the built-in digital filters present in APx analyzers already provide a sharp cut-off, so this issue is no longer a concern.
 
Traditional distortion measurements may not catch high frequency distortion problems, which can be present in class D amplifiers. That's because the usual THD+N measurement, using a low-pass cut-off frequency of 20 kHz, can't measure 3rd harmonics of fundamentals over about 6 kHz. The usual SMPTE IMD measurement is also insensitive to high frequency distortion, as it uses a fundamental of only 7 kHz. A twin-tone IMD measurement, using two high frequency fundamentals at 18 kHz and 20 kHz, solves this problem and reveals high frequency distortion problems. Both the 2700 Series and the APx Series include twin-tone DFD IMD testing capability.
 
Common Challenges:
- High frequency switching artifacts can cause distortion on an analyzer's inputs. A high quality external filter, such as AP's two channel AUX-0025 or eight channel AUX-0100, is essential to reduce high frequency switching noise that can affect measurement accuracy.
Out-of-band noise can affect noise measurements when using a standard low-pass filter. A sharp low-pass filter will keep significant out-of-band noise from affecting the results. The filters on the APx Series already have the necessary sharp slope. The optional S-AES17 filter is needed for the 2700 Series.
- High frequency distortion is missed with traditional THD+N and SMPTE IMD tests. The twin-tone (DFD) IMD test, available on the APx and 2700 Series, will reveal high frequency distortion missed by the other tests."

cab

Re: Class D versus the rest
« Reply #277 on: 6 May 2012, 08:21 pm »
One of the main reasons for a variety of experiences with many class d amps is due to their pre-filter feedback and the fact that this leads to a filter optimized for a certain load. As a result, they do not have flat frequency response with load and can sound better with certain speakers that happen to present a load near the point that the output filter was optimized for, and worse with those farther away from the design point. Hypex uses post filter feedback, as does a few other manufactures of class d amps. No coincidence perhaps that those with the better specs and reviews are usually post filter feedback designs. They generally have a flatter frequency response irrespective of load.

For further info on Hypex's measurement protocol I would refer you to their website where these issues are discussed and to Bruno himself. I am sure he would be the most capable to address any technical issues you may have with the design and/or performance of his amp.

Freo-1

Re: Class D versus the rest
« Reply #278 on: 7 May 2012, 08:52 pm »
I’m keen to understand how issues (especially with frequency and phase/impedance) were addressed.  I’ll be sure to head over to the DIY and poke around some more.  Will read the technical papers again, but it seemed a little short on some aspects (probably due to IP concerns).

kevinh

  • Jr. Member
  • Posts: 102
Re: Class D versus the rest
« Reply #279 on: 8 May 2012, 10:00 pm »
This link was posted in another thread but the tehcnical aspect as relevant to this thread.

http://www.audiophilejournal.com/hypex-ncore-monoblock-amplifiers-mola-mola/


Quote
The NCore circuit contains:
 1.A mathematically exact understanding of self-oscillation
 2.Improved comparator circuitry insures that actual behaviour matches the theoretical model as closely as possible
 3.New gate drive circuitry improves open-loop distortion at moderate signal levels while significantly reducing idle losses
 4.A new control loop ups loop gain by 20dB across the full audio range without sacrificing stability
 
Amplifiers using all four of the above will be marketed under the name Ncore.
 
Amplifiers using only the first three will still be sold under the existing UcD brand even though their internals no longer resemble that of the well-known 2001 circuit.


It appears that Bruno is addressing some of the technical issues raised in this thread. It looks like the 'next gen' UcD modules could be close to  the Ncore in performance.