Class A “class D killer” amplifier with THD less than -120dB

[garbz]

What would be lift without sarcasm

A walk up the stairs. Last thing I need from a lift is for it to say, nah you're still on the ground floor, I was only sarcastic.

Maybe I missed it, but I didn't get a sense of sarcasm.

 

[gene_stl]

Amps like these are great for multiamped systems. I am always loathe to attach 200wpc amps to my unobtanium tweeters.

I have built lots of amp kits and other electronics from scratch but landed with three similar Pioneer M-22 class A amps. They put out about 30wpc at clipping and measure like a high quality preamp. I have had two of the three on a distortion analyzer lately and the are working fine, thank goodness, in spite of being new in 1976. They are toasters idling about 250 watts iirc.

 

[Hipocrates]

With today’s inefficient loudspeakers

I don't know man... tons of options regarding high efficiency speakers, devore, tekton, zu audio or beyma, faitalpro for DIY community comes to my mind.

IMHO, a couple of these modules and a pair of coaxials beyma project, seems like fun to me and way different than just another bookshelf/hypex/purify/TPAs build.

I like the name, is fun and ''catchy''.

 

[pma]

The name of this thread is a joke, as many of you have realized. The purpose is not to bring a “killer” amplifier, but to show the possibilities of linear power amplifier design (in distortion, bandwidth, step response rise time, spectrum cleanliness) as a goal for SOTA class D amplifiers. Even the best ones approach to linear amplifiers only in mid-frequency distortion. Their one and only advantage is low power consumption and this is not enough for a SOTA design.

The amplifier shown can give much more power by simply raising the power supply voltage with unchanged bias current. Then it will pass to the class AB above 36W/8ohm or 18W/4ohm. However, in all cases the amp is not practical due to amount of heat it dissipates and for a multichannel amplifier it is absolutely unusable. I have been operating it for 2 years as a stereo amplifier in a casework with large side heatsinks, still getting hot up to some 50°C surface temperature. So it should be understood it is not a proposed direction of amplifier design, only the achievable parameters are proposed.

To add some parameters, output noise is 23uV/@22kHz BW and 17uV A-weighted. This output noise defines achievable SINAD by simple division of Vout/Vn, because distortion is considerably lower compared to noise. Voltage gain is 27dB.

Below is the THD vs. frequency plot into my complex dummy load that simulates a 2-way vented speaker, and it also simulates LF speaker impedance non-linearity by using a ferrite core inductor in LF resonant circuit. It is by far the best result with this dummy load, compared to NC252MP and class AB amplifiers I have tested.

Dummy load impedance, with EPDR plotted.

 

[DearWolf]

Nice phase shift!

Ideal for first order crossover, time coherent speakers.
Somebody who have acoustical second or above orders in loudspeaker doesn't care about amplifier bandwidth or phase response because its already messed up in speakers...

 

[dualazmak]

Than you OP @pma,

This is very interesting and informative thread, and I assume your intensive investigation would somewhat support my recent intensive amplifier exploration journey in my multichannel multi-driver multi-way multi-amplifier stereo project. The summary of my amplifier exploration can be found in my post here.

Amps like these are great for multiamped systems. I am always loathe to attach 200wpc amps to my unobtanium tweeters.

I fully agree with @gene_stl. As I wrote in my above cited post, (almost) all of the home-use Hi-Fi amplifiers, I mean integrated amps and power amps, are designed for full range operation, i.e. to cover ca. 20 Hz - 30 kHz. This means that we should be very much careful in evaluating and selecting each amplifier to directly and dedicatedly drive each of the SP drivers (in my case woofers, Be-squawkers, Be-tweeters and horn super tweeters). These Be-squawkers Be tweeters and horn super tweeters are highly efficient in response to amp's power input, and I found that HiFi Class-A (or quasi Class-A) amps with "reasonable" power reserve are very excellent candidates to drive these high-Fq SP units.

As for the possible "pros" of Class-A amps, especially in multi-amplifier systems, I referred and discussed a lot in my several posts;
- Background and discussions on possible use of class-A amplifier(s) in multichannel multi-amplifier project: #265, #267, #269, #270, #273, #275, #281

In the English Brochure for ACCUPHASE A-36, a pure class-A power amplifier, they described;
"Current feedback principle assures excellent phase characteristics in high range. As shown in the illustration, the A-36 uses the output signal current rather than voltage for feedback. Since the impedance at the current feedback point is very low, there is almost no phase shift. A minimal amount of NFB therefore results in maximum improvement of circuit parameters." (Please refer to my post here.)
I believe this would be one of the nice aspects of pure Class-A (and/or quasi Class-A) amps for tweeters and super tweeters, even though some people in ASP Forum said that above descriptions on A-36 are just marketing messages of completely nonsense contents.

BTW, how do you, OP @pma and @gene_stl, feel and evaluate the four comments/suggestions in my post here which I received from a top engineer of a small and excellent Japanese audio manufacturer, who is currently designing and developing a Purifi 1ET400A-based class-D amp. (Again, some people in ASR Forum said these comments are fully nonsense...)

 

[Geert]

You can make it also same good in class AB (see Benchmark AHB2).

The AHB2 is the reference of course, but the 'Cambridge Audio Edge A' also measures extremely well and applies a 'class XD' topology. As a result it's more power efficient than class A, while showing better THD than class AB.

THD Class AB versus XD ('XD PP' = push pull principle as applied by CA):

(Graph from 'Audio Power Amplifier design book 5th edition - Douglas Self').

You can compare measurements of the 'CA Edge A' against the 'NAD M33', utilising a Purifi class D output stage at Stereophile.com. It beats the Purify on most measurements, bandwidth for example is about double.

I wanted to bring this under the attention as 'class XD' is an interesting evolution that's usually overlooked, probably because it's proprietary to CA.

(Just mentioning this because of technical interest, no opinions about audible performance or cost effectiveness versus class D).

 

[KSTR]

[...], even though some people in ASR Forum said that above descriptions on A-36 are just marketing messages of completely nonsense contents.

Exactly.
Current-mode operation of a voltage feedback does not have any intrinsic bandwith advantage over voltage-mode voltage feedback.
Amplifiers are linear minimum-phase systems and therefore phase shift *only* depends on bandwith (actually, on the low-pass transfer function).

 

[joeren]

Can you show measurements of intermodulation distortion and preferably also some objective listenings tests that support your claim?

”Objective“ listening tests?

 

[Bjorn]

”Objective“ listening tests?

That would of course be blind tests.

However, showing measurements of IMD in a matter we know are audible would be sufficient. So far my question has been overlooked and nothing shared to back up the claims.

 

[Jukka]

... and a space heater, all in one compact, elegant package!


I jest-- the performance of this little amp looks quite good. I'd love to have the chance to hear it, especially on a pair of high-sensitivity loudspeakers!

If memory serves, Bruno said that the only advantage of class D compared to other types, is efficiency.

This amp then is not a class D killer, because it's really bad at efficiency. Yet another clickbait.

Unsubscribed.

 

[gene_stl]

BTW, how do you, OP @pma and @gene_stl, feel and evaluate the four comments/suggestions in my post here which I received from a top engineer of a small and excellent Japanese audio manufacturer, who is currently designing and developing a Purifi 1ET400A-based class-D amp. (Again, some people in ASR Forum said these comments are fully nonsense...)

The four comments are largely nonsense, and even where not nonsense, clearly the engineer is acting in a sales capacity, to promote his company's "next to be released line of products."

At this moment I don't own any class D equipment. I like the pricing of it. I don't mind the efficiency. (But anyone who says you shouldn't use a class A amp because of environmental concerns is an idiot. The whole consumer electronics and western lifestyle is a travesty in that department)

One of the reasons I resisted them is that they are like switching mode power supplies and have a "carrier frequency which is several hundred kilohertz usually. That needs to be blocked by the output inductors. I guess I have still have resistance to that just as I still prefer linear power supplies. Probably has to do with age because I do know that switchers can be made quieter BECAUSE the ripple noise is a much higher frequency.
But they don't provide as much lightning protection as a big clunky transformer. And similarly some DSD D/A s seem to put out some high frequency noise.

I will stick with class A and class A/B for whatever time I have left except possibly for woofer amps (though unlikely because I am very well stocked)

There are a lot of class D boards available at low prices for a lot of power. These did have some interest for me. But one dealbreaker is I don't see how to repair them. That heatsinked chip in the middle. Too much mystery. I don't have a deep understanding of "the magic". I have, and can read, any of the schematics of every single amplifier I have in this house (don't ask me the number. it is on the order of 20). When one of them gets sick I know exactly what to do. I am too old to break in a new amplifier class.

 

[joeren]

The four comments are largely nonsense, and even where not nonsense, clearly the engineer is acting in a sales capacity, to promote his company's "next to be released line of products."

At this moment I don't own any class D equipment. I like the pricing of it. I don't mind the efficiency. (But anyone who says you shouldn't use a class A amp because of environmental concerns is an idiot. The whole consumer electronics and western lifestyle is a travesty in that department)
One of the reasons I resisted them is that they are like switching mode power supplies and have a "carrier frequency which is several hundred kilohertz usually. That needs to be blocked by the output inductors. I guess I have still have resistance to that just as I still prefer linear power supplies. Probably has to do with age because I do know that switchers can be made quieter BECAUSE the ripple noise is a much higher frequency.
But they don't provide as much lightning protection as a big clunky transformer. And similarly some DSD D/A s seem to put out some high frequency noise. I will stick with class A and class A/B for whatever time I have left except possibly for woofer amps (though unlikely because I am very well stocked)

There are a lot of class D boards available at low prices for a lot of power. These did have some interest for me. But one dealbreaker is I don't see how to repair them. That heatsinked chip in the middle. Too much mystery. I have and can read any of the schematics of every single amplifier I have in this house (don't ask me the number. it is on the order of 20). When one of them gets sick I know what to do. I am too old to break in a new amplifier class.

I’ve never owned a class A amplifier but over the years have owned a number of Adcom, Carver, Marantz, Emotiva and other AB amps. I‘ve worked with high frequency PWM motor drives (80kHz SF), PWM power supplies and SMPS, as a technician for GE Global Research for 25 years. I’ve kept up with the technology since retiring in 2020. I took a leap of faith and curiosity in 2021 and purchased a Legacy Powerbloc2, to drive my ML 60XTis. I couldn’t be happier with the purchase for a number of reasons. I didn’t have to break my back to carry it upstairs, as I did with the last 70 lb amp, I put my ear right up the the ML ribbon tweeter and hear dead silence, for the nominal 4Ω impedance the amp will pump out 650WPC so it won’t ever run out of power on the highest transients. The amplifier sounds fantastic with very tight bass and outstanding clarity. I’m sure that there are many bad examples of class D on the market, but this isn’t one of them. From my point of view I don’t ”see” (hear) any downside. Is there still room for improvement? Absolutely. The Legacy amps PWM switching frequency is about 450kHz with a rated frequency response is 1Hz to 100kHz. Newer GaN devices can push the switching frequencies into the MHz and significantly reduce the dead time. In the end, I believe that class D will be the dominant amplifier technology.

 

[gene_stl]

I’ve never owned a class A amplifier but over the years have owned a number of Adcom, Carver, Marantz, Emotiva and other AB amps. I‘ve worked with high frequency PWM motor drives (80kHz SF), PWM power supplies and SMPS, as a technician for GE Global Research for 25 years. I’ve kept up with the technology since retiring in 2020. I took a leap of faith and curiosity in 2021 and purchased a Legacy Powerbloc2, to drive my ML 60XTis. I couldn’t be happier with the purchase for a number of reasons. I didn’t have to break my back to carry it upstairs, as I did with the last 70 lb amp, I put my ear right up the the ML ribbon tweeter and hear dead silence, for the nominal 4Ω impedance the amp will pump out 650WPC so it won’t ever run out of power on the highest transients. The amplifier sounds fantastic with very tight bass and outstanding clarity. I’m sure that there are many bad examples of class D on the market, but this isn’t one of them. From my point of view I don’t ”see” (hear) any downside. Is there still room for improvement? Absolutely. The Legacy amps PWM switching frequency is about 450kHz with a rated frequency response is 1Hz to 100kHz. Newer GaN devices can push the switching frequencies into the MHz and significantly reduce the dead time. In the end, I believe that class D will be the dominant amplifier technology.

Agreed

In another life I work with a lot of jewelers and gemologists on equipment. When LEDs came out certain microscope companies adopted them because they were much more efficient than quartz halogen lamps and were easier to implement, and also cooler. But white LEDs use the same white fluor powders as fluorescent lamps. They just use a blue LED as an exciter instead of a low pressure mercury discharge. The early LEDs had a big peak in the blue wavelength range (BAD frequency response where flat is what you want) So I was resisting. But I did know without a doubt that LEDs were an idea whose time had come. I think the same is true for class D amps. An idea whose time has come. Of course now the Color Rendition Index of LEDs at the color temp of halogens is in the very high nineties. However to have a daylight lamp (5K kelvin color temp) with a very high CRI you still need a xenon arc lamp.

I am sure a lot of the development of Class D was piggy backed on devices built for HF motor drives and similar technology.

I am not opposed to them. I just have 20 good amps that I have collected and there are enough spare that amps are not things that are going to be rotated out of stock.

I agree with you about listening for noise at drivers. Something I have always done. Noisy amps don't get to stay in my system. There is no speaker level crossover to provide attenuation.

 

[boXem]

...
I am sure a lot of the development of Class D was piggy backed on devices built for HF motor drives and similar technology.
...

The first class D amps that can be classified as "HiFi", for what it's worth, are indeed based on FETs designed for diesel engine injector drive

 

[gene_stl]

I do enjoy reading the propaganda that comes out when they make a new incremental improvement , for example when the GaN transistors came out. Since these are published by big hardcore engineering firms, although they promote their new lines of products they don't do it with snake oil , generally. So you can think to yourself, "Well the class Ds are going to get even better now" Of course they are pretty darn good already.
But how bout that Topping amp?!

 

[joeren]

I do enjoy reading the propaganda that comes out when they make a new incremental improvement , for example when the GaN transistors came out. Since these are published by big hardcore engineering firms, although they promote their new lines of products they don't do it with snake oil , generally. So you can think to yourself, "Well the class Ds are going to get even better now" Of course they are pretty darn good already.
But how bout that Topping amp?!

I’ve been following https://epc-co.com/epc. Dr John Glaser and Dr. Michael de Rooij are both former colleagues from GE Global Research, that now work for EPC. Both are outstanding engineers.

 

[joeren]

The first class D amps that can be classified as "HiFi", for what it's worth, are indeed based on FETs designed for diesel engine injector drive

I’m thinking back when I was doing single phase testing of a mult-level PWM inverter, 3100V. The load was a hulking inductor in a big faraday cage. At the end of a 30 minute run, you couldn’t put your hand on the control room metal door. Eddy currents are a bitch.

 

[dualazmak]

Thank you indeed, @KSTR, @gene_stl, @joeren and @boXem | audio, for your kind response to my post #26 above followed by your interesting and suggestive discussions. I well understood those.

I experienced and now feel that the total performance and sound quality of class-D amps still greatly depend on the amp building design engineering as a whole, of course the selection of the class-D modules i.e. which Hypex ones or Purifi ones.

Just for example, in my specific multichannel project, "at high-end extreme side" I evaluated (in my posts #228-#230 on the project thread) extraordinary expensive Dentec DP-NC400-4-EXP (class-D) multichannel (4-Ch) amplifier (four of super-cryo treated Hypex NC400 plus huge linear transformer power supply in outer boxes, not a PWM Switching Mode Pulse PS), and "at rather low-end budget side" I evaluated (in my posts #290- #291 on the thread) TEAC AP-505 class-D stereo power amp (Hypex NC122MP based TEAC version plus PWM PS) as well as its integrated version AX-505.

Through my rather subjective evaluations and comparisons between these two class-D amps, DP-NC400-4-EXP and AP-505, I essentially found and confirmed @gene_stl's points and concerns especially regarding the performance (and power reserve?) of the PS modules. I assume the (heavy-duty) linear transformer PS approach (not a PWM Switching Mode PS) for Hypex or Purifi class-D modules, like in DP-NC400-4-EXP, would be one of the interesting routes for future reasonably priced excellent class-D PWM digital amps.

In any way, at the end of my intensive amplifier exploration journey, I decided not to use class-D (at least) at present, and I have selected rather conventional class-AB and quasi class-A integrated amplifiers in my multi-amplifier project. I would like carefully keeping my eyes on the rapid progress of class-D amps including the further possible utilization of GaN FET.

 

[joeren]

Thank you indeed, @KSTR, @gene_stl, @joeren and @boXem | audio, for your kind response to my post #26 above followed by your interesting and suggestive discussions. I well understood those.

I experienced and now feel that the total performance and sound quality of class-D amps still greatly depend on the amp building design engineering as a whole, of course the selection of the class-D modules i.e. which Hypex ones or Purifi ones.

Just for example, in my specific multichannel project, "at high-end extreme side" I evaluated (in my posts #228-#230 on the project thread) extraordinary expensive Dentech DP-NC400-4-EXP (class-D) multichannel (4-Ch) amplifier (four of super-cryo treated Hypex NC400 plus huge linear transformer power supply in outer boxes, not a PWM pulse PS), and "at rather low-end budget side" I evaluated (in my posts #290- #291 on the thread) TEAC AP-505 class-D stereo power amp (Hypex NC122MP based TEAC version plus PWM PS) as well as its integrated version AX-505.

Through my rather subjective evaluations and comparisons between these two class-D amps, DP-NS400-4-EXP and AP-505, I essentially found and confirmed @gene_stl's points and concerns especially regarding the performance (and power reserve?) of the PS modules. I assume the (heavy-duty) linear transformer PS approach (not a PWM PS) for Hypex or Purifi class-D modules, like in DP-NC-400-4-EXP, would be one of the interesting routes for future reasonably priced excellent class-D amps.

In any way, at the end of my intensive amplifier exploration journey, I decided not to use class-D (at least) at present, and I have selected rather conventional class-AB and quasi class-A integrated amplifiers in my multi-amplifier project. I would like carefully keeping my eyes on the rapid progress of class-D amps including the further possible utilization of GaN FET.

Thank you as well for the feedback. There are some advantages to using linear power supplies as opposed to Switch Mode Power Supplies, which by the way work differently from PWM. Linear supplies are more robust but also much heavier and less efficient. I well engineered SMPS can easily provide the necessary power. The fundamental switching frequency must be high enough to be well out of the audio spectrum, so it may be filtered.