Mytek Brooklyn Amp power amplifier

Designed in New York City, manufactured in Poland, and barely bigger than a thick paperback, the Brooklyn Amp ($2495) is Mytek's first power amplifier. Like all of their products, it's sleek to behold, with a powerful look that suggests the company's pedigree: in addition to high-end consumer electronics, Mytek makes gear for the pro-audio market, where exceptional build quality and space-saving design are the norm.

Consistent with that last characteristic is the Brooklyn Amp's output architecture: it operates in class-D, a technology that remains controversial. In my first review for Stereophile, I evaluated the Spec RPA-W7EX Real-Sound class-D power amplifier ($5995). That hefty silver box warmed my cockles with sound that was worthy of comparison to that of tubed amplifiers: delicious spatial performance and tonality that thoroughly surprised me. The mighty Spec warranted superlatives I usually reserve for my Shindo Laboratory Haut-Brion, coming as close to that amp's humanness as one could wish for in a class-D amplifier.

The question: is class-D ready for prime time?

A demure beast designed in Brooklyn
Mytek occupies a former church in Greenpoint, Brooklyn—a building that, as recently as 2010, housed the studio of the artist Kehinde Wiley. From the street, the place looks like a crumbling homeless shelter. There I picked up not one but three Brooklyn Amps: the standard dual-mono stereo version, which produces 300Wpc; and two Amps set up in bridged mode, each working as a 600W monoblock. John Atkinson thought it made sense to listen to the Brooklyn Amp(s) hooked up to Mytek's Brooklyn DAC Plus ($2195), so I made space for one of those, too, in my ever-ready LP backpack.


Because, when in Greenpoint, record shopping is a must. This historically Polish neighborhood offers such vinyl hangs as Record Grouch, Academy Records Annex, The Thing, CO-OP 87, Captured Tracks, and, within a 10-minute walk to Williamsburg, Rough Trade. Excellent Polish bakeries abound, and Peter Pan Donut & Pastry Shop is an absolute treat. But I digress . . .

Deep below ground level, in the labyrinthine bowels of Mytek Central, is a fully operational recording studio. You know you're in a pro-audio space by the powerful air-conditioning system that bathes the studio in lovely freezing air. A smaller, well-isolated studio holds a control console, and two pairs of speakers: massive Duntech Sovereigns (90dB sensitivity), and the smaller Wilson Audio Specialties Sabrinas (87dB). Two Brooklyn Amps in bridged-mono mode were driving the former. Effortless dynamics pounced on me like two hungry tigers.

"Duntechs, with their multiple drivers and complicated crossovers, are a very difficult load," Mytek's chief designer, Michal Jurewicz, told me. "Hypex [class-D modules] cannot drive it, the amps collapse, but this Brooklyn Amp does it with ease. This came out during months of design tests I did in 2016, when we were testing many different circuits." Jurewicz also said that Mytek spent six months experimenting with more than 15 class-D amplifier modules from various OEMs, upgrading parts, adding capacitors, adjusting circuit parameters.

"Class-D is a relatively new technology, and there isn't yet enough consensus in correlating circuit properties vs subjective sound quality," Jurewicz wrote in response to my e-mailed questions. To get any such response, I had to use the journalistic equivalent of dentists' pliers. There's great competition among class-D designers, and the success of Mytek's DACs and preamps—not to mention their Brooklyn DAC+ being one of the first fully functional MQA-ready D/A converters—has put Jurewicz in the catbird seat. He'd rather not answer any questions. Trade secrets.

Though Mytek ultimately chose an amplifier module from the Danish company Pascal A/S, it's far from an off-the-shelf version. "We use one Pascal," Jurewicz wrote, "but the module is heavily modified. Major parts are replaced, and they don't sound anything like the originals (way better, more detailed and warmer). We've changed switching speeds, changed numerous parts in an attempt to significantly reduce distortion, and we beefed-up the power supplies. There are not really any classic transformers inside. It's all well-executed switching technology. That's why it's so small and light."

Inside and Out
Mytek's familiar logo, which looks to me like a backlit Rorschach ink blot, appears in the upper-left corner of the Brooklyn Amp's faceplate; a much larger version is formed by the ventilation holes punched in the top panel. The faceplate's textured surface reminds me of an optical illusion in which the pattern seems to float free of the surface when stared at long enough. On it are only a single power button with a feel that made me want to press it and keep pressing it. I resisted. After the Amp is powered off, that Rorschach jack-o'-lantern on the front panel glows faintly.

On the Amp's well-sorted, easily accessible hindquarters are: one pair each of gold-plated unbalanced (RCA) and balanced (XLR) analog input jacks, two pairs of large speaker binding posts, an IEC power inlet, a small jack for a 12V Power Up trigger, and a row of tiny DIP switches.


Two of the DIPs control gain (0dB up, +6dB down) and mode (balanced up, unbalanced down), and two others work together: "When #7 is up the AMP works in a regular stereo mode," notes the manual. "When #7 is down the AMP mode depends on dipswitch #8 setting; when #8 is up the AMP works in Bi-amp mode. When #8 is down the AMP works in a Bridge mode."

When he explained to me the inner workings of the Brooklyn Amp, Jurewicz cracked open its lid to reveal a dual-mono configuration: a tightly spaced row of eight oversize capacitors on each of two circuit boards, and some similarly large chokes. Unlike DACs or amplifiers whose impressively big cases are mostly empty, the Brooklyn Amp, which measures 8.5" wide by 1.73" high by 9.5" deep, leaves no part of its circuit boards unfilled.

Although the Amp's current owner's manual doesn't provide operating instructions, an updated version will be available soon online, Jurewicz said.

MacDougal Street Setup
In past reviews I've complained until you, dear reader, are no doubt tired of hearing about the treacherous climb to my penthouse pad, in the New York City sky. The four Myteks were together no heavier than two bags of overpriced groceries from our soon-to-be-razed local supermarket. Who needs food when there are luxury condos to build? Gotta house those new-money 30-year-olds somewhere.

I used the single stereo and the two bridged-mono Brooklyn Amps in both of my systems: the smaller rig of Thorens TD 124 turntable, Jelco TS-350S tonearm, Ortofon Quintet Bronze cartridge, and Quad S2 speakers; and my big rig of Kuzma Stabi S turntable with Stogi S tonearm and Hana EL cartridge, PS Audio NuWave DAC, LG Blu-ray player, and DeVore Fidelity Orangutan O/93 speakers.

Michal Jurewicz had loaned me a pair of Mytek Metropolis balanced interconnects to join DAC to Amps. Tellurium and Auditorium 23 speaker cables respectively snaked their ways into rigs small and big. TriodeWire Labs American interconnects also saw duty.

The Mytek Amps are small enough to fit anywhere, but when I placed one under the Brooklyn DAC+ in my small-rig rack, I was surprised when it ran relatively hot.

148 India Street
Brooklyn, NY 11222
(347) 384-2687

Bogolu Haranath's picture

"Watts are cheap" ........... Anonymous :-) ................

Those words are coming true :-) ...............

tonykaz's picture

I was one of Jim Theil & Kathy Gornick's Dealers, I spent time working with Jim, Kathy, Tom Theil at the Nandino Theil Factory.

Amps were the reason for most of my needing input from the Theils who directed me to Threshold Amps and Electrocompaniet Amplification Systems.

I never found a "cheap" Amp that could properly drive any of the Theil Line of Loudspeakers.

Jim never suggested ( to me ) that Watts were "Cheap" !

So, I'll suggest that Jim is being misquoted. I don't think Jim was ever being sarcastic because he wasn't an angry person.

High Quality Amplification has always been expensive in Dollar Cost and expensive to research, locate and own.

Which is one of the great things about Stereophile now-a-days, Stereophile has an abundance of Amplifier evaluators that go to extravagant lengths ( reminiscent of what Jim Theil and I did back in the 1980's ).

Tony in Michigan

Axiom05's picture

It drives me nuts that people can't make the effort to spell someone's name correctly. Jim Thiel.

Bob in Sarasota

rschryer's picture

I'm sure Toony didn't mean anything by it.


Roberb Schcryer

tonykaz's picture


Yes, you are correct.

I've been dysfunctional about getting Jim & Tom's last name spelled correctly, for decades now.

I still have to check my spellings.

However, both Jim & Tom enjoyed a good relationship with me and we found homes for a significant amount of their factory's output.

My misspellings are innocent, I still get their and thier goofed up.

I'll try harder, I'm not trying to upset anyone's sensabilities.

I loved THIEL stuff.

Tony in Michigan

ps. I don't own Thiels now but my old Business Partner : Dr.George Buckley still owns an early serial number pair of CS3's

jeffhenning's picture

Two thoughts about this review:

• I guess you have to crack on Hypex when your amps have more distortion and noise - hey, Pascals are still good amp modules and they've proven themselves as being rugged in the pro market. They are, though, not quite as good as nCore tech.

• If you are building a $2.5K amp with the biggest metric being driving Duntech Sovereigns (truly great 25 years ago) or the like at full blast, you've missed the point. No rational audiophile clings to that paradigm. The "World-Killer", all passive speaker concept is dead. Much like vinyl, it's kept alive by people who value status and bragging rights over true fidelity to the original signal.

It's an amp with "respectable performance." Great. Makes me wonder, though, why Ken was so enthusiastic about it. Most likely the noise floor. That is a big thing, but, given the excess of distortion in the treble, I'd never consider this. So, the amp is OK, but nothing more.

On a different, but, tangential note, when was the last time any product really got called to the mat here or in any audio publication? It happens, but quite rarely here. Elsewhere, almost never.

Personally, I'm tired of reviews that have no measurements and end with a "sounds great to me". Most especially when every product reviewed is more incredible as the price rockets skyward ("This $5k Ethernet cable offers stunning midrange!").

High fidelity is a science. The art aspect comes into play when building the product. Unfortunately, most audiophiles do not embrace science. They embrace hype.

ken mac's picture

"Much like vinyl, it's kept alive by people who value status and bragging rights over true fidelity to the original signal." Is everybody happy?

Bogolu Haranath's picture

"Eschew obfuscation" ........... Anonymous :-) .........

ken mac's picture

Oh, I see you read a different review from a different publication. Nevermind.

georgehifi's picture

Here we have a stereo Class-D amp with 250watt into 8ohms and 300 into 4ohms.
Yet into the 93db 10ohm DeVore Fidelity Orangutan O/93 speakers.
"Using two Amps in bridged-mono mode greatly increased the flesh and blood of the sound of a single Amp, with no loss of dynamics."
How is this possible one amp should have p****d it in.
All I can think of with bridged amps the damping factor is worse, distortion is higher, low impedance drive stability is far less, maybe this can "flesh and blood out for the better"??????? certainly couldn't have been the extra watts one amp was more than enough.

And this ones really interesting for me "We've changed switching speeds" has anyone more info on this statement from Mytek?
As I'm a firm believer that switching speed is the Achilles heel of Class-D at the moment at around 600-800khz, and it should be 2-3 mhz so the output filter/s can do their job properly, only then will Class-D be fully appreciated.

Cheers George

Bogolu Haranath's picture

Wikipedia has lot of info about class-D amps, if you are interested ...........

dce22's picture

Your mixing things up,
"Achilles heel of Class-D at the moment at around 600-800khz, and it should be 2-3 mhz so the output filter/s can do their job properly"

That is for Digital direct Power DACS like this:

For self oscillating analog PWM best frequency is 400kHz:

If you design Class D amp to self oscillate at higher frequency you will have real world physics problems and it will sound bad, it has been tried because most of the distortion in class d comes from the switching moment at high current so if you switch twice the rate you will double the distortion and so forth and all you will get is more flexibility on the negative feedback that you dont need in the first place you have plenty of feedback as it is.


ken mac's picture

on what I hear, and the two amps in bridged mode outperformed the single amp. Not all things correlate to measurements, to stats, to yesterday's news. What I hear outweighs all other criteria.

Ortofan's picture

... a review and test of the Rotel RB-1582 MkII.
For $1600 you get 230W/ch into 8 Ω and 400W/ch into 4 Ω.
Plus no speaker impedance dependent frequency response aberrations.

Bogolu Haranath's picture

Check out "An amplifier listening test" on Stereophile references ..........

georgehifi's picture

Better look at the very limited supply (to Kings and Presidents) $30K aud Technics SE-R1 with special newly developed limited supply high speed "GAN-Fet" transistors (from the same guys who invented the Power Mosfet)
It uses a 1.5mHz switching speed, over twice as much as what around now.

Reactions from just a couple of press/reviewers to it's sound:
"Listening to tracks that we’ve heard 100s of times — and on excellent systems at that — is now a revelation of once hidden nuance and detail. Not only are we hearing things we’d never heard before, we’re hearing it in a way we’ve never heard it before. A music system that sounds like a live performance is a tough goal to attain, but Technics’ flagship nails it."

"This amplifier delivered some of the best reproduced sound at CES 2017. I persuaded Bill Voss to rip the contents of my copy of Rutter's Requiem to his media server's solid-state drive. The broad and deep soundstage, imaging, upper midrange detail, and bass extension were thrilling during the system's rendition of "Lord, make me an instrument of thy peace" and "Pie Jesu." I appreciated the work done by the Technics staff to prepare their exhibit suite with recessed sound-absorbing panels. I left wanting to hear more."

Cheers George

dce22's picture

Technics SE-R1 are so primitive design they dont even put output filter into the feedback loop, it has rising distortion in the high frequency range and alot of phase shift.

It does not matter if you use MOS or GAN FET the actual inductance of the circut board will not allow to delever the current for the rising edge on the pulse so it will distort like a mofo.

The best design of the type of amp that Technics is trying to do is Bruno Putzeys "A True One-Bit Power D/A Converter" by spliting 2.8Mhz frequency into multi phases then sum the phases into the inductor the whole system runs at 2.8Mhz but the FET's switch at low frequency, its zero feedback, volume is controlled by moving the power supply voltage so there is no digital bits losses, and without feedback to control the output filter you get phase shift at the high frequency range just like the technics design.

Self oscillating design like ucd ncore and such, that uses the negative feedback with the output filter to create the switching oscillator and correct for the filter problems and improve the distotion spectrum and reduce the EMI and RF, and reduces parts to around 10 to 15 transistors and improves the signal to noise ration and it self regulates the power supply.

400kHz bandwith is enough for audio amp if you got multi pole negative feedback.

difference between 400kHz with two pole output filter and 2 MHz with 2 pole filter (you need two pole filter to remove switching residual) is

400kHz rate gives 2 - 50 000 hz at 0,-3 db
2MHz rate fives 2 - 100 000hz at 0,-3 db

400kHz rate gives 2 - 20 000 hz at 0,-0.25 db
2MHz rate fives 2 - 20 000 hz at 0,-0.15 db

you cant hear 0.1db difference at 20kHz
but you can damn well hear 0.000X% THD vs 0.X THD

So stop with the you need 5MHz switching rate fallacy 400 khz is the highest you need for audio so its the best rate, if you go faster the PCB inductance will chokeout the high current pulse ramp (20 amps at 9 MHz harmonic for 400khz switching) you will get hundreds of mega herz of ringing on the squarewave that will pass thru the output filter like its not there, pulse will tilt so you will get high distortion, even 400 khz is too much for that currents but you need 400 for good sound.

Hope my explanation helps Cheers...

georgehifi's picture

That's why your left with this saw noise on the 10khz square wave.

Filtered out in the last year or so with Stereophile's ("Audio Precision’s AUX-0025 passive low-pass filter") hidden so you can't see it.

If it were a much higher switching frequency even more than the SE-R1 and filtered out totally this buzz saw noise at 10khz would not be there all, even without using Stereophile's AP filter.

Cheers George

dce22's picture

"If it were a much higher switching frequency even more than the SE-R1 and filtered out totally this buzz saw noise at 10khz would not be there all, even without using Stereophile's AP filter."

Not true! output filter is a 12 db down per octave it cant surpess the switching residual and when you go into the MHz range inductors go capcitive and capacitors go inductive filter will become high pass instead of low pass.

That is why when you switch at high frequency the PCB parasitics at high currents will choke the ramp rise of the FETs
(causing the squarewaves to miss the high harmonics creating ripple at the top or down off the square)
and it will generate 100MHz ringing on the squarewave that will go thru the filter and jam FM and VHF RF singals.

About the so called "buzz saw noise" if you use a high quality scope you will see that is not a saw but filtered squarewave that looks like a sinewave but with parabolic functions instead of sine functions.

Now this parabolic like sine wave is crucial to any self ocillating switching amplifier and it has to be prestine for the amp to work with low distortion like nCore amps,

because this type of amps does not have triangle wave generator to modulate the audio (the cleaner and stable the triangle the lower the distortion in old fashion class d very difficult thing to do and requiring hundrets of transistors)

So the big invention of the ucd, ncore amps is that the "output parabolic like sine" is fed into the input with the audio output of the amp as any negative feedback using the output filters phase shift in conjunction with negative feedback control circut that corrects for the filter audio distortions and as it goes to above 200khz it releases the phase shift correction and creates a oscillator at 500kHz that is free to move around from 470kHz to 530kHz as it is needed according to the music,
because it is connected to the input signal it has non of the jittery distorting problems that old school Triangle wave modulated Class D amp have.

So if you switch at megaherz frequency you will not remove it you will just lower it at kiloherz range and you will let megaherz signals blast thru that will RF radiate from the speakers cables.

Lets say you use balance current type of circut like Crown K1 or Mark Levinson 53 that switch into phases at 500kHz slow speeds than combine them into 4MHz switch rate.

And you supress switching residual into the noise floor you are stuck with a triangle PWM amp that have alot of problems.

ML no.53 is a triangle based amp that uses 1000 transistors, equivalent self oscillating class d like nCore uses 15 transistors.

You can compare Mark Levinson 4MHz monster of components to nCore 15 transistors and one filter a on a board.

nCore1200 with Audio Precision AUX-0025

nCore400 WITHOUT Audio Precision AUX-0025
drill down on the pdf to see 4ohm and 2ohm measurments.

nCore amp is better in any way especially the nc400 with two transistor matched output stage and discrete input opamp.

SNI's picture

Nice to read your writings which I find technical insightfull.
You are right about the switching frequency, no more than about 400 KHz is required. Most manufacturers use around 500KHz, and some times a bit more or less for one channel on a stereo board.
Raising the Fs would also create more heat thus loose efficiency, due to non lenearities in the output MOSFETs.
Direct FETs are a kind of solution to that problem though. They are smd types with no pins and very low induction.
Carefull pcb layout, very low impedance supplies and short signalpath can reduce the non linearities somewhat, but there is a window for optimum performance both on Fs and deadtime.
Going beyond optimum performace will only create problems, just as you wrote.
And then again it seems, that people sometimes get af wrong impression of the meaning of measurement signals.
A squarewave is ideally a signal that contains all harmonics from DC to infinite. If you put a signal into an amplifier that is bandlimited, you will not get a squarewave at the output.
What you see is stopband behavior.
A small group og enthusiasts think just the same about DAC´s.
NOS DACs are in fact a large hoax. There are no ringing on the signal. That´s only present above the stopband, and who cares?
Passband ringing is practically non existent, and who would sit and listen to squarewaves? Squarewaves is no part of a usefull audiosignal, and is only considered to expose stability problems.
This is very important to understand.
The AES17 filter normally used to measure DAD´s and class D amps, ensures, that no frequency above a usefull audio signal (20 - or 40 KHz) is passed through the amplifier, as this will not give meaningfull performance measurements.
As time has passed, class D has becom both better and more stable, which means that in some designs lowpassfiltering of the inputsignal is avoided or minimised. This is a good thing for soundquality, but to measured performance that would be bad, if carried out wrong.
The AES has described how to do this, and Audio Precission has created filters for their analyzers, to do meaningfull measurements.

Lorenzo-Italia's picture

it would be nice if Michal Jurewicz could better articulate below statement:

"Duntechs, with their multiple drivers and complicated crossovers, are a very difficult load," Mytek's chief designer, Michal Jurewicz, told me. "Hypex [class-D modules] cannot drive it, the amps collapse, but this Brooklyn Amp does it with ease.

considering that Mytek amplifier according to JA could not yet be measured at 2ohms


while Hypex Ncore implementation actually did it succesfully


Bel Canto

theta digital..

in addition Duntech, is a first order crossover speaker 3-4ohm with rather flat impedence and phase curve vs frequency.

(couldn't find stereophile review, pls consider John Dunlavi SC VI as reference design for Duntech)

waiting for mr. Jurewicz explanation...

My Best Regards

Lorenzo from Italy

Post Scriptum
Mr. B. Putzeys, could you pls let us know about Hypex (NCore) ability on difficult speaker loads?

georgehifi's picture

Yes true because if your filtering the same 12db per octave but at 3mHz (instead of 600kHz), by the time you get to 20khz it's all but gone, not like we have now!

Cheers George

dce22's picture

Self oscillating amp is a must to remove triangle wave problems and you cant do that without letting switching residual thru the filter (UCD nCore ICEPower amps are big phase shift oscillators) to feed it back into the input so it will oscillate,

and you dont need Mhz switching because class d is a analog amp that contracts and expands the squarewave according to the music signal no resolution loss.

This is not a good class d amp so you can see the ringing at the start of the squarewave, this ringing can be removed with carefull design at 500kHz frequency but at 4MHz it will become massive ripple that will span from 100MHz - 400 MHz creating a tv transmitter and output filter will not filter these frequency inductor will go capacitive and it will connect directly to the speakers cable that will transmit into the airwaves RF.

The switching residual does nothing to the speaker, the twitter can not feel anything at 500kHz 400milivolts, the residual is like a ultrasonic flute that is playing a note very quietly in the background without interfering with the music at all it does not intermodulate with the audio or distort it in any way and id does not radiate RF because the cables are too short for the wavelength.

You can make a decent 4Mhz amp like Mark Levinson did but it will cost a fortune (4 inductor 16 output transistors per channel 1000 transistors in the PWM) and it still perform worse in listening test and on the bench in noise and distortion DC-50Khz compared to nCore 15 transisors and one inductor


jgossman's picture

Especially as electricity gets more expensive and environmentalist creep into every aspect of our lives (not always a bad thing, sez this Republican)

All I hear in this review and discussion is blah blah blah, still not as good as a good mosfet amp.. Blah blah blah, buy a Pass design from Pass, Threshold, Nakamichi, etc. Blah blah bla, the homely DH-200/220 is still one of the best measuring (and pretty good sounding amps) ever and can be had for a song. Etc, etc. Good old power hungry Tetrode and Mosfet amps still just seem to be better. And we know how, at nearly every price point, to make them sound wonderful with many if not most speakers.

Also, all the talk of switching frequency all I hear is "what's the frequency, Kenneth?" - tape hiss included.

dumbo's picture

I'm continually baffled by audio equipment these days. Manufacturer says their AMP can drive difficult loads with ease but when simply placed on a test bench and these claims verified, the bottom drops out of their claim and they look foolish.

Do these people not have test equipment of their own that they use when designing a product? I wouldn't dream of attempting to design and build audio equipment for a living without test equipment.

Submitting a product for review to a magazine like Stereophile knowing that the product specs/claims will be verified on the test bench should be at the top of the manufacturers list of things to get right before submission.

I'll give Mytek some slack though since its product here is not priced in the stratosphere unlike some other products we've seen that cost 5 figures and leave the factory with critical flaws. Amazing Stuff!

Stereophile, please don't change a thing when it comes to your unique review approach by actually measuring this equipment that you receive. Its one of the main reasons why I am a loyal subscriber. Manufacturer claims need to be put in check so the consumer can make informed purchase decisions. Too bad none of the other rags have the same philosophy as Stereophile when it comes to reviews.

Buying a product based on a wink and a smile and only using your ears afterwards doesn't fly these days with the Internet at everyone's finger tips.

allhifi's picture

Brooklyn Class 'D':

Simpler, lighter, more efficient, and cheaper; the words thrown around.

Yet, Class 'D'(estructive) amplifiers I've seen thus far, are anything but high-value. (Gran-daddy of them all, the failed ML No.53.)

So, is "D", 'better' than 'A/B' -not a chance. Cheaper? No. Is is better built, or lasts longer -not a chance.

I'm sorry, who called for this? What need does this "technology" satisfy? Why does it even exist? Oh, Isee, is this one of those hidden-camera episodes/ mid-school pranks ! (lol)

If someone can answer that, that would be insightful -even delightful.


SNI's picture

I will give it a try.
First of all Pascal will never lie about performance of their modules.
They publish everything about the product in the datasheet.
You can see for yourselv here:
Så if there are any discrepansies, it will be due to either Brooklyns modifications, Brooklyns postulated data or Stereophiles measurements, which I cannot beleive.
Actually Pascal uses the same analysers for measurement, så this is probably not the case.

Class D amplifiers of modern design have certain upsides. One of them is, that theoretically they could have 0% distortion.
The only amplifier class that can match this is class A.
They do not have any intrinsic distortion.
Class A/B and B has cross over dist.
Noise is another parameter where class D is unbeatable. It is possible to reach as far as -160 dB SNR.
Distortion in modern designs can be very low, and by far beat the competition.
On top of that efficiency is much higher, and increasing with power, thus reducing the need for cooling.
This again will make the amp much less bulkier.
A normal ½ bridge design will need only two output devices pr. channel, never the less output currents in excess of 35A can be allowed.
The cool running amp has a much longer life expectancy than any competition, the cooler it runs the longer it will last.

So to comment directly on your arguments, then:

Lighter - yes
Cheaper - yes, but depends on makers price policy
Class D better than Class A/B - yes But the amount of crap is mostly constant regardless of amplifierclass.
Lasts longer - yes
Better built - not necesarily
Simpler - absolutely not.

Self oscillating Class D amps are not simple at all.
And even worse they are counterintuitive and hard to understand.
This is why so few succeeded in making them work properly.
Also a lot of ICEPower patents prohibited the use of certain important technologies for 20 years.
So there are only very few vendors to choose from.
ICE, Pascal, Hypex, Klank, Newclass D, and hardly any others.
Cheap chineese class D amps are often app notes from ie. IRF or other vendors of gatedrivers.
In the "High End Industry" I only know of Nuforce and Mark Levinson trying to make their own class D amps, and with Nuforce I´m not even sure.
These days ICE power has become a technology vendor more than hardware.
They have sold many millions of amps for i.e. mobile phones, computers, headsets and for everything where efficiency is important.
But they also make absolutely smashing amplifier modules like i.e. the 1200AS2.
The Pascal T1 actual for this review is from Pascal stated as a 280 Watt @4/8 Ohm amp with a power input @ 230VAC.
A little less 4 Ohm power @ 115VAC.
It is not a high fidelity module, but in reality Pascals least expensive one.
Pascal is oriented on the pro market, and as such they do not really aim at recordbreaking data. They produce cheap and healthy amps for their customers.
Hypex is somewhat different, they claim very stellar data, but when further investigated then the NC500 OEM module has a gain of only 16 dB, leaving you with a deficit of 8 dB, which you´ll have to get from somewhere else.
Component quality in Hypex modules are not impressing at all. Cheap chineese electrolytic capacitors with no name and reputation to protect, Samwha they are called. ICE is always more conservative in that regard.

So you´ll find good and bad things out there. But as principle the class D amp is superior to all but class A. And the components used for class D are actually being further develloped these days, for linear amps they are not, and haven´t been for decades.

The right approach towards class D should be the same as towards any other amp. There will be good amps, and there will be bad ones, and probably a lot inbetween.

allhifi's picture

SNI: Your reply, as class "D" amplification itself, is based on nothing but theory: Zero distortion, Zero Noise, No Heat/Cool running, Superior Life expectancy etc.

Yet, you have given absolutely NOT one makes/model as an example ?

" ...So to comment directly on your arguments, then:

Lighter - yes (NO)
Cheaper - yes, but depends on makers price policy (LOL)
Class D better than Class A/B - yes (LOL)
Lasts longer - yes (Gimme'e a break)
Better built - not necesarily (Finally, a truth)
Simpler - absolutely not. (= poorer performance/longevity)

Anyone can pick-up/Google "Amplification" and spew off what you have done here. Why bother?
What have you actually answered: In the world of reality, not theoretical electronics ?

Finally, I hope you realize we are speaking of world-class, high-resolution amplification -as often sought in audiophile sound systems- and not on jamming an amplifier module into a cell phone.


SNI's picture

This is simply not true.
0% distortion may not be possible, but class D has no intrinsic distortion what so ever.
Efficiency lies above 90% for the best class D amps, ie. Hypex NC500 and ICE Power 1200AS
Noise for the ICE Power 1200AS is -160 dB or 40 µV
NC500 has 10µV noise floor.
Efiency above 40% for class A/B is rarely seen, but theory says max 66,6%.
Eficiency for class A push pull is theoretically 50%, but 20% i a realistic goal.
For class D eficiency is theoretically 100%, a bit more then 90 is seen now and then.
Life expectancy will be limited only by electrolytic capacitors life expectancy, and as the supplies for class D mostly is switch mode, then there is hardly any ripple to find, thus they live much longer.

By experience the good ones never fail.
It is pretty easy to make very fast digital protection without relays.
Så done right they never run hot, they tolerate short circuit, closes down if DC is detected, and the shut down at clipping viltage.
And most of the good ones performs better with power.
Have a look at the datasheets for NC500 and 1200AS.

Failures might be different for Hypex, I do not know, but ICE has a very good record, which probably is one of the reasons they sold hundreds of millions of ICE Power channels for active speaker, mobile phones, computers etc.
They really last and are very eficient.

For my selv I own af few class D amps, amongst them are those mentioned, but actually I now have one of our own breed.
We are a small group of boys doing stuff like that.
And our own amp actually has a noise level of -160 dB, redicuolosly low distortion, eficiency around 90% and best of all completely linear delay through the audioband. It will be pretty hard to find a linear amp with phase @ 0 degrees through the entire audioband.

Measured performance for class D is pretty much unbeatable.
But one also has to understand, that they really work in a different way, and the important things to pay attention to is completely different in class D amps.
There is in fact no reason for big and bulky powersupplies, because a good selfoscillating class D amp has very high PSRR.
The performance will not degrade gradually with power as it normally does with linear amps.

I-e. the ICE 1200AS delivers 90 seconds of full power with modest passive cooling, and that is 1.250 Wats @ 1% THD @ 4 Ohms.
And that is doon on a board 4X8 inches 3.5 inches high including powersupply.

allhifi's picture

SNI: You reference:

" class D amps, ie. Hypex NC500 and ICE Power 1200AS"

Are these actual amplifier names/and model, or simply the "D amp-modules'?

And again, great talk/theory, but where are some (repectable) reviewer/ user comments that tell us your amps (Class 'D') compete with (let alone trounce) premium A/B's from Levinson, Classe -and others ?

Give us a make/model of your beloved class "D" amplifiers ?

I for one would love to have/own/enjoy a low-cost, light-weight, bullet-proof (high efficinecy/low heat) main power amplflier to replace conventional A/B's.
Please, offer up, and back up the talk, first.


SNI's picture

You´ll find Hypex nCore in Bel Canto e.One Ref 600M, it is reviewed on this page.
PS Audio M700 is based on ICE Power 700ASC modules, also this amp is reviewed on this site.
They are not 1200AS amps, but I do not know of reviews of the 1200AS series module anywhere by now. They are pretty new to the market, and pretty expensive to, which might have some impact on its popularity.
Jeff Rowland did earlier use ICE Power, but I think they have changed that to Pascal.
Gato Dia 400 amplifier is also Pascal.

The Levinson approach was absolutely embarrasing, it was old if not very old technology, and a complete lack of understanding and exploring the benefits of class D.

And of course I will mention our own upcoming project, but that is not yet ready for anyone to have.
I hope this will enlighten you, and make you look at class D with an unbiased attitude, because the best of these amps deserve that.