ATI AT522NC Stereo Amplifier Review

[amirm]

Yeh, definitely not smart. Another complication is that it has to have settling time to properly measure THD+N since it varies so much in real time. That causes the measurements to lengthen and keep the amplifier at peak power which makes me uncomfortable. This also practically limits how many values it can try for each frequency point.

 

[LTig]

Let's not forget this chart. Who here is under 20 and can hear up to 16k? And who here is older and can only hear till 8K?

I fear I might belong to the second group.
OTOH it spares me a lot of money (no Berillium tweeters), and the fun listening to music has not deteriorated yet.

 

[March Audio]

We may want to move this to another thread as its off topic of the amp on test. @amirm @Thomas savage ?

Two things to consider here, a filter is needed to ensure accuracy of the measurement system. The second is to ask what is important to subjective quality?

In band ( nominally 20Hz to 20kHz) noise and distortion, or out of band ( >20kHz) noise and distortion? Its the former, stuff we can actually hear.

Firstly its a myth that Class d has high ultrasonic distortion. I answered this question in another forum the other day. I performed some quick and dirty measurements to demonstrate. There is no filter in use.

1KHz reference

19+20kHz IMD

30+31kHz

40+41kHz

I used IMD tests because I wanted to demonstrate the in band distortion created by ultrasonic signals input into the amp - the stuff you can actually hear. Also note that the signals used are waaaaayyyyy higher than any level you will find in real music. Below is a plot at levels you might genuinely find at 30kHz in a high res recording.

However there is obviously the question of harmonic distortion which is not accounted for in the above measurements, say a 20kHz tone with harmonics at therefore 40KHz, 60Khz etc. I will take some measurements later to show this, but its not high.


As mentioned, music simply does not have high frequency content at high levels.

This is a 96KHz Gary Burton track. This is typical and shows by 20kHz the levels are -60dB.

So if an amplifiers distortion was really bad and it created harmonic distortion at another -60dB (probably would be more like -100dB or more), then you can see these harmonics are going to be in the noise floor at -120dB.

So your tweeter response is quite irrelevant. Also there are no tweeters that respond at the switching frequency, so they simply cant create any audible in band IMD.


So you need to focus on whats important. Yes we need to measure whats going on at ultrasonic frequencies, absolutely the case, but these same myths and concerns about class d keep being raised.

Another member asked if I could write a Class D FAQ, so I will probably do so soon with lots of measurements to show whats really going on.

 

[Rusty Shackleford]

Adding square waves to amp measurements, like Stereophile, would be appreciated.

 

[restorer-john]

At 20 kHz though, distortion shoots way up and we are able to only produce 22 watts before THD+N of 0.1% is hit. Fortunately there is not much music energy there so practical audible concern is low.

This is a joke surely? Please tell me it is.

The amplifier cannot produce its rated power (375W@4R) at 20KHz? It can only produce 22W@4R /0.1% or 5.86% of its rated power?

No free passes for this amplifier and the recommendation is unwarranted in my opinion.

 

[amirm]

Adding square waves to amp measurements, like Stereophile, would be appreciated.

What power level and what frequency? And band limited as it is in my analyzer or ultra high slew rate?

 

[restorer-john]

In band ( nominally 20Hz to 20kHz) noise and distortion, or out of band ( >20kHz) noise and distortion?

Firstly its a myth that Class d has high ultrasonic distortion (say between 20KHz and 100kHz).

Can you run a full rated power 20KHz sine and plot the spectrum for us?

 

[amirm]

This is a joke surely? Please tell me it is.

The amplifier cannot produce its rated power (375W@4R) at 20KHz? It can only produce 22W@4R /0.1% or 5.86% of its rated power?

No free passes for this amplifier and the recommendation is unwarranted in my opinion.

Not quite. There is noise shaping which may be contributing to high frequency THD+N measurements as seen in the FFTs. Until I get more mileage on this test, I don't want to run with it too fast.

 

[restorer-john]

I was able to run my tough test of THD+N at 0.1% versus frequencies. Some amplifiers have died trying to run this test. Others are had to poinpoint as their distortion jumps all over the place:

Why not forget the sweep and just do spot frequencies? Less stressful on these poor little amplifiers, the time taken would be much less and they won't go bang on your bench.

Say 40Hz, 100Hz, 10KHz and 20KHz?

 

[restorer-john]

Not quite. There is noise shaping which may be contributing to high frequency THD+N measurements as seen in the FFTs. Until I get more mileage on this test, I don't want to run with it too fast.

Go old skool like me. Load the sucker, feed it 20KHz, wind up the level until it clips, back it off a touch and call it. Yes, get out your DSO/Cro.

 

[RichB]

They measure better than the majority of amplifiers out there, are very cheap for their state of the art in band performance, small, light, powerful, and are available from multiple vendors in various configurations and formats. I call this a good fashion! Customers win.

Issues are always raised by armchair analysts, while people who actually own and use them are satisfied. I am on my 3rd model of class D power amplifier having owned 3 different class AB integrateds before.

These are great advancements in amplifier design. I own an AT522NC and AHB2. The AT522NC is more powerful and the AHB2 is lighter and produces less heat in normal operation. After careful voltage level matching, I found the AT522NC to be brighter and prefer the AHB2 that is more natural to my ears.

Also, the AHB2 seems immune to power and other environmental noise that plagued toroid based PS in amps that I have owned from Sunfire, Outlaw, ATI, and Parasound (though Parasound had the least hum). Unlike the others, the AHB2 is absolutely quiet, both mechanically and sonically.

- Rich

 

[Audiocrusader]

The amp module manufacturers data sheet already has a more comprehensive set of measurements than provided here used with their SMPS1200A700 supply, minus an input buffer:

NC500 OEM Datasheet

Professionals with deep knowledge of amplifiers chose the best set of measurements to use to show off the merits of the modules.

The same suite of measurements they use in the datasheet should be used for all amps using this module. Then we would know for certain that if there's any shortcomings its due to the OEM's implementation.

 

[audimus]

Not quite. There is noise shaping which may be contributing to high frequency THD+N measurements as seen in the FFTs. Until I get more mileage on this test, I don't want to run with it too fast.

What is noise shaping?

Professionals with deep knowledge of amplifiers chose the best set of measurements to use to show off the merits of the modules.

The same suite of measurements they use in the datasheet should be used for all amps using this module. Then we would know for certain that if there's any shortcomings its due to the OEM's implementation.

What about knowing the shortcomings of the amp design itself that professionals with deep knowledge of their amplifiers choose to leave out because it shows the demerits of the modules they are vested in?

Or are you saying all those tests in addition to the others that can be chosen to objectively test amplifiers across all designs?

 

[audimus]

Also, the AHB2 seems immune to power and other environmental noise that plagued toroid based PS in amps that I have owned from Sunfire, Outlaw, ATI, and Parasound (though Parasound had the least hum).

You heard a hum in ALL of those amps? Seriously? Or are you talking about seeing them in a measurement?

Perhaps there is a market for power line conditioners after all to remove any DC component in your lines.

 

[amirm]

What is noise shaping?

This:

Or are you asking in general what noise shaping is?

 

[Audiocrusader]

What is noise shaping?



What about knowing the shortcomings of the amp design itself that professionals with deep knowledge of their amplifiers choose to leave out because it shows the demerits of the modules they are vested in?

Or are you saying all those tests in addition to the others that can be chosen to objectively test amplifiers across all designs?

It would be meaningful data for people to know the performance of the module when using with the matching supply. You will never get better performance from the module than the results in their datasheet. So there should be a reference established for all class d amps using a particular popular module. If the Hypex datasheet can't be trusted, the exact same tests they run should be ran independently for verification. And in order to do this the exact same test setup would be required. For the NC-500's, the OEM evaluation input boards can be used with the buffer bypass jumpers soldered. And the SMPS1200A700 should be used to power the modules. After the testing is complete, the datasheet results and independent results can be compared. Now we will know if Hypex is a bunch of lying hacks, or we can take the datasheet measurements from them at face value. Once this is established, try whatever other tests you want. But keep the testing standardized for all amps. Rather than cherry picking different tests for each amp.

If this is done, now we have a solid reference point. And all other amps using this module should be directly compared to this reference.

It should be clarified that the Benchmark SINAD is pretty much cheating. Because they are buying a bunch of points by having a low gain mode. Bypassing the buffers on any of these amp types will achieve the same thing. And anyone who owns one can easily do it in 30 seconds.


And let's also remember audio amplifiers are made for human ears to enjoy. Not dogs, cats and bats. And only the very best under 20 human ears can hear up to 16K. Being a self proclaimed audiophile does not make your hearing any better than everyone else. If anything it's a sign you may have suffered even more hearing damage than the average.

 

[RichB]

You heard a hum in ALL of those amps? Seriously? Or are you talking about seeing them in a measurement?

Perhaps there is a market for power line conditioners after all to remove any DC component in your lines.

I am referring to mechanical hum from the transformers that was audible from my seated position in a quiet room.

I don't believe there is an established solution for blocking DC.
I had a Panamax MX500 and not only did it not remove the hum it seemed to limit amplifier performance.
I was not about to buy an amplifiers (power regenerators) for my amplifiers.

The Benchmark engineers (and others) have found the best way to combat transformer hum is to eliminate the source

- Rich

 

[restorer-john]

Professionals with deep knowledge of amplifiers chose the best set of measurements to use to show off the merits of the modules.

Exactly. They are in the business of making their amplifiers look as good as they can. There's no way they are going to publish any numbers that paint their product in a poor light. Hence, the huge number of disclaimers, annotations and little numbers that lead to 'fine print'.

The same suite of measurements they use in the datasheet should be used for all amps using this module. Then we would know for certain that if there's any shortcomings its due to the OEM's implementation.

No way. Test the modules in accordance with established standards and report the results without fear or favor. Hypex have 'created' their own 'standards' and are holding them up as gospel. Headline advertised power- 700W@4R. Fine print continuous power "typically one fifth" but then they even derate that to 100W@4R.

Power bandwidth- 35KHz. Fine print power bandwidth testing limited to 5Khz @100W because the output caps will likely go up in smoke.

It should be clarified that the Benchmark SINAD is pretty much cheating. Because they are buying a bunch of points by having a low gain mode

We all know that- several of us screamed it in the Benchmark review. But then again Hypex are the kings of low impedance, low gain, driven by a stupid low impedance source for their numbers aren't they? Hardly innocent eh?

What's good for the goose is also good for the gander it would seem, in this dodgy post-standards world.

 

[amirm]

What's good for the goose is also good for the gander it would seem, in this dodgy post-standards world.

Wait a sec. There is not a company in the world providing the level of documentation Hypex is. They come from chip companies so have documented their amplifier module as such. They did not just throw some random numbers out there.

What Hypex has done is to turn class D design into a mathematical problem that is solved first, then implemented.

For these two things, they deserve a lot more praise than scolding.

 

[Audiocrusader]

Exactly. They are in the business of making their amplifiers look as good as they can. There's no way they are going to publish any numbers that paint their product in a poor light. Hence, the huge number of disclaimers, annotations and little numbers that lead to 'fine print'.



No way. Test the modules in accordance with established standards and report the results without fear or favor. Hypex have 'created' their own 'standards' and are holding them up as gospel. Headline advertised power- 700W@4R. Fine print continuous power "typically one fifth" but then they even derate that to 100W@4R.

Power bandwidth- 35KHz. Fine print power bandwidth testing limited to 5Khz @100W because the output caps will likely go up in smoke.

View attachment 39607



We all know that- several of us screamed it in the Benchmark review. But then again Hypex are the kings of low impedance, low gain, driven by a stupid low impedance source for their numbers aren't they? Hardly innocent eh?

What's good for the goose is also good for the gander it would seem, in this dodgy post-standards world.

Hypex does not make any claims other than what is claimed in their data sheets. No different than an IC opamp, DAC chip or any IC for that matter. They provide a large suite of measurements that are more than enough to satisfy their potential OEM clients. If not, they wouldn't be so successful.

You still don't seem to realize that other than the NC-400, Hypex sells OEM subcomponents for professional amp builders to use as a subsection of their amplifier products. These are not turnkey, ready to go amplifiers. When you read the datasheet of a DAC chip, you don't see the exact same specs that will be seen in every DAC on the market that uses the chip. It's up to the manufacturer who incorporates the module into their design to release whatever kind of test data they want. The reason the input impedance and gain is so low is because it's not a finished product! These modules are designed this way so the OEM can make their own input buffers and voice the amplifiers to whatever taste they want, or try to make them as neutral as they want. This is how OEM's differentiate their products from their competitors. If any OEM finds the Hypex datasheet is insufficient for their needs, they can run whatever battery of tests they want during the evaluation period of the modules. And whatever data the OEM decides to share with their clients is all based on the marketing strategy they choose to go with.

If you look at big companies like Harman, with their flagship Mark Levinson lineup they choose to release very little measurement data on the products. And Harman is worshipped as the pinnacle of audio science around here. But nobody ever talks about that.