Amplifier comparison: March Audio P452 (Purifi 1ET400A) vs Buckeye Hypex NC502MP vs AIYIMA A07 (TPA3255) vs Onkyo TX-NR737 (125W Class AB)

[f1shb0n3]

I'll use this thread as my personal blog in The Quest to find The Answer to The Question about Amplifiers. My subjective one, informed by the best science known to me. Both the question and the answer are to be figured out, my first revision of the question is: "What are the audible differences between different power amplifiers?"
I have a few amps laying around and using my near-field to mid-field desk setup of KEF R3, I'm equipped with all gadgets and cables to make a proper comparison of amplifiers using my own ears.

Disclaimer: I'm not an electrical engineer, not even an experienced audio enthusiast. I'm doing this test for my own understanding and education and to compare the gear I have at hand. I'm also not picking favorites here - each of these amps work great for its intended purpose and is a part of an audio system in my home.

Products involved

• March Audio P452 Class D amplifier using Purifi 1ET400A chip with Alan's custom input buffer. Amir's review of a comparable model by Nord, I expect P452 to measure not too far from it.

Spoiler: Measurements

• Buckeye NC502MP Class D amplifier using all-in-one mains-powered Hypex module. Amir's review, teardown.

Spoiler: Measurements

• AIYIMA A07 Class D amplifier using Texas Instruments TPA3255 chip with 32V 5A power supply. Amir's review.

Spoiler: Measurements

• Onkyo TX-NR737 AVR with 125W Class AB amplifiers. No measurements are available to my knowledge, looking for comparable ones for reference.

Diagram of the setup

Picture

Status so far
Volume matching is a serious challenge, still working to confirm I'm within 0.1dB.
Heard distinct differences between Purifi and Hypex and between Purifi and Onkyo AVR so far with my best attempts in volume matching.
Details on specifics comparisons and exact settings will follow.

Updates from the thread below

• Level-matched Purifi vs Onkyo with oscilloscope, not hearing differences afterwards, continuing to listen.

 

[maty]

Please, with very good recordings, DR > 11.

And acoustic and / or electric instruments. You know, jazz, folk, orchestral...

With today's recordings of popular music all computer manipulated the audible differences should be much smaller, I say.

 

[restorer-john]

So you have your speakers on a chinese restaurant lazy susan? Awesome. I've got one I use for heavy amplifier repairs on my lab bench.

I applaud your efforts, but attempting to compare amplifiers with that dodgy switcher (pressing both switches at the same instant for both channels) is doomed to failure.

You need to electrically balance the levels. Why even bother with the mini-dsp at all for comparison purposes- it will be a bottleneck you don't need?

 

[f1shb0n3]

So you have your speakers on a chinese restaurant lazy susan? Awesome. I've got one I use for heavy amplifier repairs on my lab bench.

Allows for faster comparison between speakers in the exact same position, audio memory being short and all. Works out well, only had to use replace with heavy duty rotation hardware. Eventually motorized and remote controlled will be perfect

I applaud your efforts, but attempting to compare amplifiers with that dodgy switcher (pressing both switches at the same instant for both channels) is doomed to failure.

Indeed when switching two switches simultaneously I get short breaks in both channels that were a bit confusing in the beginning, but was easy to learn to ignore. I'd appreciate suggestions for a better switch - at Amazon many had same ground connected to all black terminals and could not find a single-button one that would switch 2 channels simultaneously without common ground.

You need to electrically balance the levels. Why even bother with the mini-dsp at all for comparison purposes- it will be a bottleneck you don't need?

I tried measuring voltage at amplifier with 1kHz square wave for Purifi and Hypex and found that it's not representative of the perceived volume. Couldn't differences in output impedance between amplifiers affect perceived volume at same voltage? Still unsure of the "proper way" to volume match when comparing amplifiers.

MiniDSP SHD is great tool for volume matching in my use case - with 2 DAC outputs, separately digitally attenuate-able in 0.1dB increments. PEQ can be applied separately to each too which will be very useful to try to compensate for the 0.5dB dip at 12-20kHz in the frequency response linearity of Hypex when comparing vs Purifi. I think this one might be audible and curious if it's "correctable".
I also use it for Dirac Live room correction - it improves my listening experience in with all speakers in all settings. When comparing amplifiers will be doing it with and without Dirac enabled. I'll eventually integrate a pair of stereo subs with the MiniDSP, after I'm done with all fun comparisons I plan to do.

 

[restorer-john]

I tried measuring voltage at amplifier with 1kHz square wave for Purifi and Hypex and found that it's not representative of the perceived volume.

Use a sine wave at 100Hz with your DMM. (assuming you are using a DMM and many are only accurate with sines at low frequencies)

 

[raindance]

125 watt class A amplifiers - what, does the Onkyo weigh 100lbs?

 

[restorer-john]

125 watt class A amplifiers - what, does the Onkyo weigh 100lbs?

He clearly means Class AB.

It weighs 11.5kg according to the specs.

 

[abdo123]

Use a sine wave at 100Hz with your DMM. (assuming you are using a DMM and many are only accurate with sines at low frequencies)

@f1shb0n3 or even better, a cheap oscilloscope.

56.04US $ 53% OFF|FNIRSI 1C15 Professional Digital Oscilloscope 500MS/s Sampling Rate 110MHz Analog Bandwidth Support Waveform Storage|Oscilloscopes| - AliExpress

Smarter Shopping, Better Living! Aliexpress.com

www.aliexpress.com

 

[f1shb0n3]

@f1shb0n3 or even better, a cheap oscilloscope.

56.04US $ 53% OFF|FNIRSI 1C15 Professional Digital Oscilloscope 500MS/s Sampling Rate 110MHz Analog Bandwidth Support Waveform Storage|Oscilloscopes| - AliExpress

Smarter Shopping, Better Living! Aliexpress.com

www.aliexpress.com

I want one, happy to be needing it Can you recommend a reasonable more expensive one from Amazon? I’m confused when I see the variety of different models with features I don’t understand if I’ll ever need..

 

[tomtoo]

Use a sine wave at 100Hz with your DMM. (assuming you are using a DMM and many are only accurate with sines at low frequencies)

Why it should be accurate for just comparing? It will give a wrong number, but if both voltages the same it will give the same wrong number for both. Or do you see this different?

 

[restorer-john]

Why it should be accurate for just comparing? It will give a wrong number, but if both voltages the same it will give the same wrong number for both. Or do you see this different?

He was advocating using a square wave at 1kHz.

 

[abdo123]

I want one, happy to be needing it Can you recommend a reasonable more expensive one from Amazon? I’m confused when I see the variety of different models with features I don’t understand if I’ll ever need..

The one i linked is fairly good for its price and should be more than good enough for this purpose (or anything really in the audible band). it's also small and portable.

 

[tomtoo]

He was advocating using a square wave at 1kHz.

Thats not what i would do. But i not see how it would change the situation. He is not looking for the absolute number, just if both voltages are the same.

Its like having a ruler thats wrong made and shows instead 1m 1.20m.

Useless for a absolute number but perfectly usable to see if two objects have the same size.

 

[boXem]

Thats not what i would do. But i not see how it would change the situation. He is not looking for the absolute number, just if both voltages are the same.

Its like having a ruler thats wrong made and shows instead 1m 1.20m.

Useless for a absolute number but perfectly usable to see if two objects have the same size.

The square wave is a sum of sines. So the frequency response of the amplifier will have an effect on the voltage read by the DMM.
1 kHz sine via oscilloscope or 100 Hz sine via DMM are more reliable.

 

[PierreV]

A bit of lateral thinking might help here

If you pass a blind test between amplifiers, there will be an endless debate about the proper level matching techniques you did not use or executed poorly that will never be settled.

I would suggest matching levels by ear until you reliably fail the blind test.

 

[Blumlein 88]

Use a sine wave at 100Hz with your DMM. (assuming you are using a DMM and many are only accurate with sines at low frequencies)

Could be his meter is reacting slightly to the switching frequencies of those class D amps. Might need a simple filter on the meter.

Oops didn't notice the square wave part. Yeah use a sine wave.

 

[tomtoo]

The square wave is a sum of sines. So the frequency response of the amplifier will have an effect on the voltage read by the DMM.
1 kHz sine via oscilloscope or 100 Hz sine via DMM are more reliable.

Yes the FR of the amp will also have a impact on overall loudness.
But this was not my point.
My point was its not importend to use a highly accurate rms meter. As long you just compare two levels. It will show shit, but the same shit if the input level is the same.

 

[f1shb0n3]

Yes the FR of the amp will also have a impact on overall loudness.
But this was not my point.
My point was its not importend to use a highly accurate rms meter. As long you just compare two levels. It will show shit, but the same shit if the input level is the same.

Related question for the electrical engineers here - could different amps measuring same voltage on a competent meter/oscilloscope with 1kHz or 100Hz sine wave, produce different volume due to their output impedance differences or some other factor?

 

[boXem]

Related question for the electrical engineers here - could different amps measuring same voltage on a competent meter/oscilloscope with 1kHz or 100Hz sine wave, produce different volume due to their output impedance differences or some other factor?

If the matching is done with speakers plugged (it should), no difference at the matching frequency. But since the output impedance is fequency dependant, it could happen at different frequencies.

 

[f1shb0n3]

If the matching is done with speakers plugged (it should), no difference at the matching frequency. But since the output impedance is fequency dependant, it could happen at different frequencies.

An eye-opener point for me - never thought I have to measure with speakers plugged-in, I was measuring at amplifier +/- with multimeter. There are so many ways to get stuff wrong..
Thanks!