Why is EVERY amp on this polish website at low signal output 20 dB worse than @amirm´s measurements

[totti1965]

A Guy from Poland who is running a yearly HiFi show in Warschau does pretty much the same thing as Amir does:
Measuring amps (some hundreds of them!)
What puzzles me, is that all his measurements with small signal (look at 100 milliwatt) are showing about 20 dB worse results as @amirm!!!
E.g. look at the measurements of the NAD C 316 BEE which is as many, many amps measured by both of them!

NAD C 316 BEE (wzmacniacz zintegrowany)

Obserwuję urządzenia NAD od ponad dwóch dekad i cieszę się, że niemal każdy ze wzmacniaczy stereo, w tym C 316BEE, wciąż ma w sobie ...

audio.com.pl

100 milliwatts starting with about minus 50 dB or so (0,3%)
But @amirm ´s measurements are starting with about minus 70 dB or so ….

NAD C 316BEE V2 Integrated Amplifier Review

This is a review and detailed measurements of the NAD C 316BEE V2 integrated amplifier with phono stage and headphone jack. It is on kind loan from a member and costs US $399. The unit feels just heavy enough to not feel cheap. Controls have a nice feel and can be managed with a remote...

www.audiosciencereview.com

It is a systematically difference which turns up in EVERY measurement when you compare audiosciencereview results of amps with this interesting polish website!

Wzmacniacze stereo - testy, ceny i sklepy | AUDIO

Wzmacniacze stereo - zaawansowane i wiarygodne testy napisane przez profesjonalistów. Porównania i porady. Przeczytaj, zanim kupisz!

audio.com.pl

Nearly every amp starts about minus 50 dB THD + noise at the polish site (0.3% THD + noise) when measured at 100 milliwatts …. And then decreases not so good as on measurements at the audiosciencereview website…


Some differences between two audio websites are ok, but 20 dB…. That is factor 10!

I just want to understand what technical reason is going on here…..

 

[DVDdoug]

when measured at 100 milliwatts …

I believe Amir measures power amps at (or relative to) 5W. A higher signal makes a better signal-to-noise ratio! Distortion MIGHT also get better up to the point of clipping.

 

[staticV3]

I just want to understand what technical reason is going on here…..

He doesn't use the same test equipment as Amir and doesn't use the same test parameters as Amir.

A worse audio analyzer will will have more inherent noise and so will higher measurement bandwidth or lower FFT count.

Unless the hardware and software parameters are matched, the resulting measurements cannot be directly compared.

 

[totti1965]

He doesn't use the same test equipment as Amir and doesn't use the same test parameters as Amir.

A worse audio analyzer will will have more inherent noise and so will higher measurement bandwidth or lower FFT count.

Unless the hardware and software parameters are matched, the resulting measurements cannot be directly compared.

Are not both of them sweeping a 1 kHz tone from 50 (@amirm) or 100 milliwatts (the polish guy) to full power and then plotting the SINAD?

 

[totti1965]

I am just puzzled because the difference is so huge!

 

[staticV3]

I am just puzzled because the difference is so huge!

That's par for the course with audio measurements.

 

[Timmy]

Unless the hardware and software parameters are matched, the resulting measurements cannot be directly compared.

Shouldn’t that be reason enough to define measuring standards which would lead to results that very well can be compared? Actually, I would be surprised to learn that those standards do not exist already.

 

[Mnyb]

Also look up FFT gain if you look at levels between the spikes in some FFT this can be a bit misleading vs the actual noise levels.

 

[Bamboszek]

Shouldn’t that be reason enough to define measuring standards which would lead to results that very well can be compared? Actually, I would be surprised to learn that those standards do not exist already.

Lots of organizations tried to introduce some kind of standardization in audio measurements - FTC, IEC, DIN, CEA to name few.
The problem is that those norms are often contrary and manufacturers rarely follow any.
Input level, output level, channels driven, load impedance, bandwidth, weighting and so on affect results.

"How to write (and read) audio specifications" paper by Audio Precision is a good introduction to this topic.
It's used to be available on AP website, now I see it can be accessed here - https://sengpielaudio.com/Mathew-AudioPrecisionSpecifications.pdf

 

[totti1965]

Ah, and by the way the guy from Poland is forum member @elberoth:
@amirm knows and likes him:

Musical Fidelity A1 mk3 and B200 vintage amps measurements

I have just measured those two. The MF A1 has a cult following, and was designed by Tim de Paravicini Here are the details: 1. Musical Fidelity A1 mk 3 (1994; 20W Class A amp according to the manufacturer), fully restored with modern parts: Rated power (1% THD + N, 1 kHz) [W] 1 channel...

www.audiosciencereview.com

This is what makes me looking to his cool polish site!

Next week is his next audioshow in Warsaw (Warschau), Poland:

Dla odwiedzających - Bilety - Audio Video Show: 25–27 października 2024

Audio Video Show jest największą w Polsce wystawą sprzętu audio-video oraz kina domowego.

audioshow.pl

 

[NTK]

There are also large discrepancies between the measurements by audio.com.pl and SoundStage! Take the example of the NAD M10V2. Here are the THD+N vs power measurements from audio.com.pl

NAD M10 V2

Trzy lata po premierze wzmacniacza M10 pojawia się jego następca. Identyczny symbol, uzupełniony tylko przypisem V2, sugeruje, że nie jest ...

audio.com.pl

Here are the SoundStage! measurements. The SoundStage! THD+N levels were also significantly lower. My guess is that the measurements by audio.com.pl could be limited by their equipment and/or setup.

SoundStageNetwork.com | SoundStage.com - NAD Masters M10 V2 Integrated Amplifier-DAC

www.soundstagenetwork.com

 

[Timmy]

look at the measurements of the NAD C 316 BEE

A humble attempt to visually multiply the two (correct?) measurement charts could look like this:

 

[totti1965]

A humble attempt to visually multiply the two (correct?) measurement charts could look like this:

View attachment 399572

Yes but you have to make the polish measurements even a bit smaller!
1% is exactly minus 40 dB
0.1% is exactly minus 60 dB
0.01% is exactly minus 80 dB

But after you have done it, you see how large is the difference between the polish measurements and any other I have seen so far (stereophile, ASR, L7, Erin’s audiocorner, all those Chinese manufacturers….)

I think, it is mostly because @elberoth forgotten to lower the Bandwith a bit. So it is not measured with a Bandwith of 22 kHz or 45 kHz or even 88 kHz or so but perhaps with the full 1.2 MHz, the audioprecision is capable of….
So you measure a lot of things, nobody can hear (Ultrasonic Noise).

 

[Timmy]

Yes but you have to make the polish measurements even a bit smaller!
1% is exactly minus 40 dB
0.1% is exactly minus 60 dB
0.01% is exactly minus 80 dB

That’s »Bahnhof« to me, but I try to follow. The polish chart begins to blur now, I fear

 

[totti1965]

That’s »Bahnhof« to me, but I try to follow. The polish chart begins to blur now, I fear

View attachment 399579

Oh! Perfect! Thank you! This is amazing! Are you using pages from apple to do this?
Nice comparison of the two measurements.
And by the way: the difference seems to be more something like 30 dB at low volumes. That is a factor of 30!
With higher volumes the difference starts to getting smaller, but even 20 dB (factor 10).

 

[amirm]

That kind of curve tends to indicate non-linearity in the dummy load. You can tell this when it happens at higher powers where complex thermal effects cause modulations.

 

[Timmy]

the difference seems to be more something like 30 dB at low volumes. That is a factor of 1.000!

Please don’t consider my pics to be any »evidence«. I’m not even sure that I took the right screenshots (that’s what it is) for this to compare

 

[totti1965]

Please don’t consider my pics to be any »evidence«. I’m not even sure that I took the right screenshots (that’s what it is) for this to compare

I checked it. These were the right screenshots! Which program did you use to match the two screenshots? I must learn it to do this by myself!

 

[totti1965]

That kind of curve tends to indicate non-linearity in the dummy load. You can tell this when it happens at higher powers where complex thermal effects cause modulations.

Fun fact is, that even these plots are looking as some melted stuff!

 

[Bamboszek]

Yes but you have to make the polish measurements even a bit smaller!
1% is exactly minus 40 dB
0.1% is exactly minus 60 dB
0.01% is exactly minus 80 dB

But after you have done it, you see how large is the difference between the polish measurements and any other I have seen so far (stereophile, ASR, L7, Erin’s audiocorner, all those Chinese manufacturers….)

I think, it is mostly because @elberoth forgotten to lower the Bandwith a bit. So it is not measured with a Bandwith of 22 kHz or 45 kHz or even 88 kHz or so but perhaps with the full 1.2 MHz, the audioprecision is capable of….
So you measure a lot of things, nobody can hear (Ultrasonic Noise).

According to his post, measurements are done with Neutrik A2-D analyzer. Quite dated to today's standards.

I checked it. These were the right screenshots! Which program did you use to match the two screenshots? I must learn it to do this by myself!

Acquiring data with WebPlotDigitizer and then comparing them directly on the new plot should be probably even more readable.