Benchmark AHB2 / Class D Purifi Eigentakt / Mark Levinson 333 - Listening impression & Conclusion

[Geert]

To address the blind listening issue, apparently it was all sighted listening.

Please explain:

Most of the test was blind. Especially when we want to confirm the differences on what we heard.

 

[Matias]

I think he meant ABX, that is, first listening to the amplifiers sighted to learn how each amplifier sounds, and then blinded to try to recognize those difference without knowing which one is playing.

 

[MasterApex]

I think he meant ABX, that is, first listening to the amplifiers sighted to learn how each amplifier sounds, and then blinded to try to recognize those difference without knowing which one is playing.

Correct...that is what we did.
Hard to tell AHB2 vs 333, very easy to identify Purifi on songs G-I.

 

[Matias]

Correct...that is what we did.
Hard to tell AHB2 vs 333, very easy to identify Purifi on songs G-I.

Join me in waiting for the new high power Purifi modules that Bruno is creating. They still did not finish the amp, so no specs or release dates, but it is ongoing. Hopefully one day you can repeat this test with a higher power Purifi amp and see how it goes.

 

[rdenney]

The more experience I have (and my audio interest already goes back to the 70's), the more persuaded I am that more power is better, even at moderate listening volumes, to make sure those sharp peaks don't find compression, soft clipping, or even hard clipping. I've just had too many experiences where bigger amps of modest performance sound "better" than supposedly better amps of lower power. Level-matching at 1KHz is a steady-state match--it doesn't say anything about whether levels will be matched during the peaks.

If we level-match at 87 dB SPL (which is loud) at, for typical speakers, 2.83 volts (1 watt into 8 ohms), peaks might be as much as 106 or 108 dB SPL for recordings with high dynamics, but even those might be measured over a long enough period to gloss over the true peak. Those peaks would require 128 watts, and that's into 8 ohms. I think of amps in that power category as the minimum for playing really dynamic music loudly. For reference loudness, and for speakers with lower impedance (or for the part of the spectrum where their impedance is lower) it takes more.

Rick "believing that many amp comparisons find compression or clipping even though the testers don't think they do" Denney

 

[boXem]

@MasterApex I had a few questions previously. I would be interested in knowing the answers in order to confirm or not what I suspect
- level matching at 1 kHz with RMS voltmeter or peaks recording
- model of the Purifi amp you tested
- volume attenuation setting for each amp

 

[Matias]

@MasterApex- model of the Purifi amp you tested

VTV Purifi Eigentakt EVAL-1

 

[boXem]

Join me in waiting for the new high power Purifi modules that Bruno is creating. They still did not finish the amp, so no specs or release dates, but it is ongoing. Hopefully one day you can repeat this test with a higher power Purifi amp and see how it goes.

The more experience I have (and my audio interest already goes back to the 70's), the more persuaded I am that more power is better, even at moderate listening volumes, to make sure those sharp peaks don't find compression, soft clipping, or even hard clipping. I've just had too many experiences where bigger amps of modest performance sound "better" than supposedly better amps of lower power. Level-matching at 1KHz is a steady-state match--it doesn't say anything about whether levels will be matched during the peaks.

If we level-match at 87 dB SPL (which is loud) at, for typical speakers, 2.83 volts (1 watt into 8 ohms), peaks might be as much as 106 or 108 dB SPL for recordings with high dynamics, but even those might be measured over a long enough period to gloss over the true peak. Those peaks would require 128 watts, and that's into 8 ohms. I think of amps in that power category as the minimum for playing really dynamic music loudly. For reference loudness, and for speakers with lower impedance (or for the part of the spectrum where their impedance is lower) it takes more.

Rick "believing that many amp comparisons find compression or clipping even though the testers don't think they do" Denney

Well, I have nothing against selling you guys higher power amplifiers at a premium, but my amplifiers have a clipping indicator (voltage and current) and until now none of my customers did get it engaged on the Purifi based one.

...

Thanks! So the DC at the output triggering speaker distortion is a first valid hypothesis.

 

[Matias]

@boXem | audio Depends on the speakers and material, really. He tested using B&W 802D3 which are large speakers, and Tchaikovsky 1812 canons are the worst. Easy to clip in this particular situation.

 

[rdenney]

Will the clipping indicator persist long enough to see if the clipped waveforms only last 6 or 8 ms? Seems to me that would be enough to change the character of the "articulation" (as musicians would put it). It's measurable distortion, of course, but the measurement protocols might miss that scenario.

Rick "never trusting clipping indicators as test equipment, but rather as speaker-damage warnings" Denney

 

[boXem]

@boXem | audio Depends on the speakers and material, really. He tested using B&W 802D3 which are large speakers, and Tchaikovsky 1812 canons are the worst. Easy to clip in this particular situation.

I have customers with B&W 802D3.

 

[boXem]

Will the clipping indicator persist long enough to see if the clipped waveforms only last 6 or 8 ms? Seems to me that would be enough to change the character of the "articulation" (as musicians would put it). It's measurable distortion, of course, but the measurement protocols might miss that scenario.

Rick "never trusting clipping indicators as test equipment, but rather as speaker-damage warnings" Denney

Yes. The indication duration is not related to the clipping duration.

 

[rdenney]

boXem, that's good that you have happy customers with moderately difficult speakers, but not definitive of the issue I raised, which may be specific to this test scenario.

Rick "wondering what the dynamics are of those cannon blasts--actually, I think I have that recording" Denney

 

[boXem]

boXem, that's good that you have happy customers with moderately difficult speakers, but not definitive of the issue I raised, which may be specific to this test scenario.

Rick "wondering what the dynamics are of those cannon blasts--actually, I think I have that recording" Denney

That's fine, I will contact you when I have high power monoblocks for sale

 

[Gorgonzola]

Please explain:

I guess I missed that MasterApex' comparisons including blind listening: my apology to him.

 

[VintageFlanker]

but my amplifiers have a clipping indicator (voltage and current) and until now none of my customers did get it engaged on the Purifi based one.

That's because I didn't test one yet...

 

[boXem]

That's because I didn't test one yet...

Well, I believe that one of the HCFR testers is even crazier than you

 

[mocenigo]

Interesting report.
[...]
Also Purifi has the lowest output impedance and lowest distortion (IMD) on the treble. This probably correlates with the perceived rolled off treble.

In fact, a lot of people expect distortion (even IMD) to present itself as a very unpleasing sonic signature, While this may be the case if it is mostly on the bass, one must understand that the most perceptible distortion is mostly on harmonics (of the difference of the two considered frequencies for the IMD), and therefore it will ADD strength to the signal at higher frequencies. This will obviously ("mathematically") imply two things:
1) increased volume at higher frequencies - and therefore a Purifi module will sound as if treble had been rolled off; and,
2) accordingly, higher separation at the treble, because the added artifacts will be more independent of each other, possibly resulting in a wider image, even though less stable

A signature aspect of high-tier Class D amplifiers is that the image is stunningly stable, not particularly wide, but with a deep stage, and sounding a bit rolled off in the treble with respect to high bandwidth AB amps.

Some key aspects for me are: stability of the staging, and, for orchestral classical music, very extended yet smooth treble, with a dark tone for strings. These are telltales of a realistic take and rendition of the recording, the latter confirming the quality of Purifi and Hypex modules.

You can get a wider-than-recorded scene, but it won't be sculpted, it will become larger at larger sound pressures and shrink when softer. Just use Class A. Well one Class AB can vary from very close to Class A to very close to Class D.

Roberto

 

[SIY]

In fact, a lot of people expect distortion (even IMD) to present itself as a very unpleasing sonic signature, While this may be the case if it is mostly on the bass, one must understand that distortion is always on harmonics, and therefore it will ADD strength to the signal at higher frequencies. This will obviously ("mathematically") imply two things:
1) increased volume at higher frequencies - and therefore a Purifi module will sound as if treble had been rolled off; and,
2) accordingly, higher separation at the treble, because the added artifacts will be more independent of each other, possibly resulting in a wider image, even though less stable

A signature aspect of high-tier Class D amplifiers is that the image is stunningly stable, not particularly wide, but with a deep stage, and sounding a bit rolled off in the treble with respect to high bandwidth AB amps.

Some key aspects for me are: stability of the staging, and, for orchestral classical music, very extended yet smooth treble, with a dark tone for strings. These are telltales of a realistic take and rendition of the recording, the latter confirming the quality of Purifi and Hypex modules.

You can get a wider-than-recorded scene, but it won't be sculpted, it will become larger at larger sound pressures and shrink when softer. Just use Class A. Well one Class AB can vary from very close to Class A to very close to Class D.

Roberto

Uh huh.

 

[mocenigo]

Uh huh.

What?
(Now I see, I added some clarification to my ramblings.)