Topping PA5 Review (Amplifier)

[voodooless]

If you don’t think phase is audible flip polarity of one speaker for 180deg. That’s an extreme case I know but despite a room having reflections, the direct sound from the speaker is very important. Even on non bass heavy content music the “phasiness” is instantly detectable (it is in the 1kHz to 6kHz range). I can tell when speakers are accidentally flipped with ears instantly.

That is obviously different. First off, it's for all frequencies, not just HF. Secondly, the change is 180 deg, not the 30 to 40 deg at 20 kHz that we see in this example. Thirdly, you keep one speaker's phase in tact. Obviously, you'll get interference from that.

 

[sarumbear]

If you don’t think phase is audible flip polarity of one speaker for 180deg. That’s an extreme case I know but despite a room having reflections, the direct sound from the speaker is very important. Even on non bass heavy content music the “phasiness” is instantly detectable (it is in the 1kHz to 6kHz range). I can tell when speakers are accidentally flipped with ears instantly.

How about reversing both speakers?

 

[Paco De Lucia]

I have an old amplifier (class D) which suffers from wandering soundstage in its old age. Its imagining was amazing until it started wandering around from second to second. Reading the last few pages of this thread I suspect it was a phase problem. I wonder what components have failed or worn out to bring this age related behaviour.

 

[SylphAudio]

Please take into account that it depends on load complex impedance and amplifier circuit topology. Please let me answer you by measurements of amplitude frequency response and phase frequency response of a conventional class AB amplifier vs. TPA3255 AIYIMA A07 - and think about it. If more discussion is needed, a separate thread would be an option, I can paste my measurements there. Is this audible in a DBT test? Yes, not easily, but with better than 5% uncertainty. Is this because of phase response? Just related to, as a derived parameter. Can we read this from 4/8 ohm resistive response? With difficulties.

View attachment 172230

View attachment 172231

P.S.: new plots with soundcard FR elimination

Why is phase change not apparent when amplitude FR clearly had had a big change because of the LC filter resonance? I thought douglas self said that phase FR is complety defined by its amplitude FR, or his statement is only applicable with class a/b topologies?

Edit: AIYIMA A07, -40+ degrees at 20kHz is as good as Hypex NC400 phase shift. Now I'm curious on how this PA5 or other PFFB TPA3255 compares.

If you don’t think phase is audible flip polarity of one speaker for 180deg. That’s an extreme case I know but despite a room having reflections, the direct sound from the speaker is very important. Even on non bass heavy content music the “phasiness” is instantly detectable (it is in the 1kHz to 6kHz range). I can tell when speakers are accidentally flipped with ears instantly.

Sound waves will cancel each other out and the effect is more detectable by the ears in low frequencies. Apparently the sound stage becomes wider though, I thought it is because the bass was gone and I'm hearing more mids and highs, but if you remove the bass content in music, the effect is not the same. Is it because the sound waves in the room or ear response is now messier ? I think our ears can really detect if there's something's wrong that's why some people prefer full range speaker driver for best soundstage and imaging. (most people don't care)

 

[KSTR]

Yes, you tested A + B together, not these factors separately as far as I know. The question is: which one of the two is audible in this case: phase or amplitude. Or did I totally miss that test?

Test magnitude response with pink noise which is phase-insensitive and very revealing of even the most minute magnitude differences.
Test phase response... trickier as it depends on what aspect of phase errors you want to test for...

 

[KSTR]

P.S.: new plots with soundcard FR elimination

had already started wondering... ;-)

 

[pma]

had already started wondering... ;-)

2 channel measurement in Steps works properly and eliminates soundcard intrinsic amplitude and phase response, however in REW it is not the case. Even if you make measurements with timing reference and A over B with soundcard response, the phase plot result is wrong. So back to Arta/Steps, sometimes.

 

[xrk971]

How about reversing both speakers?

This will be more apparent on music with sharp percussion and kick drums. Especially if speaker is transient perfect. Pressure rarefaction from a kick drum sucking the woofer in on the initial punch is more audible on some types of speakers than others. Air’s response to compression (woofer moving out) vs rarefaction (woofer moving in) is asymmetric (Boyle’s law) and is what gives rise to second harmonic distortion from sound waves naturally propagating through air as a medium.

 

[xrk971]

I have an old amplifier (class D) which suffers from wandering soundstage in its old age. Its imagining was amazing until it started wandering around from second to second. Reading the last few pages of this thread I suspect it was a phase problem. I wonder what components have failed or worn out to bring this age related behaviour.

Variable channel crosstalk (from a possible intermittent left/right short on audio input) can cause this too. And non steady phase from a malfunctioning amp can also as well.

 

[Paco De Lucia]

Variable channel crosstalk (from a possible intermittent left/right short on audio input) can cause this too. And non steady phase from a malfunctioning amp can also as well.

Interesting, it has a built in DAC so the shorted input is unlikely so its probably something else, I'd love to know as i'm handy with a soldering iron )) thank you ))

 

[maty]

Phase shift in class D amplifiers. How it affects the sound?

Feedback is not a bandaid, it is a design method to make circuits more effective, controllable & reliable. Feedback is also a naturally occurring reality, which we use every day with our movement as humans which is modified by our sensory feedback..

www.diyaudio.com

Icebrick IcePower Engine-400
https://icepower.dk/products/amplifier-power-modules/icebricks/

 

[KSTR]

@maty, what are you trying to show us, that Hilbert was right?
First order LP gives 6degree phase shift at 1/10th the -3dB frequency, a well-known rule of thumb (for EE's).

 

[KSTR]

This will be more apparent on music with sharp percussion and kick drums. Especially if speaker is transient perfect. Pressure rarefaction from a kick drum sucking the woofer in on the initial punch is more audible on some types of speakers than others. Air’s response to compression (woofer moving out) vs rarefaction (woofer moving in) is asymmetric (Boyle’s law) and is what gives rise to second harmonic distortion from sound waves naturally propagating through air as a medium.

Just that this distortion (if there is any, really) is not the reason why kick drum sounds different when inverted.

 

[SylphAudio]

My personal conclusion:

1. Class D amplifiers have improved a lot of the years but Class D Amps are still no match to good design Class AB Amp.in term of sound quality and dynamic handling
2. Benchmark AHB2 is more versatile (good for all music) and more dynamic than the Class D Purifi Eigentakt
3. Class D Purifi Eigentakt is suitable for center channel, its slight compression and unnatural bias toward midrange makes dialogue more clear It is not an ideal amp for life classical music reproduction.
4. Mark Levinson 333 is the best despite its old age. It is the most life-like, with more weight on vocal , and more at ease when playing life recording classical orchestra at higher volume level
5. There needs to additional scientific measurements that can explain why the benchmark AHB2 sounds better than Purifi Eigentakt in my environment and why the levinson sounds more dynamic and detailed on loud classical music. Traditional SINAD, THD , IMD measurement using sinewave do not have the musical waverform complexity and can not explain the contrast of what we heard among these amplifiers in our environment. What we heard can not be explained with just THD measurement as they all sound good (not distorted) but different. As below certain THD, the delta is not audible.

Although this is only anecdotal and not a controlled test, I think phase response has to do with his listening impressions, so I am very insterested with phase response graph of AHB2 and Purifi as well.

 

[maty]

That a great manufacturer of class D amplifier modules specifies the phase should attract our attention. Presumably they advertise it as something to consider when choosing.

That said, phase should be measured in class D amplifiers. Amir has only measured Hypex NC400. Maybe we will have surprises.

 

[KSTR]

That a great manufacturer of class D amplifier modules specifies the phase should attract our attention. Presumably they advertise it as something to consider when choosing.

Just a shortcut to what's immediately obvious, plus marketing.

 

[maty]

Phase shift in class D amplifiers. How it affects the sound?

Here's a probably misplaced spanner. Whilst you lot are discussing the nuances of phase shift in Class D amplifiers I'm left wondering about current slew rate in the output filter inductor. Let's say it can just support a full peak to peak 20KHz sine wave with nothing else going on. Now add...

www.diyaudio.com

 

[voodooless]

Test magnitude response with pink noise which is phase-insensitive and very revealing of even the most minute magnitude differences.
Test phase response... trickier as it depends on what aspect of phase errors you want to test for...

I'm not that interested in pink noise though. That is not music.

 

[KSTR]

2 channel measurement in Steps works properly and eliminates soundcard intrinsic amplitude and phase response, however in REW it is not the case. Even if you make measurements with timing reference and A over B with soundcard response, the phase plot result is wrong. So back to Arta/Steps, sometimes.

No issue with REW for me:

(The top end magnitude warping above 40kHz is because that is an IIR filter in this example, 2nd order LP @20kHz, Q=1)

"Set T=0 at IR start" must be selected, then A/B works perfectly (as long ASIO buffer size isn't changed in between, or different AD or DA reconstruction filters, etc). Also, REW can measure with loopback reference as well.

 

[KSTR]

I'm not that interested in pink noise though. That is not music.

"Normal music" will far less revealing. All depends on what you want to know...