Amp freq response: Sonos vs Peachtree

  • 12 April 2019
  • 84 replies
  • 2229 views

There have been questions about the high frequency response of the Sonos Amp. User hodgeal posted a very interesting plot showing the frequency response of the Sonos Amp using the software REW. It didn't show excessive high frequencies in that single test.

I had the opportunity to compare the Sonos amp to my Peachtree amp using the same software (which is very useful and easy to use - thanks hodgeal), and this provides a comparative test of the Sonos Amp frequency response.

The Peachtree Nova 125 SE amp is well-respected in the audio community. It is a class D digital amp, same as the Sonos Amp, but as far as I can tell uses different chips. The test uses Duntech Marquis speakers and a Studio Projects C1 microphone. The mic is setup on-axis at a distance of 120cm, which is pretty standard. But none of the details should matter because this is a comparison under identical conditions - I simply swapped amps and repeated the test.

And the results?



As you can see, the amps are almost identical. Ignore low frequencies - they vary from test to test even with the same amp below about 30Hz. At high frequencies, above about 4kHz, the Peachtree gets gradually brighter, ending up being about 1.5dB louder at 20kHz. This would be imperceptible to pretty much everyone.

Note that there is a slight increase in high frequencies above about 7kHz for both amps, so this could be the speakers. However, I think this is due to the frequency response of the microphone - the specifications show a similar slight increase.

Cheers, Peter.

84 replies

I don't understand how frequency responses of mics and speakers come into play, and if they do, how is this a frequency response measurement of an amp? Isn't a signal generator at the input end and a voltmeter/oscilloscope across the speaker terminals of the amp the way to do this?
Ideally one would use an oscilloscope to measure just the amp. I don't have one. Even then, one would need to pay attention to the impedance, so the oscilloscope or voltmeter looked like a speaker load. Voltmeters typically have a very high resistance - I don't know how that would change the measurement.

The measurements that hodgeal and I have posted include the speaker and microphone responses. On their own they reveal little about the amp, unless it is noticeably bright as I understand some people have claimed. However, the comparative plots are useful, since they change one variable only. In my case, I change just the amp, leaving everything else the same. The difference between the two plots is then attributable to the difference between the amps regardless of the influence of the speakers or microphone.

Cheers, Peter.
So what you have found is that both amps react with the speakers in the same way and if one is bright, so is the other to the same extent. Certainly that is a lot more than those claiming brightness for the Sonos amp with nothing to back up a subjective and probably defective assessment.
Userlevel 7
Badge +20
Thanks, Peter. Interesting information, so much better than simplistic qualitative assessments (including my own).
It would be interesting to know of any name brand amp that delivers different results beyond measurement error differences if dropped into this set up; that would disprove the claim that amps do not have sonic signatures as long as they don't clip.

Not that I expect you, Peter, to do this, just saying that anyone that wants to so disprove that statement can do so beyond merely saying so.

Ditto for DACs.
Userlevel 2
Badge +1
Nice work! Pretty strong evidence that the brightness reputation is groundless. Maybe some reviewers did not have the amp running with all the settings configured for flat output.

P.S. would you mind plotting just 200-20000Hz and blowing up the vertical scale?
Of course the thing about the Sonos Amp is that with a built in DAC as well as DSP, it is more than just an amp. Those that have claimed brightness have also noted that they were able to tame it by using the EQ which should mean that brightness was found with initially flat EQ. And then there is the case of one that found that the sound righted itself after a power down/power on cycle. All of which muddies the waters somewhat.
P.S. would you mind plotting just 200-20000Hz and blowing up the vertical scale?



Done!
Of course the thing about the Sonos Amp is that with a built in DAC as well as DSP, it is more than just an amp. Those that have claimed brightness have also noted that they were able to tame it by using the EQ which should mean that brightness was found with initially flat EQ. And then there is the case of one that found that the sound righted itself after a power down/power on cycle. All of which muddies the waters somewhat.

When I first set up my amp, it took a while to realise that loudness was turned on by default. Perhaps this has caused reviewers to note brightness, since loudness boosts the high frequencies as well. I also noted elsewhere that the sub was turned on by default, although no sub was attached. The latter doesn't affect the sound, but loudness does, particularly at lower volumes (I'm assuming it works this way. hodgeal only tested at 50% volume. Perhaps there is a software bug here? I might try and test this when I get time.)

In any event, it's a bit silly to have loudness and sub turned on by default. Loudness may be reasonable for Play 1's, but not the amp with your own speakers. Perhaps the fact that loudness is enabled should be shown somewhere prominently in the app, like next to the volume control.

Kumar, do you have links to the reviews you mention that complain about brightness?

Cheers, Peter.


In any event, it's a bit silly to have loudness and sub turned on by default.


Can other AMP owners confirm that the default settings after initial setup is indeed Loudness=ON?

IIRC with the Connect and Connect:AMP the loudness is disabled by default.
Peter,
One that talks about the effect of the loudness thing:
https://www.whathifi.com/reviews/sonos-amp

One that says it is bright and claims that this is also how other reviews have found it:

https://www.theverge.com/2019/3/7/18254803/sonos-amp-review-speaker-specs-features-price

Another one that says the same about brightness:

https://www.soundandvision.com/content/sonos-amp-streaming-amplifier-review-page-2

No surprise, none have any measurements attached.

I suspect that posters here that say the same thing have read these and other such reviews and been influenced by these. There is nothing to be said about one poster that found it bright, but found that burning in takes care of that:-).

As to the loudness thing - the usual design for that is to boost the lower mid to low side of things at low sound levels to bring back some depth/richness to the sound that is otherwise missing at those levels. This boost tapers off as sound levels approach 50%. Treble is usually not boosted by the loudness switch. I have no idea how Sonos has built this function for the Sonos Amp.

Can other AMP owners confirm that the default settings after initial setup is indeed Loudness=ON?

At least one of the reviews I linked confirms this.
Peter,
As to the loudness thing - the usual design for that is to boost the lower mid to low side of things at low sound levels to bring back some depth/richness to the sound that is otherwise missing at those levels. This boost tapers off as sound levels approach 50%. Treble is usually not boosted by the loudness switch.


I disagree. The usual design of a loudness switch is to boost both bass and treble. The reason is that human hearing is less sensitive in both the bass and treble frequencies at low volumes. This has been traditionally demonstrated by the "Fletcher-Munson curves", although I believe they have been superseded by the similar "Equal-loudness contours".

Cheers, Peter.
Does your graph allow you to see what Sonos actually does via loudness?

See this for my understanding:

"The most famous, well-used measurement for plotting our perception of loudness against the frequency of tones is the Fletcher-Munson curve(s) of equal loudness, published in 1933. The graphs were updated in 1956 by D.W. Robinson and R.S. Dadson. They were refined again in 2003 and adopted by the International Standards Organization as ISO 226. The purpose of the graph is to show that for humans to consider two pitches equally loud, the amount of energy necessary to produce the tone at one frequency may be completely different than at another. The multiple lines, spaced in 10 dB increments, also show that the energy/frequency differences are steeper at lower intensity levels and flatten out at extremely high intensities, though not quite as much as Fletcher and Munson first determined.
The "loudness button" on a stereo amplifier is intended to boost bass frequencies at lower volume levels where the curve is the steepest."

Perhaps it is the 2003 revision that drives the boost to just the bass frequencies. In my listening as well, the impression conveyed by the loudness switch is a higher presence of the lower end at low volumes. But it would be interesting to see what a measurement shows, for Sonos.

It is quite possible that my ears don't hear much treble in the first place, for the increase to be noticed!
I see that the other LS50 thread has a graph that shows what the loudness toggle on Sonos Amp does and it shows a noticeably higher boost at the low end compared to that at the treble end.
That's at 50% volume levels; I wonder how it would look at the 20% mark.
Here are graphs of the effect of the loudness control on the Sonos Amp at volumes of 25%, 50% and 75%. The legend indicates which plot is which, but it should be fairly clear. The largest effect is at 25% volume, where the bass is boosted by about 5dB and the treble by about 2.5 dB. At a volume of 50% the effect is roughly halved. At a volume of 75% the loudness control has no effect. The reason there is more boost to the bass than the treble is because our ears lose sensitivity at low volumes more at bass frequencies than at treble frequencies (still true for Equal loudness ISO 226:2003 curves).



Again, note that it is only differences that are meaningful in this plot. Pairs of lines at the same volume have been shifted vertically for clarity, and as discussed earlier, a single line includes the effect of the speaker and microphone (as well as mic pre-amp and sound card response, room effects at low frequencies etc).

Cheers, Peter.
Very nice, thank you!
I see that there are two lines for 75%, but are seen as one because they overlap completely?
This bears out what I hear in other Sonos units - a palpable elevation of bass and even some vocals, and I expect that I don't hear the smaller boost to the treble given my already age affected hearing of those frequencies. And because a lot of my music is low/midrange dominated jazz, with vocals not being the opera/soprano kind.
I am little surprised to see that at 50%, which is moderately loud where my play units are concerned, there is still a significant effect of the loudness toggle and that it takes levels to be as high as 75% for it to vanish.
I see that there are two lines for 75%, but are seen as one because they overlap completely?

Correct.

I am little surprised to see that at 50%, which is moderately loud where my play units are concerned, there is still a significant effect of the loudness toggle and that it takes levels to be as high as 75% for it to vanish.

Agreed. I would rarely listen at this level.

The volume taper used by Sonos is quite different to most amps. My Peachtree reached very high volumes at about 20%, but didn't change as much thereafter. The Sonos Amp is just warming up at 50%, but gets much louder after that. I have run it at 100%, and was surprised at the lack of any distortion. I never ran the Peachtree at 100%, but I probably could have. Just wasn't game!

Cheers, Peter.

The volume taper used by Sonos is quite different to most amps.

It is more honest and therefore makes the volume control useable across the entire range of its motion, allowing for finer control at the lower end unlike most amps of today that seek to sound more powerful by very quickly reaching peak distortion free sound levels.
Peter, see the linked review page that has the most detailed comments in elaboration of the brightness finding. Strangely enough, the reviewer did not see this go away by toggling Loudness to off, but by a radical reduction of treble in the EQ control.

Of course the review also talks about things like the amp being more dynamic than another, to the preference of the reviewer. What dynamic means is a mysterious part of the equally mysterious PRAT attributes of amplifiers that some reviewers love to talk about: Pace, Rhythm, Attack, and Timing.

https://www.cnet.com/reviews/sonos-amp-review/2/


In any event, it's a bit silly to have loudness and sub turned on by default.


Can other AMP owners confirm that the default settings after initial setup is indeed Loudness=ON?

IIRC with the Connect and Connect:AMP the loudness is disabled by default.


If I recall correctly, Loudness was enabled after initial setup both times I have set up an Amp.

FWIW, I don't think the Amp is brighter than other amps. Mine replaced a Grace Design M920 + Hafler power amp. I did have to futz with the sub crossover and level when I switched to the Amp, but once I did, the setup up sounds the same to me.
Userlevel 2
Badge +1
Thanks again Peter for your measurements. Something else occurred to me. The DDFA technology used in the Sonos Amp is known to have a high end frequency response that is load dependent. For example, see the curves below from measurements of the NAD M2, which also uses DDFA (from https://www.stereophile.com/content/nad-m2-direct-digital-integrated-amplifier-measurements). Now consider that real speaker impedance at high frequencies may be far from nominal 4 ohms or 8 ohms, and possibly the high end response could be somewhat variable with different speakers. So I wonder if this could partly explain what some reviewers are hearing.


[Fig.2 NAD M2, 4 ohms speaker compensation, frequency response at 2.83V into: simulated loudspeaker load (gray), 8 ohms (left channel blue, right red), 4 ohms (left cyan, right magenta), 2 ohms (green). (2dB/vertical div.)
Now consider that real speaker impedance at high frequencies may be far from nominal 4 ohms or 8 ohms, and possibly the high end response could be somewhat variable with different speakers. So I wonder if this could partly explain what some reviewers are hearing.

That's a really interesting thought. Presumably all amplifiers have to be able to handle the frequency-dependent impedance of a whole range of speakers. I seem to remember my Duntech Marquis have impedance ranging from 3-6 ohms (nominally 4 ohms) at different frequencies. If DDFA-based amps are more susceptible to this variable load, I can imagine there could be some issues. However, presumably all DDFA-based amps would be affected in a similar way. I'm not sure if this has come through in reviews of different brands. Of course there could always be design choices unique to Sonos that exacerbate the problem.

Cheers, Peter.
Userlevel 2
Badge +1
I seem to remember my Duntech Marquis have impedance ranging from 3-6 ohms (nominally 4 ohms) at different frequencies.

Interesting. This fits what the M2 shows if we assume the Sonos Amp is dialed in for lower impedance. Then the amp might be a little bright with higher impedance speakers. I ordered some power resistors so maybe I can do some simple tests with dummy loads. But first I emailed Sonos to ask if they have any curves that they are able to share. I don't know if Sonos people are monitoring this thread and might chime in?

DDFA uses some feedback taken right at the speaker connections and so I thought it would compensate for different loads. But the M2 tests seem to show that this might only work well over a limited range of loading. Maybe Qualcomm has improved this with newer DDFA technology?
Peter, if you can answer this after measurements, questions that arose in another thread:

What happens at the speaker terminals of the Sonos Amp if a blank jack is plugged into the Sub out jack? Are the speaker terminals then denied frequencies below the crossover setting of 80Hz even if there is nothing at the other end of the jack? And then does turning the Sub on/off in room settings for the Sonos amp in the Sonos app have any impact on what happens at the speaker terminals? Are low frequencies restored to the speaker terminals with the jack plugged in, but the Sub setting on off?

Reply