Revel C52 Speaker Review and Measurements

[Krunok]

My findings on the audibilty of excess phase:
First of all, like anything subtle (phase errors sure fall into the subtle category), it takes some training and skills to hear it. Just like's Amir's outstanding capabilities to identify subtle codec artifacts in 320k MP3's vs original WAV required a long training period, phase error detection skills need the same level of dedication, so to say.
The problem area of excess phase is mainly below 1kHz or so where our hearing is not only a frequency/level analyser but also judges actual waveform and the waveforms get distorted by excess phase (and by the normal minimum phase of LF rolloffs also, that is). This has a slight effect on the "timbre" of steady-state signals that are asymmetric (even order overtones dominating), organ notes are a good example for this.
For "bass transients" like kick-drums, plucked upright bass and slapped electric bass etc, excess phase at LF spreads out the energy in time, the lower freqcuency components arriving later. This makes the transient sound less punchy, compact and impactful and also slightly sharper/spikey (the HF content gets masked less by the LF content). This is probably the most drastic effect one will hear after while. Subwoofer crossovers at the typical 80Hz do a lot of damage here, as do other bass to low-mid crossovers in the 200...400Hz range, the higher the order the worse.
I also found a sharpening / more pin-point stereo imaging and stage depth when excess phase is removed, but this is likely a side-effect of the reduced "computing power" we need apply to seperate transient incidents because those are more compact.
I also found that LF room modes etc don't have too much masking effect, the difference still is heard even when the bass region is quite corrupted in the time domain by room effects. On the other hand, phase errors are usually more benign when listening with headphones, not something one would expect and I don't have a good explanation for this.

My overall take: when designing speakers, don't sacrifice other more important aspects just for a notably better phase response. Linear phase can easily installed afterwards by inversely pre-distorting the signal with a FIR-"phase-unwrapper". Linear phase nomally means minimum phase, of the equivalent single driver speaker, but in some cases a bit of phase unwrapping of the minimum phase bass roll-off can be beneficial to speed up the bass response, for ported boxes with rather high cutoff (50Hz'ish) and the ususal 2nd order electrical subsonic filter found in active monitors, making the roll-off 6th-order.

Well, thank you for taking time to give such a detail answer!

First, let me comment that Amir, as an ex-Microsoft executive, is very good in self promotion. The thing that he can distinguish recording of jungling keys in mp3 vs wav is really nothing to what his wife can from the kitchen tell about speaker's response. Luckilly for us she has decided to step out of his shadow and help him evaluate the speakers.

What is your overall take when doing EQ regading phase alignement/correction? I did some in-room mesaurements and according to them it turns out that not only magnitude response but also the phase response varies from point to point. As modern EQ tools (and rePhase, if one chooses to do correction mannualy) beside FR also adjust the phase response at LP, would you say that this also results in subtle but audible gains in SQ, or is it just a marketing gymmick/waste of time?

 

[BYRTT]

Ah, those are perfect! What is the process of importing external data into it?

Think @napilopez had it covered into post #132 thanks for that, it needs one txt/frd file per step but as said its creator is open to suggestions and often dial in changes in a day or so. Seems you have tons of work load to do therefor made below dirty quick guide should you want to try it out, in example used your shared txt file for the nineteen horizontal directivity steps of NHT Pro M-00 but i had to manual use MS Excell to export each row as a seperate text file.

 

[Laserjock]

Would be interesting to know how the Performa3 C208 compares.

is this speaker (C52) a sealed design also?

 

[BYRTT]

Looks low end roll off for the on axis measurement is close to a 2nd order BW Q 0,85 @94Hz, those numbers point to its sealed design.

 

[MZKM]

Looks low end roll off for the on axis measurement is close to a 2nd order BW Q 0,85 @94Hz, those numbers point to its sealed design.

It is a sealed design if you look at the photos.

But yeah, if you wanna go the hard way of calculating the roll-off, you can do that.

 

[DDF]

@Krunok and @KSTR , some more grist for the mill:

Its the inner ear envelope (after all its non-linearities) that matters in this.

From https://secure.aes.org/forum/pubs/conventions/?elib=19404 "Audibility of Loudspeaker Group-Delay Characteristics"
Authors: Liski, Juho; Mäkivirta, Aki; Välimäki, Vesa
Affiliations: Aalto University, Espoo, Finland; Genelec Oy, Iisalmi, Finland(See document for exact affiliation information.)
AES Convention: 144 (May 2018) Paper Number: 10008
Publication Date: May 14, 2018

Loudspeaker impulse responses were studied using a paired-comparison listening test to learn about the audibility of the loudspeaker group-delay characteristics. Several modeled and six measured loudspeakers were included in this study.​

​

The impulse responses and their time-reversed versions were used in order to maximize the change in the temporal structure and group delay without affecting the magnitude spectrum, and the subjects were asked whether they could hear a difference. Additionally, the same impulse responses were compared after convolving them with a pink impulse, defined in this paper, which causes a low-frequency emphasis.​

​

The results give an idea of how much the group delay of a loudspeaker system can vary so that it is unlikely to cause audible effects in sound reproduction. Our results suggest that when the group delay in the frequency range from 300 Hz to 1 kHz is below 1.0 ms, it is inaudible. With low-frequency emphasis the group delay variations can be heard more easily.​

Also some additional anecdotal experience from John Siau at Benchmark:
https://www.audiosciencereview.com/...f-benchmark-ahb2-amp.7628/page-17#post-180486
"If the high-frequencies arrive first, they are accentuated. If the high frequencies arrive late, the impression will be that they are rolled off"

 

[BYRTT]

It is a sealed design if you look at the photos.

But yeah, if you wanna go the hard way of calculating the roll-off, you can do that.

Agree a first look in the photos it looks sealed design but sometimes seen some ports or acoustic resistors is implemented in invisible way, i'm not the best calculator guy so didn't go the hard way but used member pos free Rephase program to linearize the roll off knee and from there comes the numbers .

 

[Krunok]

@Krunok and @KSTR , some more grist for the mill:

Its the inner ear envelope (after all its non-linearities) that matters in this.

From https://secure.aes.org/forum/pubs/conventions/?elib=19404 "Audibility of Loudspeaker Group-Delay Characteristics"
Authors: Liski, Juho; Mäkivirta, Aki; Välimäki, Vesa
Affiliations: Aalto University, Espoo, Finland; Genelec Oy, Iisalmi, Finland(See document for exact affiliation information.)
AES Convention: 144 (May 2018) Paper Number: 10008
Publication Date: May 14, 2018

Loudspeaker impulse responses were studied using a paired-comparison listening test to learn about the audibility of the loudspeaker group-delay characteristics. Several modeled and six measured loudspeakers were included in this study.​

​

The impulse responses and their time-reversed versions were used in order to maximize the change in the temporal structure and group delay without affecting the magnitude spectrum, and the subjects were asked whether they could hear a difference. Additionally, the same impulse responses were compared after convolving them with a pink impulse, defined in this paper, which causes a low-frequency emphasis.​

​

The results give an idea of how much the group delay of a loudspeaker system can vary so that it is unlikely to cause audible effects in sound reproduction. Our results suggest that when the group delay in the frequency range from 300 Hz to 1 kHz is below 1.0 ms, it is inaudible. With low-frequency emphasis the group delay variations can be heard more easily.​

Also some additional anecdotal experience from John Siau at Benchmark:
https://www.audiosciencereview.com/...f-benchmark-ahb2-amp.7628/page-17#post-180486
"If the high-frequencies arrive first, they are accentuated. If the high frequencies arrive late, the impression will be that they are rolled off"

Nice read, thank you @DDF!

What bugs me here is that I'm not sure what is the right way to display GD in REW. Should some smoothing be applied or not? Should FDW window be applied or not? Maybe @JohnPM can help with answers to these questions..

Let me illustrate. Here is GD of my left speaker as measured from LP (4m from the speaker) with no FDW:

And here it is with FDW of 13 cycles:

Bot graphs are with 1/6 smoothing applied.

Obviously reflections affect GD in a big way.

 

[Laserjock]

Being sealed brings up another topic since a lot of speakers come with port plugs.

Feasible/interest to run another set of numbers with ports plugged if so designed as an option?

 

[Krunok]

Being sealed brings up another topic since a lot of speakers come with port plugs.

Feasible/interest to run another set of numbers with ports plugged if so designed as an option?

You plug the ports to reduce LF response when it is exaggarated because of wall proximity or corner position. In other words, speakers were designed with having port opened in mind and plugging the ports is a simple and rudimentary way of room EQ. As speakers are measured in the Klippel rig with no walls in proximity I really see no need for it. The same goes for HF shelve filter that many speakers have - probably the best idea when measuring such speaker is to put the filter in 0 (neutral) position.

 

[Thomas_A]

Just plugging the ports will cause other negative effects of LF response unless there are other ways to adjust the response. If there is an exaggerated bass the speaker is not properly designed for its use in a room regardless if it is sealed or ported, alternatively, the speaker is designed for a placement near or off the front wall but placed incorrectly.

 

[SDX-LV]

Think @napilopez had it covered into post #132 thanks for that, it needs one txt/frd file per step but as said its creator is open to suggestions and often dial in changes in a day or so. Seems you have tons of work load to do therefor made below dirty quick guide should you want to try it out, in example used your shared txt file for the nineteen horizontal directivity steps of NHT Pro M-00 but i had to manual use MS Excell to export each row as a seperate text file.

View attachment 46352

This is amazing, I think with this program and your explanation it should be possible to convert all the data from https://www.princeton.edu/3D3A/Directivity.html into Spinoramas !!!
I was waiting for this but did not know how, because they have decent anechoic measurements of KEF LS50, Genelec One and several other important speakers. Until Amir measures those speakers - this is the best source of Spins out there

Now if only there would be more time in a day of human life...

 

[DDF]

Nice read, thank you @DDF!

What bugs me here is that I'm not sure what is the right way to display GD in REW. Should some smoothing be applied or not? Should FDW window be applied or not?

Hi Krunok, hey, my pleasure.

Sorry, I don't use REW (MLSSA user here). But if you want to linearize the phase of your speaker, it has to be anechoic or quasi anechoic plus all excess (linear phase component) removed first. The smoothing applied to phase will be determined by your FFT resolution (window length and sample rate) and will limit phase resolution just as it does amplitude. I don't think this is a big deal as you shouldn't be seeing any sharp peaks or dips in group delay in any reasonable speaker, not that you want to eq anyway.

If I were doing this, I'd probably perform a quasi anechoic measure + near field splice (adjusted for diffraction), and phase EQ that. Diffraction as seen in one point in space (not integrated sound power wise) is minimum phase so modelling it this way is OK.

I think this is how SoundEasy performs phase linearization, for example http://www.bodziosoftware.com.au/Ultimate_Equalizer_Manual_5.pdf

 

[Senior NEET Engineer]

I don't necessarily disagree with you, but your last line does bring up a point I've been thinking about for some time: the more I listen to speakers, the more I feel the gap between great expensive speakers and great inexpensive speakers is really way smaller than the magnitude of their price difference. Of course, all of us measurements-liking folks know price isn't an indicator of quality, but it just becomes more obvious in my practical experience all the time.

What's your room acoustics like? Dedicated room that is enclosed and symmetrical, especially left to right symmetry? Good % of broadband absorption and diffusion? Good ETC? Good T60? 10+ feet to rear wall? Any deficiency in these will create a masking effect on speakers that makes them more similar than different.

 

[Krunok]

Hi Krunok, hey, my pleasure.

Sorry, I don't use REW (MLSSA user here). But if you want to linearize the phase of your speaker, it has to be anechoic or quasi anechoic plus all excess (linear phase component) removed first. The smoothing applied to phase will be determined by your FFT resolution (window length and sample rate) and will limit phase resolution just as it does amplitude. I don't think this is a big deal as you shouldn't be seeing any sharp peaks or dips in group delay in any reasonable speaker, not that you want to eq anyway.

If I were doing this, I'd probably perform a quasi anechoic measure + near field splice (adjusted for diffraction), and phase EQ that. Diffraction as seen in one point in space (not integrated sound power wise) is minimum phase so modelling it this way is OK.

I think this is how SoundEasy performs phase linearization, for example http://www.bodziosoftware.com.au/Ultimate_Equalizer_Manual_5.pdf

That was how I tried to do it the first time but then I realised phase and step response doesn't look good at all at LP (which is 4m from the speakers). There were also some cancellation in the bass in response when measuring at LP both speakers playing.

Then I measured phase (via sweep) from LP, applied FDW of 13 cycles and 1/6 smoothing and corrected it. Now it looks like this when measured from LP:

Left speaker at LP, phase (gery) , minimum phase (bue dotted) and excess phase (black) shown:

Phase of both speakers at LP:

Step response of both speakers at LP:

As you can see now everything looks fine when measured at LP, bass cancellations when measuring both speakers playing at LP are gone, but if I repeat quasi anechoic measurement the way you described it doesn't look so nice. But that is the case with magnitude response as well - if I do quasi anechoic measurement with room EQ filter for LP active it wouldn't look nice as well.

It seems to me that when I based my phase corections on pseudo anechoic measurement I was doing phase correction of the speaker and that this way does correction of the phase at LP. Pretty much the same as with magnitude response.

Is that so or I'm getting the things wrong?

 

[Krunok]

What's your room acoustics like? Dedicated room that is enclosed and symmetrical, especially left to right symmetry? Good % of broadband absorption and diffusion? Good ETC? Good T60? 10+ feet to rear wall? Any deficiency in these will create a masking effect on speakers that makes them more similar than different.

Not that I generally disagree with your points but If speaker would require to be placed 10+ feet from the rear wall very small number of living rooms would satisfy that criteria.

 

[thewas]

But that is the case with magnitude response as well - if I do quasi anechoic measurement with room EQ filter for LP active it wouldn't look nice as well.

That is normal because of 2 reasons
a) with just a normal sweep and a FDW of 13 cycles you include also some reflected sound
b) your Castle loudspeakers don't have perfect directivity behaviour
so by correcting the sum of direct and reflected sound (which doesn't have the same shape) at the listening position will make the direct sound not linear and according to many like should be avoided in higher frequency regions. That's why many recommend to increase the weighting of direct sound in higher frequencies by methods like reducing the FDW cycles at higher frequencies or using spacial averaging or using windowed or anechoic measurements for correcting that region.

 

[Krunok]

That is normal because of 2 reasons
a) with just a normal sweep and a FDW of 13 cycles you include also some reflected sound
b) your Castle loudspeakers don't have perfect directivity behaviour
so by correcting the sum of direct and reflected sound (which doesn't have the same shape) at the listening position will make the direct sound not linear and according to many like should be avoided in higher frequency regions. That's why many recommend to increase the weighting of direct sound in higher frequencies by methods like reducing the FDW cycles at higher frequencies or using spacial averaging or using windowed or anechoic measurements for correcting that region.

It has nothing to do with the directivity characteristic of my speakers - that is normal because measurement at LP which includes reflections will always differ from quasi-anechoic reflections taken near field. Besides, room EQ in frequency domain is always done with reflections included.

I kindly ask you to stay away from this discussion.

 

[napilopez]

Not that I generally disagree with your points but If speaker would require to be placed 10+ feet from the rear wall very small number of living rooms would satisfy that criteria.

The rear wall is the one behind you, not the speakers =]

What's your room acoustics like? Dedicated room that is enclosed and symmetrical, especially left to right symmetry? Good % of broadband absorption and diffusion? Good ETC? Good T60? 10+ feet to rear wall? Any deficiency in these will create a masking effect on speakers that makes them more similar than different.

To be clear, I said the gap between 'great' expensive and affordable speakers is smaller than presumed. In other words, I'm not completely sure I could tell an expensive speaker with exotic materials and design from a cheap one if they're both on the "good" side of the frequency and directivity spectrum - which is what blind tests have shown us anyway.

I'd be curious to see evidence otherwise though! Like, I wonder what the threshold is for small, non-frequency related improvements in an expensive speaker to lead to consistent preference. The Olive preference forumula is great, but still only 86 percent accurate in its generalized form. I wonder what the other 14 percent is =].

I think my room sounds pretty good. My apartment has a mostly open floor plan. 15-18ft irregular ceilings throughout most of it, but the speakers are located under a slatted loft that reflects some sound and lets some through. It's over 20 feet from my listening position to an irregular rear wall (aka my kitchen), and I also have an irregular front wall (custom shelving with books, and trinkets). Symmetrical left and right other than some funiture. Biggest potential issue imo is the room is fairly narrow at 13 feet, but I preferred this over having speakers along the long wall. T60 (per REW Topt) hovers a bit below 0.4s and ETC looks decent enough to me.

That aside, and I'm not denying the value of taking care of a room, but research tells us people are able to rank speakers consistently in a variety of different rooms, so in that sense I disagree your premise. Though I suppose you could argue even if the ordering of preference is the same, the gap between the speakers might be larger in an ideal room.

 

[thewas]

Besides, room EQ in frequency domain is always done with reflections included.

On of the reasons it is so problematic to EQ loudspeakers on the higher frequencies based on LP measurements.

As Toole says:

It is essential to note that this is the room curve that would result from subjectively highly-rated loudspeakers. It is predictable from comprehensive anechoic data (the "early reflections curve in a spinorama). If you measure such a curve in your room, you can take credit for selecting excellent loudspeakers. If not, it is likely that your loudspeakers have frequency response or directivity irregularities. Equalization can address frequency response issues, but cannot fix directivity issues. Consider getting better loudspeakers. Equalizing flawed loudspeakers to match this room curve does not guarantee anything in terms of sound quality.

and

E-r-r-r-r, have we not been saying that the small peaks and dips in room curves are likely to be caused by non-minimum phase phenomena, most likely caused by reflections and cannot be equalized. To two ears and a brain they are innocuous spaciousness, not coloration. It is attempts to "correct" such fluctuations that lead to degradation of well designed loudspeakers. So, above the transition frequency, small details in steady-state room curves should be ignored because unless you have comprehensive anechoic data on the loudspeakers you don't know what caused them

I kindly ask you to stay away from this discussion.

As fas as I now it this a free and open forum, if you don't like what I am posting just don't read it.