Infinity RS152 Review (Surround Speaker)

[Juhazi]

https://www.dolby.com/about/support/guide/speaker-setup-guides/7.1-virtual-speakers-setup-guide/
https://www.dolby.com/about/support/guide/dolby-atmos-speaker-setup/

https://www.dolby.com/about/support/guide/speaker-setup-guides/

I believe that bipoles are mentioned only because they are so popular and historical.

 

[MZKM]

Dolby says bipoles are still fine if you are seated close to them, which is the case in many homes.

No mention for Atmos, but they say no dipoles:
https://www.dolby.com/siteassets/te...atmos-installation-guidelines-121318_r3.1.pdf

 

[Vladimir Filevski]

These two measurements showing the exact same slopes are case closed for me.

You are closing too early. Anti-Covid measures?
- From anechoic measurements in post #1 (thanks, Amir!): maximum difference between 650 Hz dip and 2-8 kHz tweeter plateau is exactly 15 dB. The "650 Hz" dip is very wide - between 300 Hz and 900 Hz.
- From in-room measurement in post #61 (thanks, Worth Davis!): maximum difference between 650 Hz dip and 2-5 kHz tweeter plateau is exactly (and only!) 8 dB, including the peak at 4 kHz. Much better! Because of the dip between 6 and 8 kHz (due to two-tweeter interference) I suspect this imeasuring is done with microphone in only one position (i.e. not moving-microphone technique). Also, peak at 5 kHz may be from the same two-tweeter interference and maybe is not "real", so maximum difference is 7 dB. Taking the "0 dB" line as reference, the "650 Hz" dip is between 500 Hz and 900 Hz - big improvement over anechoic 300 Hz and 900 Hz.
- From in-room measurement in post #65 (thanks, spacevector!) maximum difference between 650 Hz dip and tweeter peak is exactly 11 dB, (left surround) and 9 dB (right surround). We can say it is 10 dB average - again, better than anechoic. Also, the "650 Hz" dip is between 550 Hz and 900 Hz - big improvement over anechoic!
- From in-room measurement with moving-microphone technique in post #117 (thanks, Chromatischism!) maximum difference between 650 Hz dip and tweeter peak is exactly 7 dB - again, much better than anechoic. Also, the "650 Hz" dip is between 400 Hz and 900 Hz - improvement over anechoic!

So, we can conclude this:
1. Tweeters between 2 kHz and 8 kHz are too "hot" - both in anechoic and in-room measurements. Why are loudspeaker manufacturers so obsessed with overly loud tweeters?
2. Dip at about 1.5 kHz is real. It can be made much smaller with some sort of phase-plug in front of the woofer cone, as a part of the mounting back plate. Plastic is cheap.
3. The "650 Hz dip" is (kinda) real, but not so severe as in anechoic measurements. In room (and on-wall) it is not so deep and wide. Also, if the high-pass filter of tweeter was done right, the "650 Hz dip" largely disappears.

There is a clear dip here, yes the room modes(and comb filtering, which is this speaker's fault) are hiding how bad it is, but they are not hiding its presence. This sort of extremely broad dip simply does not happen due to room modes alone.
Not sure why you think it isn't there. It 100% is.
View attachment 124691

Oh, cmo'n! You intentionally rejected much of the frequency response. Overall trend is very clear from the complete in-room frequency response:
View attachment 124686
As we can see, below 300 Hz are peaks and dips from room modes, and our reference horizontal line can be anywhere between 72 dB and 76 dB. If we choose the 72 dB line, than there is no "650 Hz dip", but there is excessive tweeter output (which we know is true). If we choose the 76 dB line, than the tweeter output is neutral (which we know is not true!) and there is a wide "650 Hz dip" depression between 400 Hz and 900 Hz, but only 4-6 dB deep. For me, the only real problem is too hot tweeter.

 

[edechamps]

Seriously, did you not read the rest of the thread? You have it totally wrong, and just about everything you claimed as incorrect is in fact correct. You would do well to take the time to understand just what the Klippel system is, how it works and what its limitations are. It is not a perfect system, there are specific limitations about how it works and in particular the assumptions the system makes in projecting the measured field into the room are limiting. Sadly it does not model the expansion of the hologrpahic field it measures into the room, it just makes some very basic calculations of rough aggregate values off the walls. It just expands the calculated polars into the room and applies some simple weightings. It does not model phase, and basically models the speaker in the room as monopole. Thus is only valid for a conventional box speaker in free space in the room. It gets in-room respose of panels and other di-pole speakers wrong in-room because of this.
A speaker that requires the loading of a wall and is intended to utilise that wall as part of the radiating system, one designed to radiate into a half space, is not modelled by Klippel's software. The RS152 measured here is designed to include a wall as an intrinsic part of the speaker's operation. Klippel's software has no way of understanding this, there is no mechanism for even telling it that there should be a wall. Without a wall, the measurements, whilst interesting do not, and cannot, reflect actual intended use case operation.

To be fair, what you're describing here are problems with CEA-2034, not the NFS. If you had measured the speaker in a real anechoic chamber and applied CEA-2034 to these measurements, you would (in theory at least) get the same results. Shoot the message, not the messenger.

Maybe there are ways to "tweak" CEA-2034 for speakers mounted directly on a wall. Measuring with the speaker mounted on a baffle might do it. The measurements would of course show zero output for rear angles but perhaps the CEA-2034 averages might still work fine despite that.

 

[MZKM]

To be fair, what you're describing here are problems with CEA-2034, not the NFS. If you had measured the speaker in a real anechoic chamber and applied CEA-2034 to these measurements, you would (in theory at least) get the same results. Shoot the message, not the messenger.

Maybe there are ways to "tweak" CEA-2034 for speakers mounted directly on a wall. Measuring with the speaker mounted on a baffle might do it. The measurements would of course show zero output for rear angles but perhaps the CEA-2034 averages might still work fine despite that.

Here is what the standard says:


6 Directivity Response – In-Wall, In-Ceiling, On-Wall and On-Ceiling Loudspeakers

The device under test shall be tested in an anechoic, or hemi-anechoic, chamber. It shall also be mounted in (for in-wall/in-ceiling DUTs) or on (for on-wall/on-ceiling DUTs), a rigid surface that represents a wall, ceiling or floor to yield a 2 π measurement space, or in a baffle. It may be easier to extend the size of the rigid surface if it is laid horizontally on the floor of the test chamber. If the manufacturer specifies a particular rear box design then this rear box shall be installed during testing, otherwise the cavity behind the device under test shall be no larger than 40 liters, which approximates the cavity between the studs in a typical wall. If a baffle is used it must be at least 1.98 m high by 1.98 m wide, and the device under test shall be mounted in the center of the baffle.
The microphone used for taking measurements shall be 2 m from the device under test.
All of the requirements regarding the measurement environment, measurement apparatus, measurement distance and input signal described in Section 5.1 also apply here.
With the device under test mounted as described above, one of the following measurement methods shall be used for collecting data.


6.2 Data Acquisition – Rotating Microphone

Record the front hemispherical (180°) horizontal and vertical polar measurements at ten degree increments. These measurements shall be taken by moving the measurement microphone from one angular position to the next in sequence around the vertical axis as shown in Figure 1 until the front 180° have been covered. Then by moving the measurement microphone from one angular position to the next in sequence around the horizontal axis as shown in Figure 1 until the front 180° have been covered.
Instead of moving a single microphone it may be preferable to use 37 separate stationary microphones (note that the microphone directly in front of the DUT would be used twice, once for each axis). The DUT remains stationary during this test.
Also, instead of moving the microphone around the horizontal axis it may be preferable to rotate the DUT 90° about its reference axis and repeat the original rotation of the microphone about the vertical axis to obtain the required measurements.

 

[Chromatischism]

No mention for Atmos, but they say no dipoles:
https://www.dolby.com/siteassets/te...atmos-installation-guidelines-121318_r3.1.pdf

Yeah, dipoles were a non-localization trick from the 5.1/7.1 days. But bipoles are still ok.

 

[Sancus]

s. In room (and on-wall) it is not so deep and wide. Also, if the high-pass filter of tweeter was done right, the "650 Hz dip" largely disappears.


Oh, cmo'n! You intentionally rejected much of the frequency response. Overall trend is very clear from the complete in-room frequency response:
View attachment 124686
As we can see, below 300 Hz are peaks and dips from room modes, and our reference horizontal line can be anywhere between 72 dB and 76 dB. If we choose the 72 dB line, than there is no "650 Hz dip", but there is excessive tweeter output (which we know is true). If we choose the 76 dB line, than the tweeter output is neutral (which we know is not true!) and there is a wide "650 Hz dip" depression between 400 Hz and 900 Hz, but only 4-6 dB deep. For me, the only real problem is too hot tweeter.

Yikes, people really are intent on rejecting the basic facts in this thread. I'm done with it. The speaker isn't good, and the desperation of people trying to pretend the anechoic results are irrelevant is not a good look either, but I'm very familiar with this sort of delusion from the Magnepan threads and it's not worth engaging with tbh.

 

[Deleted member 2944]

Yikes, people really are intent on rejecting the basic facts in this thread. I'm done with it. The speaker isn't good, and the desperation of people trying to pretend the anechoic results are irrelevant is not a good look either, but I'm very familiar with this sort of delusion from the Magnepan threads and it's not worth engaging with tbh.

I wonder how the NFS would characterize and old Allison Model Two. That would be fun, eh?
How about a Klipschorn??
How about numerous other speakers that are designed assuming an augmentation of some sort.....via room boundary, or location, or mounting, or some other aspect??

An NFS test takes away that augmentation. How can you reconcile that disparity, meaningfully, in all cases?
Think about that a little bit.

Dave.

 

[Vladimir Filevski]

Yikes, people really are intent on rejecting the basic facts in this thread. I'm done with it. The speaker isn't good, and the desperation of people trying to pretend the anechoic results are irrelevant is not a good look either, but I'm very familiar with this sort of delusion from the Magnepan threads and it's not worth engaging with tbh.

?
The essence of all my posts in Magnepan LRS thread was that anechoic Klippel NFS measurements tells much of the story. And for every other dipole loudspeaker.
My personal opinion is that for really accurate prediction of in-room performance of true dipole speakers (Magnepan, etc), Klippel NFS need a little bit different weighting coefficients for the back and side measurements.
For accurate prediction of performance of in-wall speakers Klippel NFS has bespoke panel which replaces the actual wall. If the anechoic measurements is all you need, why on earth Klippel made that panel?
Infinity RS152 is not good, but in a view of low price, is passable. If the tweeters are attenuated about 4 dB, it would be a good (but not great) surround speaker.

 

[Juhazi]

Perhaps it's just me, but I I think that now discussion is mixing potatos and tomatos.

Klippel NFS gives 3D soundfield data that is same as anechoic free-field sound dispersion of the unit. That "raw data" is very good, reliable and informative per se. Directivity overlays and spectrograms are excellent also, reflecting anechoic free-field behaviour.

CTA-2034 is another thing, it has it's own rules and basically we see presentation for a normal forward-radiating small free-standing speaker (typical scenario) CTA gives specifications for measuring on/-inwall speaker like MKZM refers in post #145. The problems is that Amir didn't follow those instructions, so in this case CTA-2034 gets invalid data and gives irrelevant analysis.

As we have seen in many posts, room/spot measurements are fundamentally different from CTA prediction that Amir shows. This is very obvious and follows rules of physics and acoustics very well.

Sadly now again we are in a situation when Amir can't take any extra measurements, that would verify the issues in question. This is exactly why I'm not so enthusiastic about this website - guestions remain unanswered and are open to endless debate, misunderstandings and heated battles. But so it goes...

 

[Vladimir Filevski]

I concur that Klippel NFS gives accurate anechoic data in the whole audio spectrum, and CTA-2034 give good prediction of in-room behavior of conventional loudspeakers.
For dipole loudspeaker weighting should be a little bit different, so for now there might be some difference between CTA prediction and the real-world in-room measurements. But to state

... room/spot measurements are fundamentally different from CTA prediction that Amir shows.

is overblown.
True, predictions for on-wall loudspeakers measured so far here are not accurate in the low and mid-low region, because there was no panel imitating the wall. But we (should) know how to extrapolate the anechoic data in those cases, do we?

 

[Francis Vaughan]

But we (should) know how to extrapolate the anechoic data in those cases, do we?

Extrapolation from the spinorarma data is not possible as minimally we have lost phase information. The full holographic reconstruction of the sound field might contain enough information, but you would need to include the enclosure shape and apply modeling based on either CFD or BEM to understand how the radiation from the mid-bass is happening. The wavelengths cover a very wide range and the gap between the driver and the wall is too small to gain any accuracy otherwise. The sounds does not just reflect off the wall. Indeed at the wavelengths involved and the scales at play, reflection is simply impossible. We are in the domain of fluid dynamics.
It is impossible to separate the speaker from the wall. The wall is an intrinsic part of the acoustic design. It isn’t a conventional on-wall speaker with the mid-bass driver facing forward.

 

[MZKM]

Sadly now again we are in a situation when Amir can't take any extra measurements, that would verify the issues in question. This is eactly why I'm not so enthusiastic about this website - guestions remain unanswered and are open to endless debate, misunderstandings and heated battles. But so it goes...

The Klippel doesn’t have an on-wall measuring solution. So Amir would have had to jerry rig a drywall wall with studs on the rear and mounted the speaker on to that. Which he ain’t going to do.

 

[Juhazi]

The big picture is, that 90% of people reading/following or occasionally passing by ASR tests, don't understand the problemacy associated with loudspeaker measurements, and they look at only the first test post and perhaps 2-3 after that. This lack of in-room (installed per instructions) measurements applies to RS152 specially, small minimonitors and Magnepan. As well multiwoofer floorstanders' bass issue was solved just a week ago! Original tests and invalid results, including Preference Ratings are still out there...

To be fair, same problemacy appears at almost every review/test that is based on anechoic measurements, and even open field ground plane measurements. Room/boundary influence can be calculated to some extent, but with unconventional designs that guessing is too difficult. Best example is valueing bass reflex tuning over closed box subwoofers, and dipole speakers, corner horns, line arrays, on-wall speakers are problematic as well..

Loudspeaker quality assessment is very difficult, anechoic measurements are not the definitive measure by themselves! The reviewer should try to get deeper in the design and how it is supposed to be used. Established magazine reviews are golden (Stereophile, Soundstage, Stereo.de etc.) and some hobby reviewers like @hardisj and @napilopez do a really good job!

 

[amirm]

Established magazine reviews are golden (Stereophile, Soundstage, Stereo.de etc.) and some hobby reviewers like @hardisj and @napilopez do a really good job!

Stereophile? All of their bass measurements are exaggerated due to their near-field measurements.

Soundstage measurements are incomplete, low resolution and very hard to read.

And nobody remotely creates as much authoritative measurements as we do. I tested 120+ speakers last year alone. And you dare to compare what we do to the trickle that some of those rags put out with millions of ads?

If there is so much suffering going on for you in reading our reviews, perhaps you want to stop and go elsewhere.

 

[amirm]

You think you can even get other reviewers interest in testing stuff like this? Where were you before I tested this speaker? How much did you know then compared to now?

I am just amazed a the sense of entitlement some of you show who have not contributed one thing to the forum. I pay for the gear, do the testing, post the results, and a member shells out the money to send the gear, and you think it is your place to complain about it?

 

[Juhazi]

OK, I'll quit posting totally. I tried to delete my account, but seems like I'm not smart enough to do that.
Goodbye!

 

[Vladimir Filevski]

It isn’t a conventional on-wall speaker with the mid-bass driver facing forward.

I was referring to conventional on-wall loudspeakes, which Amir tested so far: Focal OV706V, OSD AP6CV, Polk Atrium 4, Revel M55XC, RSL Outsider II. All of them have mid-bass driver facing forward. Although only anechoic measurements were presented, we can easily extrapolate what will happen when they are mounted on-wall, like boosted lows and some comb-effects.
Infinity RS152 is very different, and I clearly acknowledged that in all of my posts in this thread.

Klippel NFS nearfield measurements and farfield computations are just fine, the problem is that some loudspeakers are designed to be mounted on the wall (or in the wall). There is no computation substitution for a real wall.
Klippel NFS has module with bafle for half-space measuring in-wall loudspeakers, and maybe it can be modified for measuring on-wall loudspeakers.

 

[More Dynamics Please]

So based on test results away from a wall on which it's designed to be mounted for optimum performance @amirm concluded that he didn't know if this design is ever going to work and opened up discussion for everyone to analyze the results and tell him their opinions.

Conclusions
I have none to offer other than cringing at the sight of two tweeters in opposing fashion trying to produce sound. And a woofer in the back which is also supposed to do midrange duty. Is this ever going to work? I don't know. You analyze it and tell me!

 

[Chromatischism]

I think what Amir does for the site and the hobby is tremendous. Let's not diminish that. I certainly wouldn't want Amir to start avoiding potentially controversial products and just stick with "safe to review".

I think there's value in seeing the differences between Klippel vs installed measurements for a speaker design like this. We have enough data to conclude that the installed speaker has the same general trend as the anechoic, which is what we would expect. The discussion though should be about why the irregularities in the Klippel measurements appear to smooth out to a degree once the speaker is installed. Clearly there were design decisions done at Harman that can explain this.