An attemp to make subjective and objective analysis of Bowers&Wilkins 685 S1 bookshelf speakers

[ReDFoX]

Hello ASR, this is my first attempt posting something here, so I'm sorry in advance for mistakes that I could made. English isn't my primary language so I'll try keeping things as clear as possible.
I've been in "read-only mode" and recenly realised that I gained more knowledge over this year than in 5 previous years when I was usually watching YT and visiting "ordinary" non-english speaking forums and sites. Therefore, about 3 month ago I felt confident in myself enough to actually try and make quasi-anechoic measurements in ordinary flat. My first 2 test subjects were JBL 4312mkii and Pioneer CS-66G. While I don't think that those measurements were properly made and processed, I compared them to existing data from manufaturers and it showed a surprising amount of "truth" in my homemade stuff. So, here it goes...

Spoiler: Summary

This is one of those goofy ahh B&W speakers

Subjective part
A friend of mine gave me those 685s for sell, so took the opportunity to measure them. I'm kinda biased towards B&W in a negative way due to brand's sound, however I tried to give them a chance. Well, their interior looks really cool and reminds me of my pair of KRK RP7G4, unfortunately B&Ws have straight cut corners with no rounding at all which leads to higher edge diffraction. In general, the design is quite typical for two-way systems: a conical LF/MF dynamic driver (with a surprisingly soft rubber suspension), a dome tweeter without a waveguide (another dissapointment), and, of course, a bass-reflex design with frontal port. I should also note the unusual chamber behind the HF driver, the fixed “bullet” on LF/MF driver and indentations at the inlet of the port, which, apparently, should reduce the turbulence.
I started with the most difficult thing for these bookshelves - a direct comparison with my 4" monitros probably from some German brand. Aaand, 685 sucked. The difference was so huge I started to wonder if I applied wrong eq to them somehow. Worth mentioning, I have a very small room with RT of around 200-250ms and room modes all the way up to ~400Hz (this 2 factors will become crucial later). My monitors were tuned to harman without bass shelf and both the highs and lows on 685 were unrealistically boosted. On the one hand, this created the impression of sound “in the face”, as I like, on the other hand, you immediately understand why the vocal (for example) jumps out and transients are emphasized, and you come to the realization that there's something fundamentally wrong with the sound.
Here's what I wrote during the first audition: "mumbling, no mid frequencies, chymous top.". I tried to deviate from the axis and it seemed to me that part of the spectrum of the upper mid was damped to a lesser extent, which could indicate problems with directivity (and we'll see this later).
In addition, there was a feeling that the port ofbass-reflex system was incorrectly tuned - like too high, in a region where the natural roll-off of LF/MF driver hasn't started yet. Anyway, for some reason the "speed" of bass on these is just "non-existent".
Despite all this, it is worth mentioning the excellent power handling of those: I did not manage to achieve audible degradation even when used one 685 in mono with low crest factor musical material.
In general, I tested the 685 with a few more tracks that I know well, and compared to those unknown 4" gray monitors and noticed that 685s have a much worse separation of sound layers and, in general, the picture is kind of “pushed in”. Whether this is due to the fact that the comparison was made with monitors with a linear phase response, or is it purely frequency dependent stuff that our brain throws at us, honestly, I don’t know, but the images separation is much worse: in the situation where I "see" 3 layers on monitors, on 685s, the first 2 simply “stick together” and the vocals, for example, simply jump out at the listener at certain moments.
All in all, 685 gave me an impression of speaker that compansates for its owner age-dependent hearing loss. Unfortunately for manufacturer, I'm 19 and the amount of highs that this thing produced in nearfiled was enough to cause an ear fatigue, even for such "straight" room curves lover as me.

Spoiler: some pics of them that I made

Objective part
I hope that someone more knowledgable will correct mistakes (if I've done such) in measurements processing since I still has no concrete proof of their reliability.

Spoiler: full PB-REC chain

Native Instruments Komplete Audio 6 mkii -> Yamaha A2000A -> B&W 685 S1 -> Sonarworks Xref 20 -> Native Instruments Komplete Audio 6 mkii

I took a total of 17 measurements, I guess.
The distance was ~70cm for vertical and horizontal measurements. The room I measured them in is very reflective and on a smaller scale of overall size. That lead to a gate time of 3.5ms. This is criminally small, bit unfortunately I have no opportunity to measure something outside since I live in a block of flats near the city centre. SPL numbers are kinda meaningless since I don't have any spl meters nor usb microphones with SPL calibration. An app on my phone showed levels around 70-80 dBC, but I have no trust in its measurements.

Uh-oh, that's one scary graph. Not only it shows a V-shaped FR, but also tells us that this is unEQuable speaker. Why? Look at the 5-9K region, where tweeter basically radiating in 180 degrees and I heard that in sighted test. It means that we can't pull this region on on-axis FR, because we'll create a peak in ER spectrum. This leads to high room-dependency of the speaker (remember I told that my listening room has RT of about 250ms?).

Looking at polar plots we can clearly see a directivity mismatch and that strange looking resonance. This is the speaker I'd give for those who believe that you can eq everything. No, you can not. I've tried making PEQ filters for this thing, but that HF bell would be impossible to nail since all rooms have different amount of ER so the perception of that region would highly depend on amount of reflection/absorbtion.

After deriving biquad coefficients, I've loaded filters into my DDRC-24 and gave 685s a last chance. It's important, that I listened to them in a different room than in my first test and their tonality without PEQ has changed! I heard that HF peak as kind of a high-shelf boost due to huge aount of undapmpened early reflections. After filters appliction I was both releaved and saddened: my filters did a great job of bringing back those scooped mids, but it was something in highs that I can't describe that still gave me that impression of inflated upper mids.
Maybe I'm doing something wrong, so feel to poke me in any msitakes or inaccuracies that I could possibly made.
This is a comparison between 3 versions that I recorded ~1m in listening position with an absorbtion panel behind. This is for purely entertainment reasons because omnidirectional microphone != our ears+brain

I hope that it will help somebody one day and stop them from buying such things. Such situation and speaker are another confirmation of the need for publicly available measurements
Measurements + graph data

 

[thewas]

Looking at polar plots we can clearly see a directivity mismatch and that strange looking resonance. This is the speaker I'd give for those who believe that you can eq everything. No, you can not. I've tried making PEQ filters for this rhing, but that HF bell would be impossible to nail since all rooms have different amount of ER so the perception of that region would highly depend on amount of reflection/absorbtion.

To quote Floyd Toole:

"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."

Source: https://www.audiosciencereview.com/...ut-room-curve-targets-room-eq-and-more.10950/

 

[ReDFoX]

To quote Floyd Toole:

"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."

Source: https://www.audiosciencereview.com/...ut-room-curve-targets-room-eq-and-more.10950/

And this is exatly the reason I called it "unEQable". Idk if this word can be even used in such context, but I wanted to emphasize that no amount of equalization will fix this particular model. IMO, Toole's "Sound Reproduction" and presentation at CIRMMT are the best introduction to loudspeaker/room side of sound reproduction

 

[fineMen]

Grant it a second thought. With speakers that shiny in visual appearence it is quite unlikely to see them used by stereo-lovers. Stereo, as an effective addition to the general sound reproduction, isn't of prime importance for many. I speculate even, that only a very tiny minority actually cares. Stereo is a secondary layer, and that was it. Except for all unsolved problems and nasty nasty complications in the production process in the studio.

That said, people, nearly all, don't listen "on-axis". For the general public the overall sound, reverberation included, counts. The room-filling qualities. Listening even from another room upstairs while dancing with the vacuumcleaner. Subdued mids are easy to the ear. When listening "on-axis" as sort of a self-confirming act, the accentuation of certain frequency registers makes a difference, only to indicate that this was, in case of a correct discipline, the right thing to do. Until some weeks later.

I don't like this attitude, because it is a fraud. "The stereo" is a fraud. Speakers like this exploit it. But tell people, and they say they were not willing to miss something out. In this regard the common man isn't that different from an audiophile.

 

[Nivag123]

Hope links are allowed. A pretty extensive review of the existing response, and a few levels of tweaks to the crossover for various measured improvements.

www.audioexcite.com » Bowers & Wilkins 685 review part 1

 

[thewas]

Thanks for sharing, very nice article which shows that very good drivers for the price class are unnecessarily compromised by a too minimalistic crossover.

 

[Music1969]

And this is exatly the reason I called it "unEQable". Idk if this word can be even used in such context, but I wanted to emphasize that no amount of equalization will fix this particular model. IMO, Toole's "Sound Reproduction" and presentation at CIRMMT are the best introduction to loudspeaker/room side of sound reproduction

Even if used nearfield , not worth EQ'ing at all? Won't improve the sound quality if you EQ correctly?

 

[thewas]

Even if used nearfield , not worth EQ'ing at all? Won't improve the sound quality if you EQ correctly?

In nearfield use in my experience equalising it based on the anechoic LW would improve it.

 

[Music1969]

In nearfield use in my experience equalising it based on the anechoic LW would improve it.

I would absolutely think so too

 

[ReDFoX]

Hope links are allowed. A pretty extensive review of the existing response, and a few levels of tweaks to the crossover for various measured improvements.

www.audioexcite.com » Bowers & Wilkins 685 review part 1

Yeah, discovered that review right after I finished writing my subjective part. Still, it's interesting how 2 different people's opinions about "out of the box" sound quality are pretty much identical

Even if used nearfield , not worth EQ'ing at all? Won't improve the sound quality if you EQ correctly?

As I said, PEQing definitely helps 685, but that upper-mid "anomaly" can't be properly EQed which ultimately leads to the feelling that something is not right with the sound. I haven't tested EQed version in my 200ms RT room, but even at 1.5m with RT of ~500ms I could hear that off-axis behaviour and this is one the reasons I recorded that comparison video

 

[Penelinfi]

It depends on the model, but many B@W seem to play the woofer quite high in frequency, making use of the controlled breakup of the cone.

So you could be hearing the breakup, even if it's not a big peak, there would be surface/material reflections.

You could also be hearing the higher frequencies interfering with the tweeter in a mish mash of directivity and phase issues.

I don't tend to listen up loud, so B&W has generally provided an interesting sound that lends out its shortcomings over time.

 

[Ron Texas]

To quote Floyd Toole:

"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."

Source: https://www.audiosciencereview.com/...ut-room-curve-targets-room-eq-and-more.10950/

This is one reason why people should not buy speakers on preference score alone. There are speakers available which have excellent dispersion but their preference score is lowered by easily fixable peaks in their frequency response. Dips are not so easy to fix. One other thing to consider is Toole weighted each octave equally. There is going to be an occasional speaker with a good preference score but it is off in an octave which makes it sound unsatisfying. The best example I can think of is the review of the KEF R3 here on ASR.

 

[Curvature]

Consider getting better loudspeakers.

So polite it cuts to the bone.

 

[thewas]

This is one reason why people should not buy speakers on preference score alone. There are speakers available which have excellent dispersion but their preference score is lowered by easily fixable peaks in their frequency response. Dips are not so easy to fix. One other thing to consider is Toole weighted each octave equally. There is going to be an occasional speaker with a good preference score but it is off in an octave which makes it sound unsatisfying. The best example I can think of is the review of the KEF R3 here on ASR.

Fully agree, I own both the original LS50 and the LS50 Meta since each model was released and the first one can be quite transformed with EQ due to its decent directivity (and both can be improved even more with subs, which can also be seen on the corresponding scores).

 

[ReDFoX]

This is one reason why people should not buy speakers on preference score alone. There are speakers available which have excellent dispersion but their preference score is lowered by easily fixable peaks in their frequency response. Dips are not so easy to fix. One other thing to consider is Toole weighted each octave equally. There is going to be an occasional speaker with a good preference score but it is off in an octave which makes it sound unsatisfying. The best example I can think of is the review of the KEF R3 here on ASR.

100% agree. While it may be useful to have a single metric of "quality" (like SINAD) for comparison graphs, but it's extremely misleading for begginners and those, who have little knowledge about what people can hear an how
They'd think that an 50$ Edifier with a tonality score of 2.1 or a 70dB SINAD amp will sound like trash, but they wouldn't. I heard 0 difference in my HD650s between a dac built in motherboard, Komplete Audio 6 mkii and E-MU 1616, but there's definitely a huge gap between them, in terms of "raw" performance. Moreover, I had a pair of Edifier R980T for almost 2 years before switched to RP7G4 (later KH80) and I enjoyed listening to music through 980s as much as I do now, driving KH80s

 

[ferrellms]

This is one reason why people should not buy speakers on preference score alone. There are speakers available which have excellent dispersion but their preference score is lowered by easily fixable peaks in their frequency response. Dips are not so easy to fix. One other thing to consider is Toole weighted each octave equally. There is going to be an occasional speaker with a good preference score but it is off in an octave which makes it sound unsatisfying. The best example I can think of is the review of the KEF R3 here on ASR.

So, the objective and empirically derived preference score does not always correlate with your personal subjective listening preference? This is an argument against people using it for choosing speakers? Can you back up the claim "There are speakers available which have excellent dispersion but their preference score is lowered by easily fixable peaks in their frequency response" with real evidence?

Preference score is the best place to start for choosing speakers.

 

[Ron Texas]

So, the objective and empirically derived preference score does not always correlate with your personal subjective listening preference? This is an argument against people using it for choosing speakers? Can you back up the claim "There are speakers available which have excellent dispersion but their preference score is lowered by easily fixable peaks in their frequency response" with real evidence?

Preference score is the best place to start for choosing speakers.

Preference score is a good place to start, but budget and dynamic range are also factors. What seems to be going around here is a lot of people put the preference score on a pedestal and conclude they need Genelec speakers. As to the "claim" there are lots of forum members who have said the same and it's a logical extension of what Toole already proved. If you don't like it, ignore it.

I gave the R3 test here as an example of a speaker with a great preference score which does not do it for our host. The JBL 708P is an example of an active speaker with an unimpressive preference score which our host loved. Remember, this unit has built in EQ and its preference score could be raised with a bit of tweaking. It can also play loud.