Elac BS U5 Slim 3-way Coaxial Speaker Review

[Shazb0t]

Could the distortion issues be mitigated by crossover modifications? @Dennis Murphy

 

[m8o]

For a second time we see that the magic of coaxial drivers tends to be more for the eye than the ear.

What my Tannoy System 8 NFM II has taught me is ...

What they are best at is a *uniform* tonality in the mixer's seating position as he/she moves their head forward/back, left/right, up/down (within reasonable angular limit). No changes of tonality or phasing (crossover lobing), and other such auditory artifacts created from crossover interaction with both drivers, their phase, and physical offsets.

Producing a more *uniform* sound reproduction in a wider 3-D space. *Not* necessarily a more perfect one.

...thought I've tried to explain this before. Maybe it was another forum or FB. Dunno. I try to explain it often.

 

[BN1]

I am hoping that Elac plans an update to the Uni-Fi speaker line. As the Debut line has moved more upscale in performance and price the Uni-Fi line seems to be squeezed out in their model lineup. I like Elac's approach to the Debut of using decent components and improving their peformance in steps. Hopefully, the Uni-Fi's follow this improvement process in months to come.

 

[KaiserSoze]

Could the distortion issues be mitigated by crossover modifications? @Dennis Murphy

It is peculiar the way those two prominent distortion peaks occur exactly on the crossover points. One thing that this causes me to wonder about is whether the woofer is even isolated within the cabinet from the midrange-tweeter. And it would be nice to understand more about how the distortion measurements are taken because on the face of it, you can hardly avoid thinking that this is actually sympathetic resonance between coupled, underdamped drivers. But if that were the case, you would expect to see peaks in the response curves, especially the lower one, where the distortion shows up over a frequency range much too wide to be obscured by smoothing in the frequency response measurements. It would thus seem all but certain that it really is distortion due to over-driving the smaller of the two drivers at each crossover point, and that it appears as a peak because with frequency below the crossover point the excursion of the small driver decreases rapidly with decreasing frequency. It thus seems likely that substantial improvement would be achieved by moving the crossover points higher, from 270 Hz to 500 Hz and from 2.7 kHz to 3 kHz. But then you have to wonder why Elac didn't notice these distortion peaks and didn't bother to adjust the crossover points.

It would be interesting to learn how good or bad it would turn out if for the lower crossover the components were all adjusted such that the same impedance effects occur at twice greater frequency, and similar for the upper crossover except for the components to have the same impedance effects at frequency 15% higher than the original frequency.

 

[Dennis Murphy]

Could the distortion issues be mitigated by crossover modifications? @Dennis Murphy

Something is certainly off. If these things had deeper bass response, I might take a crack at it. But with the current woofer and cabinet, it just seems like an overly complex speaker.

 

[KaiserSoze]

Something is certainly off. If these things had deeper bass response, I might take a crack at it. But with the current woofer and cabinet, it just seems like an overly complex speaker.

I may be in the minority by caring about the polar response in the vertical plane, but it has always bothered me that the sound you get from a speaker depends on where you stand or sit when listening to it. I recall thinking that this isn't the way it should be even when I was very young and listening to a big old console phonograph player. I like the concentric driver approach for fundamental reasons, and find it unfortunate that this speaker came very close to being a great speaker but then fell short of the mark because the designers chose the crossover points without first studying the distortion characteristics of the drivers. It appears that KEF is the only manufacturer that takes concentric drivers seriously enough to produce a truly serious product. But maybe Elac will approach it more seriously the next time around. If they want to compete with KEF in this space they'll have to.

 

[Shazb0t]

I would be interested if someone was willing to tinker with the crossover points to bring the distortion peaks down. Aside firm these distortion peaks, I think it's fair to say that the measurements for this speaker are actually quite good. Especially considering that they can be had for ~$300 a pair for the vinyl wrap model.

 

[JW001]

One thing that this causes me to wonder about is whether the woofer is even isolated within the cabinet from the midrange-tweeter...

https://www.lowbeats.de/site/wp-content/uploads/2016/11/2016-11-Elac-UniFi-B5-Koax-Housing.jpg

 

[KaiserSoze]

https://www.lowbeats.de/site/wp-content/uploads/2016/11/2016-11-Elac-UniFi-B5-Koax-Housing.jpg

Thank you, but after I thought a few seconds about it, I mentioned that if those peaks were due to sympathetic resonance of undamped drivers, the peaks would show up in the routine frequency response plots. It is evidently just plain old distortion due to the driver excursion becoming too great at the low end of the respective ranges. Maybe the designer said to himself, "If it's good enough for the woofer, it's good enough for the midrange and the tweeter!" The only reason I can think of, for why they didn't choose higher crossover points, is that it would have put some hills and valleys in the off-axis response, and they wanted the off-axis response to be as smooth as possible. Other than that, I can't think of any reason why they wouldn't have moved the crossover points higher, especially the tweeter, which doesn't need to move very far.

 

[Shazb0t]

I think this is an interesting speaker to see if these kind of distortion peaks are really audible to most people! The PIR and dispersion of these speakers is rather good with just the distortion peak flaws.

 

[KaiserSoze]

I think this is an interesting speaker to see if these kind of distortion peaks are really audible to most people! The PIR and dispersion of these speakers is rather good with just the distortion peak flaws.

This isn't an easy sort of test to do. You can't just increase the volume and ask the listener to yell, "Stop!" when they hear distortion. You can conduct an informal test to see loud a given speaker will play before you start to hear distortion. I believe it is useful to do that sort of informal, personal test, but you can't assume that other people will get the same result. To test the audibility of distortion of this particular speaker in a controlled, scientific way, you would need to introduce the distortion in a controlled way, into a speaker that is known to be essentially distortion-free. The distortion you introduce would need to match the measured distortion of this speaker in most every conceivable way. And there are other difficulties, for example, the volume of the erstwhile clean speaker would need to match the volume, of the speaker being evaluated, at which the measured level of distortion occurs. All in all it is apparent that this is a very tricky and difficult thing to do. As such it might be best to just experiment and learn for yourself how your sensitivity to distortion compares to the published tables, and not be concerned when other people make statements about what level of distortion is or isn't audible.

 

[Sergei]

The only reason I can think of, for why they didn't choose higher crossover points, is that it would have put some hills and valleys in the off-axis response, and they wanted the off-axis response to be as smooth as possible. Other than that, I can't think of any reason why they wouldn't have moved the crossover points higher, especially the tweeter, which doesn't need to move very far.

According to http://noaudiophile.com/ELAC_UB5/, "The crossover looks like 3rd order 18db/octave crossover all around." If that's correct, the required tolerances for the crossover and transducers components are bound to be tighter than for a second-order one.

I believe KEF LS50 uses a simple phase-reversing second-order crossover, which is less sensitive to components deviations. Yet, obviously, in this case each transducer has to be able to cover a wider frequency band without distorting too much.

So, perhaps Andrew Jones went for a more sophisticated crossover, which in theory would narrow the bands that the transducers need to cover without distortion? Yet the realities of manufacturing and QC doomed this design in practice?

This hypothesis could explain why the demo speakers sounded better (they used accurately built crossovers), and why the two production speakers, which were supposed to be identical, exhibited different distortion profiles.

 

[KaiserSoze]

According to http://noaudiophile.com/ELAC_UB5/, "The crossover looks like 3rd order 18db/octave crossover all around." If that's correct, the required tolerances for the crossover and transducers components are bound to be tighter than for a second-order one.

I believe KEF LS50 uses a simple phase-reversing second-order crossover, which is less sensitive to components deviations. Yet, obviously, in this case each transducer has to be able to cover a wider frequency band without distorting too much.

So, perhaps Andrew Jones went for a more sophisticated crossover, which in theory would narrow the bands that the transducers need to cover without distortion? Yet the realities of manufacturing and QC doomed this design in practice?

This hypothesis could explain why the demo speakers sounded better (they used accurately built crossovers), and why the two production speakers, which were supposed to be identical, exhibited different distortion profiles.

I suppose it is that in order to avoid an unintended notch or peak at the crossover point, the indicated tolerances are tighter for steep slopes, compared to slopes that are less steep. But I'm not sure how this relates to the issue here, unless the hypothesis is that the slopes ended up being less steep than intended and that this is the reason why the tweeter is working harder than intended in the vicinity of its crossover to the midrange, and why the midrange is working harder than intended in the vicinity of its crossover to the woofer. Certainly this hypothesis is not unreasonable, however the peaks in the distortion plots indicate that the actual crossover points are right where they were intended to be, at 270 Hz and 2.7 kHz. The question that thus wants to be considered is the likelihood that the components would be far enough out of tolerance to cause the slopes to be less steep than intended, without also causing the crossover points to be lower than intended. I'm not at all sure, but my gut (which is often unreliable) is that if the components were so sloppy in tolerance as to cause the slopes to be more shallow than intended, that there would also likely be an evident shift in the crossover point. But there's just no way to know. This would be an easier nut to crack if the specs had indicated that the upper crossover point is supposed to be at 3 kHz and the test data gave strong evidence that the actual point is 2.7 kHz.

In any case, I am still of the opinion that the potential of this coaxial speaker is great enough such that if I were as proficient at crossover tweaking as some people here are, I would be inclined to take this on as a project.

 

[KaiserSoze]

I just happened to notice a pair of slightly used UB5 speakers for sale on US Audio Mart, for $300 the pair. The seller is located in Reno NV. For anyone with any interest in modifying the crossover to mitigate the distortion, this might be an opportunity. Note please that I've never bought anything through this site and I have no idea what the risks might be.

 

[rebbiputzmaker]

I suppose it is that in order to avoid an unintended notch or peak at the crossover point, the indicated tolerances are tighter for steep slopes, compared to slopes that are less steep. But I'm not sure how this relates to the issue here, unless the hypothesis is that the slopes ended up being less steep than intended and that this is the reason why the tweeter is working harder than intended in the vicinity of its crossover to the midrange, and why the midrange is working harder than intended in the vicinity of its crossover to the woofer. Certainly this hypothesis is not unreasonable, however the peaks in the distortion plots indicate that the actual crossover points are right where they were intended to be, at 270 Hz and 2.7 kHz. The question that thus wants to be considered is the likelihood that the components would be far enough out of tolerance to cause the slopes to be less steep than intended, without also causing the crossover points to be lower than intended. I'm not at all sure, but my gut (which is often unreliable) is that if the components were so sloppy in tolerance as to cause the slopes to be more shallow than intended, that there would also likely be an evident shift in the crossover point. But there's just no way to know. This would be an easier nut to crack if the specs had indicated that the upper crossover point is supposed to be at 3 kHz and the test data gave strong evidence that the actual point is 2.7 kHz.

In any case, I am still of the opinion that the potential of this coaxial speaker is great enough such that if I were as proficient at crossover tweaking as some people here are, I would be inclined to take this on as a project.

Curious, have you listened to these speakers? Is this distortion audible, does it actually exist off axis at a normal listening position? I am thinking that the results might differ when used as intended. ymmv

 

[KaiserSoze]

Curious, have you listened to these speakers? Is this distortion audible, does it actually exist off axis at a normal listening position? I am thinking that the results might differ when used as intended. ymmv

A fundamental assumption that underlies most of what is done on this site is that the only reliable way to assess the sound quality of a speaker is by taking objective measurements. This is the only sort of speaker evaluation that is reproducible and that accommodates corroboration of findings by multiple, independent evaluators.

The question of audibility of distortion is inherently difficult for several reasons to include the fact that the level of distortion depends on how loudly the speakers are being played. But the measurements that Amir took clearly indicate a level of distortion that is appreciably greater than what is typical for loudspeakers generally, and easily above the threshold that is generally considered audible and significant. At 96 dB and 1 m distance, THD as measured by Amir is greater than 5% at the lower crossover point (270 Hz) and greater than 3% at the higher crossover point (2.7 kHz). There is another peak at about 8 kHz which is similarly greater than 3%. Did you look at these measurements, at the start of this thread, before you made up your mind to challenge the idea that the level of distortion is problematic with this speaker?

My personal contribution to this thread has been fairly limited, mostly to sharing my opinion that there is a lot to like about this speaker and that as such it might be worth someone's trouble to modify the crossover such that the crossover points are moved higher in frequency. As such, I doubt whether my posts in this thread were the most appropriate posts for you to have attached a comment questioning the audibility of distortion at the levels observed in this speaker.

For what it's worth, the reason my response to you is blunt is that I did not like your having written, "...the results might differ when used as intended.." This is especially irksome for me, because it implies that anyone who prefers to take an objective, measurement-based approach to speaker evaluation is pursuing a Quixotic quest, because this isn't the way the speaker is supposed to be used. This runs strongly contrary to the fundamental idea that the right way to assess the sound quality of a loudspeaker is to take objective measurements. When a car is placed on a dynamometer to measure the engine performance, is this a flawed endeavor since it isn't how the car was intended to be used? When a video monitor is subjected to controlled measurements to assess black level, dynamic range in luminosity, accuracy of color rendering, etc., is this a flawed endeavor since this isn't how the monitor was intended to be used? Are you suggesting that loudspeakers are different in this respect, that for some reason the approach of objective measurement is not valid when it comes to loudspeakers? What exactly are you saying? That measured distortion isn't meaningful unless it is confirmed audible by listening for it when seated at a "normal listening position"? How do you define "normal listening position"? How loudly should the speakers need to be played? Should double-blind methodology be followed? Are you sure that it is not sufficient to observe that the measured levels at a standard volume level are well above the levels that have been determined, through a considerable amount of controlled tests including some you can do on line, to be audible? Why exactly wouldn't this be sufficient?

 

[rebbiputzmaker]

A fundamental assumption that underlies most of what is done on this site is that the only reliable way to assess the sound quality of a speaker is by taking objective measurements. This is the only sort of speaker evaluation that is reproducible and that accommodates corroboration of findings by multiple, independent evaluators.

The question of audibility of distortion is inherently difficult for several reasons to include the fact that the level of distortion depends on how loudly the speakers are being played. But the measurements that Amir took clearly indicate a level of distortion that is appreciably greater than what is typical for loudspeakers generally, and easily above the threshold that is generally considered audible and significant. At 96 dB and 1 m distance, THD as measured by Amir is greater than 5% at the lower crossover point (270 Hz) and greater than 3% at the higher crossover point (2.7 kHz). There is another peak at about 8 kHz which is similarly greater than 3%. Did you look at these measurements, at the start of this thread, before you made up your mind to challenge the idea that the level of distortion is problematic with this speaker?

My personal contribution to this thread has been fairly limited, mostly to sharing my opinion that there is a lot to like about this speaker and that as such it might be worth someone's trouble to modify the crossover such that the crossover points are moved higher in frequency. As such, I doubt whether my posts in this thread were the most appropriate posts for you to have attached a comment questioning the audibility of distortion at the levels observed in this speaker.

For what it's worth, the reason my response to you is blunt is that I did not like your having written, "...the results might differ when used as intended.." This is especially irksome for me, because it implies that anyone who prefers to take an objective, measurement-based approach to speaker evaluation is pursuing a Quixotic quest, because this isn't the way the speaker is supposed to be used. This runs strongly contrary to the fundamental idea that the right way to assess the sound quality of a loudspeaker is to take objective measurements. When a car is placed on a dynamometer to measure the engine performance, is this a flawed endeavor since it isn't how the car was intended to be used? When a video monitor is subjected to controlled measurements to assess black level, dynamic range in luminosity, accuracy of color rendering, etc., is this a flawed endeavor since this isn't how the monitor was intended to be used? Are you suggesting that loudspeakers are different in this respect, that for some reason the approach of objective measurement is not valid when it comes to loudspeakers? What exactly are you saying? That measured distortion isn't meaningful unless it is confirmed audible by listening for it when seated at a "normal listening position"? How do you define "normal listening position"? How loudly should the speakers need to be played? Should double-blind methodology be followed? Are you sure that it is not sufficient to observe that the measured levels at a standard volume level are well above the levels that have been determined, through a considerable amount of controlled tests including some you can do on line, to be audible? Why exactly wouldn't this be sufficient?

My question was more general not personally directed at you please don’t take it that way. I did pick up a pair, open box, the regular cabinet for $250. They do have some issues but they are much better than this review might make people believe. I think the issues that these measurements point out are not actually noticeable when listening or running sweeps. IMO. (they do have limited headroom though) If the distortions noted in the review were really an issue they would’ve been addressed by the designer, not overlooked. IMO... A normal listening position, a stereo pair and listening off axis at a reasonable distance, does make a difference imo. Ymmv

 

[BN1]

Anxious to see how the new Elac Uni-Fi 2.0 stacks up against this older review. Looks like the speaker has had a complete redesign with new cabinet, new drivers, and new cross-over. Some preliminary reviews have indicated a flatter in-room response with a slightly raised treble compared to the older version. I would assume that Elac would not release a new speaker unless it was somewhat better than the speaker that it replaces.

 

[Shazb0t]

Anxious to see how the new Elac Uni-Fi 2.0 stacks up against this older review. Looks like the speaker has had a complete redesign with new cabinet, new drivers, and new cross-over. Some preliminary reviews have indicated a flatter in-room response with a slightly raised treble compared to the older version. I would assume that Elac would not release a new speaker unless it was somewhat better than the speaker that it replaces.

Curious as well! The PIR for the original shows the tonality is around where it should be without requiring a boosted treble response. All Elac really needed to do was tame the distortion spikes at the crossover points.

 

[981CS]

I would assume that Elac would not release a new speaker unless it was somewhat better than the speaker that it replaces.

It's Elac. There's *always* a new speaker from them...for better or worse.

They are in the business to sell speakers (and pay for all that AJ marketing). They can't sit around on years old designs that people have already formed an opinion about and aren't really buying anymore (unless they go on a big sale, like has been happening quite often here lately).