Revel F328Be Speaker Review

[Stephen]

Luckily, I downloaded it.

See attached pdf or view photos below. Cover and last page are not included.


View attachment 93326View attachment 93327View attachment 93328View attachment 93329View attachment 93330View attachment 93331

Harman should also present the audiogram (hearing test) of the 9 listeners. something that looks like that:

Otherwise everyone hearing is different, and we all have preferencies. You can always average the global taste of a group but there is absolutly no guaranty this will suits you. Unless you persuade yourself to like it. That happens all the time when someone invest in an extremely expensive system. And the more expensive it is, the more persuative it is... this is only psychology.

 

[andreasmaaan]

On a completely unrelated note to the discussion thus far, I think it's quite obvious that the older aluminium dome used in the non-Be range of Revel Performa speakers is significantly cleaner than the beryllium dome used in this speaker (which FWIW appears not to differ too much from the Be tweeter used in the F226Be etc).

Note in particular the levels of H3 (bright blue) and H5 (light green) between 2kHz and 4kHz.

And I can't say it's at all obvious to me how the tweeter/waveguide is a significant improvement in terms of dispersion vs the older (Al-based) models. Perhaps slightly wider dispersion in the 9kHz-11kHz range?

F208:

F328Be:

 

[BYRTT]

.....@BYRTT , my in room measured response is after I'd done an Anechoic EQ on the speaker using Amir's measurement data (of Listening Window) if that makes any difference to your observations at all.....

Love you active research if theres a chance mirror of Amir's curves looks work okay onto other samples imagine such trials should be good education and if measurements plus ears tell its not far cant imagine should make difference to observations..

.....(I see your point BYRTT re sharp roll off of bass on the JBL 308 which explains why you can't get anywhere near flat to 20Hz with any room gain).

Yes its a deep crater to fill up when slope is so steep and boundary gain often mostly push up a spike on the high Q knee seen in below example, from upper row left side its Amir's raw analyze of 309P MKII / F328Be / C52 and lower row is same but added a exactly same linkwitz transform that happen bring C52 into a perfect 2nd order slope @20Hz, notice scales look 70dB deep to get the picture down there..

(@BYRTT , I just noticed you traced my Predicted Response after RoomEQ into your graph rather than the actual measurement, the actual measurement is the non-highlighted curve)

Ahh my mistake but in it looks mostly a high Q 3dB @50,7 & 5dB @155 difference think theres no need to re model..

 

[pozz]

Nah that just some made up average like everything else. Some people hear 22kHz some 20kHz. Some hear 20Hz others 16Hz. There is a spectrum. Not being able to hear a note below 20Hz does not mean that you can't feel it. Most people have never experienced a sound system that can give you the proper force needed for <20Hz but the difference between 20Hz and 10Hz is real.

"Under ideal laboratory conditions, humans can hear sound as low as 12 Hz[11] and as high as 28 kHz"
Olson, Harry F. (1967). Music, Physics and Engineering. Dover Publications. p. 249. ISBN 0-486-21769-8. Under very favorable conditions most individuals can obtain tonal characteristics as low as 12 cycles.

There's a good, but rare book on infrasound written in the 1970s that I've been trying to get: https://www.worldcat.org/title/infrasound-and-low-frequency-vibration/oclc/3169375

The claims I've read are that you can still sense ("hear" is not quite the right word) very, very low frequency vibrations (sub 10Hz) at very high SPLs (>130dB). No direct research that I could quote yet though.

 

[aHaidc]

Luckily, I downloaded it.

See attached pdf or view photos below. Cover and last page are not included.


View attachment 93326View attachment 93327View attachment 93328View attachment 93329View attachment 93330View attachment 93331

Magico A3 are not good enough speakers for the test, and I think the A3s would be even worst on ML amp, too 'dark'. The Revels are sure to have better synergy with ML.

 

[aHaidc]

@amirm Would you measure the Revel Salon 2?

 

[Steve Dallas]

@amirm Would you measure the Revel Salon 2?

He answered this a few pages back. His Salon 2s are upstairs in a loft and very heavy. His NFS is downstairs in the garage. His stairs are fine wood. He has a bad back. He has no way to very carefully move one down there without help, which is not available during the pandemic.

 

[phoenixdogfan]

He answered this a few pages back. His Salon 2s are upstairs in a loft and very heavy. His NFS is downstairs in the garage. His stairs are fine wood. He has a bad back. He has no way to very carefully move one down there without help, which is not available during the pandemic.

He'll get it done in his own time. I think the really interesting question here is how the wider dispersion designs like the Revel stack up against the cardoid narrower dispersion designs like the D&D 8cs, Kii 3s, and GGNTKT M1s. Would love to see one or more of these designs make it onto a real Klippel. Understand we may not have too much longer to wait if hardisj finds a more elegant way to control digital volume than he currently has in his testing rig. If anyone has something like a miniDSP SHD Studio they can lend out for a few weeks please help the man out!

 

[hardisj]

Understand we may not have too much longer to wait if hardisj finds a more elegant way to control digital volume than he currently has in his testing rig. If anyone has something like a miniDSP SHD Studio they can lend out for a few weeks please help the man out!

FWIW, testing the 8C is no problem (other than maybe the verticals thanks to the cardioid cancellation vent). I was just looking for a more elegant setup for my listening tests. But I solved that. Been demoing the speakers all week. Plan to start testing them soon. I'm ready to go right now but I am holding off because I might be changing my measurement rig a bit and don't want to change in the middle of taking measurements. That, and finding the time between my work schedule and good weather.

 

[Robbo99999]

Love you active research if theres a chance mirror of Amir's curves looks work okay onto other samples imagine such trials should be good education and if measurements plus ears tell its not far cant imagine should make difference to observations..

Hi BYRTT, not sure what you mean here, are you saying it would be interesting to see if my Anechoic EQ using Amir's data of the JBL 308p Mkii would improve other samples of 308p and produce equally close in-room measurements? You mean it would be interesting to see if there's much production variance from one 308p to the next? Subjectively I'd say my Anechoic EQ is an improvement, and particularly with female voices, that's what I've noticed the most (perhaps fixing the crossover was the biggest part of this) - just seems to bring the timbre & detail of female vocals into a more present & realistic sound whilst also retaining smoothness (not harsh)....also the overall tonal balance is good with just the Anechoic EQ so certainly broadly from head to toe on the frequency curve I think it's pretty spot on.

 

[BYRTT]

@BYRTT
As others have pointed out, that spreadsheet is an approximation for maximum possible gain assuming sealed room and no losses, rarely if ever the case (IME, with drywall construction and typical some open room flow, not even close).

That aside though, I think it's fair to question the response given the variance in Revel's own design targets there. Assuming the Klippel is giving perfect results (and even Amir isn't 100% sure as he states in this thread), the different Revels have notably different low frequency tunings with the F328Be the softest of the bunch.

The 3 8" woofers provide large dynamic range and that no doubt helps support the impression of loud authoritive bass: they will just play bass loudly.

But all these different low frequency tunings can't be "right". Its a very worthy question and observation. Note that a higher f6 will often drive a stronger bass hump as a trick to give the impression of deeper bass, and we're seeing some of that here, but it doesn't seem to be the complete story

F328Be:
View attachment 93309

F208:
View attachment 93310

M106:
View attachment 93311

M16:
View attachment 93312

Fine but think there is much more to this than that low end extension target curve is looking weak in people eyes compared what they got used to look at and can afford because much stuff is cheap these days and many have the finances, think in there is really so many protests from users that a relative simple model to illuminate why it can happen Revel shoot for that target was on its place, also honestly Amir told loud and clear there is plenty of clean low end and then user responses critize him how he can tell about something that isnt visible in the charts, animation show whats going on and why some speaker systems for decades had could produce impressing bass notes including acoustic instruments, feel free anybody mean whatever one like better and use few to tens of subwoofers but please see the reality why some have impressing natural low end on less complicated systems.

About the spreadsheet yes pressurization gain can be all over the place in reality so lets omit it but per boundary for example the floor is probably spot on and Amir's curve of F228Be needs no more that the avarage of the three woofers distances to floor to exceed manufacture curve and get into teritory reproduce basic tone of 41Hz from a normal bass guitars open E string..

Right for first time ive seen it sounds in this thread Amir isnt 100% sure for low end curve but who can withstand that pressure when it goes on an on thread after thread and only because many manufacture curves is different, see Erin's analyze for F226Be was also different and Amir had third sample of KH 80 so close to manufacture curve when temparature was summer, will guees for tech interested that third party analyzes as Amir and Erin support us is the better clean picture of acoustic output compared manufacture brochure where data could be interpolated for many reasons.

3 times 8 inchers, large dynamic range, different frequency tunings, and F328Be is the softest of the bunch in low end tuning, what i think is going on for this premium speaker and why it sounds so good to Amir even anechoic low end is looking lean is if you rotate per transducers impedance graph left 90º, ignore resistence curve, mirror the phase curve we get how that transducer manipulate time domain 20Hz-20kHz because transducer is a current device and impedance isnt flat, now it can be for some that per transducer inside intended passband that the time distortion is a non noticeable signature, but when we in a 3.wayer sum three different manipulations of time domain or say signatures we get measureable distortions here and there because they really so different, when measureable distortion drop to fantastic low numbers as for F328Be think stuff as this is brought in sync as close as possible and probably why it sounds so great and is so expensive to build, outcome is more than usual system coherence for harmonics down to their basic notes.

 

[killdozzer]

Man, Revel gets a lot of love! I had a chance to buy the Concerta. Didn't know much about them back then.

These:

 

[andreasmaaan]

...if you rotate per transducers impedance graph left 90º, ignore resistence curve, mirror the phase curve we get how that transducer manipulate time domain 20Hz-20kHz because transducer is a current device and impedance isnt flat...

@BYRTT would you mind explaining what you mean here? Sorry, it's just that I think this sounds interesting, but I don't quite get what you're saying.. Thx.

 

[aarons915]

Luckily, I downloaded it.

See attached pdf or view photos below. Cover and last page are not included.

That's it, so they were actually 1 preference point away but I still wonder how much of that preference is explained by the bass difference. Either way I would bet that they were so close to the Ultima series and actually beat the Studio 2, which is already discontinued, so the Be line will either be their top line now or the Ultima 3 line will be announced soon.

 

[DDF]

Fine but ...Amir told loud and clear there is plenty of clean low end and then user responses critize him ....

Thanks for the detailed response @BYRTT. I agree with you that such criticism is a tiresome topic. You'll find I didn't

...per boundary for example the floor is probably spot on and Amir's curve of F228Be needs no more that the avarage of the three woofers distances to floor to exceed manufacture curve

The point I made was that Revel changed the bass tuning. As science minded people, we wonder "why"? I think @andreasmaaan made a very good case that it was to reduce the box size. I think Amir's subjective comments about his back pain suffered moving these brutes around reinforce that supposition Good discussion, I think.

...Amir isnt 100% sure for low end curve but who can withstand that pressure when it goes on an on thread after thread...

I agree with you that such criticism is a tiresome topic. I briefly mentioned it only to illustrate that if this inaccuracy did exist, it doesn't matter when comparing the relative low frequency responses of Revels speakers

...outcome is more than usual system coherence for harmonics down to their basic notes.

I spent many years studying the psychoacoustic literature, and I don't think group delay improvements (if they exist) are making any significant impact here. I think its a bit simpler:
1. The bass tuning trades off upper bass for slower bass roll off, allowing bass to be audible to lower frequencies
2. Three woofers can reduce the audible effect of the floor bounce as the 3 drivers fill in the response notch for each other

 

[richard12511]

I'm still not entirely convinced that the bass measurement is correct.

If I assume the bass measurements are correct:

While I appreciate @BYRTT's fantastic animations, I have to disagree with his conclusion that such a rolled off bass response(kinda starting at 200Hz) is a good thing, or something that Revel engineers were striving for intentionally. I'm sure floor bounce and LF gain will increase the overall in room bass output, but what we're seeing here really is a bad anechoic bass response. Also, having too much bass is not really a bad thing, as it can always be brought down with EQ. However, too little bass isn't fixable with EQ(at least not without limiting headroom).

If a rolled off anechoic bass response like we see here were really the preferred bass response(due to LF gain), then that should have showed up in the Harman blind tests, but it didn't. The blind tests showed that people overwhelmingly preferred either an anechoically neutral bass response, or an anechoic bass response that was slightly boosted above neutral. This speaker is already down 3dB by 67Hz(Revel claims the -3 is 34Hz ). I'm sorry, but if we're judging this speaker based on the Harman/NRC science(as we have every other speaker), then this is a very poor bass response for a speaker of this size and cost, and I think that(combined with the higher resolution of 2000 measurements) is what is dragging the score down lower than expected. We were very critical of the Magnepan LRS for it's bass rolloff starting around 300Hz, so I see no reason why we shouldn't also be critical of this speaker, which appears to be shelved down starting at 200Hz.

As for the most likely cause for the bad bass response? I guess I agree with @andreasmaaan in that it's probably a matter of the box being too small. It seems the engineers had to sacrifice sound quality for the sake of aesthetics , which is sad to hear.


That said, I still don't truly believe these bass measurements are accurate. I keep coming back to Amir's subjective comments. Either Amir's ears were completely off that day(tricked by expectation bias?) and Revel is fudging the -3dB point by 33Hz, OR the bass measurements are wrong. I still see the latter option as the more likely option. I find it really hard to believe Revel would design a 51", 51kg, $16,000 loudspeaker...with a bass response that's already down 6dB below the preferred response by 56Hz. I also find it really hard to believe that Revel would be lying about the -3dB point by 33Hz.

 

[restorer-john]

Everything about speakers, rooms and people's preference is complicated. IME people tend to like what they are used to and certainly my initial reaction to a speaker with a very different response to my current speaker, particularly on a favourite test track, is inevitably going to be at the very least sceptical, at first.

Never a truer word was written.

 

[Robbo99999]

I'm still not entirely convinced that the bass measurement is correct.

If I assume the bass measurements are correct:

While I appreciate @BYRTT's fantastic animations, I have to disagree with his conclusion that such a rolled off bass response(kinda starting at 200Hz) is a good thing, or something that Revel engineers were striving for intentionally. I'm sure floor bounce and LF gain will increase the overall in room bass output, but what we're seeing here really is a bad anechoic bass response. Also, having too much bass is not really a bad thing, as it can always be brought down with EQ. However, too little bass isn't fixable with EQ(at least not without limiting headroom).

If a rolled off anechoic bass response like we see here were really the preferred bass response(due to LF gain), then that should have showed up in the Harman blind tests, but it didn't. The blind tests showed that people overwhelmingly preferred either an anechoically neutral bass response, or an anechoic bass response that was slightly boosted above neutral. This speaker is already down 3dB by 67Hz(Revel claims the -3 is 34Hz ). I'm sorry, but if we're judging this speaker based on the Harman/NRC science(as we have every other speaker), then this is a very poor bass response for a speaker of this size and cost, and I think that(combined with the higher resolution of 2000 measurements) is what is dragging the score down lower than expected. We were very critical of the Magnepan LRS for it's bass rolloff starting around 300Hz, so I see no reason why we shouldn't also be critical of this speaker, which appears to be shelved down starting at 200Hz.

As for the most likely cause for the bad bass response? I guess I agree with @andreasmaaan in that it's probably a matter of the box being too small. It seems the engineers had to sacrifice sound quality for the sake of aesthetics , which is sad to hear.


That said, I still don't truly believe these bass measurements are accurate. I keep coming back to Amir's subjective comments. Either Amir's ears were completely off that day(tricked by expectation bias?) and Revel is fudging the -3dB point by 33Hz, OR the bass measurements are wrong. I still see the latter option as the more likely option. I find it really hard to believe Revel would design a 51", 51kg, $16,000 loudspeaker...with a bass response that's already down 6dB below the preferred response by 56Hz. I also find it really hard to believe that Revel would be lying about the -3dB point by 33Hz.

I agree, bass measurements are wrong, was the first thing I thought when I saw the spinorama and then the listening review stating good bass. Room gain isn't really an argument because other speakers have room gain too, although BYRRT did show that a slow roll off amplifies room gain beyond what would normally be expected but I think that was under ideal conditions too.....but I don't think it's that, I think there's too much evidence (incl Revel's specs not matching the spin, which you pointed out) showing that the measurements are likely off for towers.

 

[infinitesymphony]

This speaker is already down 3dB by 67Hz(Revel claims the -3 is 34Hz ).

Yep, straight from the F328Be manual:

So we're almost an octave off from the stated specs.

It looks like they were having this same conversation at Audiogon back in February. Some people say they fill the room with incredible bass, others say they're bass-shy. The former seem to have large rooms, but that could be a coincidence; not everyone mentions room size.

Could these be speculative measurements based on projected in-room response rather than anechoic or semi-anechoic responses (i.e. is that something Harman/Revel would do)?

Edit: I see now that @ctrl covered many of these points with detailed info in an earlier post.

 

[Steve Dallas]

Are we holding the Klippel NFS to too high a standard here, regarding bass response measurements of floorstanders? The NRC states their measurements are inaccurate below ~150 Hz. Axciom says their chamber is inaccurate below ~150 Hz, they would need a chamber 6 to 8 times its size to correct that (still requiring compensation and splicing), and they built a 100' tall outdoor tower platform for bass measurements as a result. Stereophile publishes a bass response disclaimer with every set of measurements. KEF uses some kind of compensation and splicing in their published measurements, and they willingly disclose this.

Could it be that Amir has found the lower limit of the device in terms of physical speaker size in his room on that machine, and we simply need to accept it like we do with most other measurements? Or figure out a methodology for compensation and splicing similar to what others do?

It seems to me 20+ pages of exploration of why the measurements show thin bass response, when Amir subjectively disagreed with that assessment is a little... weird.