Revel F35 Speaker Review

[Blumlein 88]

I actually disagree - Looking at the overall picture of the measurements I'm inclined to think the formula is working pretty well. I don't think they can or should be tied too closely to the impressions of one listener but overall the ones that measure poorly don't sound great and the ones that measure decently do. This is the expected result. AFAIK there hasn't been a speaker with terrible measurements that sounded great to Amir, and again, it's the impression of one listener.

Obviously though, I don't doubt Harman looks at far more things when designing their speakers than whatever has the best preference score. I mean, as far as I know, there's no reason to believe they actually calculate the preference score when designing new speakers - it's all about the blind tests.. But the principles are there.

Edit: Perhaps more important to me personally, the preference scores correlate quite closely with my interpretations of the data.

I play this little game now of guessing preference ratings based on the measurements, and I've gotten pretty good at it. And with one exception (the Harbeth), they're pretty in line with measurements I'd consider being great, good, decent, meh, or bad.

Well, I knew better than to think a formula like this would allow one to pick the better speaker down to two decimal places. And if nothing else it seems to work well at showing you which designs can be dismissed. That said in some cases Amir's reaction seems off vs the relative scores. I'd say I haven't learned enough about interpreting the data carefully that comes from this testing. Which is what I'd hope this testing and the spin data could achieve which is for someone to judge likely speakers to audition or not in a narrow relative price and usage range.

 

[Sancus]

Well, I knew better than to think a formula like this would allow one to pick the better speaker down to two decimal places. And if nothing else it seems to work well at showing you which designs can be dismissed. That said in some cases Amir's reaction seems off vs the relative scores. I'd say I haven't learned enough about interpreting the data carefully that comes from this testing. Which is what I'd hope this testing and the spin data could achieve which is for someone to judge likely speakers to audition or not in a narrow relative price range.

There are a lot of problems with the listening tests. As was already established, room modes are a big one. Second to that is everyone loves to cite the no-LFE scores because they're bigger numbers, but no listening tests have been conducted with a subwoofer, let alone with multiple subwoofers and room EQ, which is what would be required to get close to the PIR in real world use. The Pioneers did perform well enough in Wirecutter's blind tests to merit a budget pick, so that is some support for their score.

It also seems extremely clear to me at least that the Genelecs are the best speaker tested, despite them not getting a "golf panther", which is backed up by both the measurements and the score. They don't have that much bass output, which restricts their overall SPL, but I really just don't agree with marking them down for that, because if you really need more bass SPL you're going to use a sub or two(or buy 8351b/8361a.. or both). And the max SPL is clearly stated by manufacturer specs. Of course, they're pricy, but for that price they come with built-in room correction, which is quite a big advantage.

So I don't really see the score being terribly off. It's important to remember that you really need at least 1 point between speakers, if not 1.5 points, to guarantee that the majority of people "should" prefer one to the other. Quibbling over decimal points is a waste of everybody's time.

 

[richard12511]

I actually disagree - Looking at the overall picture of the measurements I'm inclined to think the formula is working pretty well. I don't think they can or should be tied too closely to the impressions of one listener but overall the ones that measure poorly don't sound great and the ones that measure decently do. This is the expected result. AFAIK there hasn't been a speaker with terrible measurements that sounded great to Amir, and again, it's the impression of one listener.

Obviously though, I don't doubt Harman looks at far more things when designing their speakers than whatever has the best preference score. I mean, as far as I know, there's no reason to believe they actually calculate the preference score when designing new speakers - it's all about the blind tests.. But the principles are there.

Edit: Perhaps more important to me personally, the preference scores correlate quite closely with my interpretations of the data.

I play this little game now of guessing preference ratings based on the measurements, and I've gotten pretty good at it. And with one exception (the Harbeth), they're pretty in line with measurements I'd consider being great, good, decent, meh, or bad.

ASR reviews have definitely decreased my confidence in Harman's ability to predict user preference based on spinorama measurements, but (at least in my case) I think it's because I misunderstood what the correlation metric in Olive's study meant. I had previously taken the results of the study to mean that Harman could look at any two spinorama graphs and predict the double blind preferred speaker 86% of the time, or 99% of the time with equalized bass. I realize now that those 86% and 99% figures don't mean what I thought they did.

Right now, I think I view the score as something that can tell you if two speakers are in the same range of sound quality, but very little beyond that. I'm moving more towards the traditional subjectivist view of "you have to listen", albeit blind, and really only if the two speakers are within a point or a point and a half. It's really weird, and I kinda hate the fact that it seems I'm moving more towards the subjectivist viewpoint.

I agree with you in that my intuition from looking at the graphs also seems to track really well with the score. I've also been doing the "try to guess the score" game. I knew that the Genelec was by far the best speaker measured(and still is). I correctly guessed that the two Revel speakers would probably land in the mid 7 range. I guessed high 7 range for the KEF R3. It ended up low 8s, but I did correctly identify that it was the second best measured speaker yet.

 

[Dimitri]

I view the score as something that can tell you if two speakers are in the same range

My two line summary:
A)The better the score the easier they can be persuaded to behave in a space with room correction.
B)The lower the score the more "personality" they have and no DSP in the world is going to tame/change it.

There might even be an implicaton that as far as "preferences" go, one might even like the sound of B no-matter how "correctly" A is equalized.

 

[napilopez]

Well, I knew better than to think a formula like this would allow one to pick the better speaker down to two decimal places. And if nothing else it seems to work well at showing you which designs can be dismissed. That said in some cases Amir's reaction seems off vs the relative scores. I'd say I haven't learned enough about interpreting the data carefully that comes from this testing. Which is what I'd hope this testing and the spin data could achieve which is for someone to judge likely speakers to audition or not in a narrow relative price range.

Amir has also said though that he's working on trying his subjective impressions to the measurements, so it's a learning process for all of us.

Imo the measurements are useful for purchasing decisions. It's a vote of confidence you'll enjoy a speaker, similar to what @MZKM said about rotten tomatoes scores. Of course, the highest scoring speakers are also likely to sound best.

Right now, I think I view the score as something that can tell you if two speakers are in the same range of sound quality, but very little beyond that. I'm moving more towards the traditional subjectivist view of "you have to listen", albeit blind, and really only if the two speakers are within a point or a point and a half. It's really weird, and I kinda hate the fact that it seems I'm moving more towards the subjectivist viewpoint

I wouldn't say you're moving towards a more subjectivist viewpoint, you just have a better understanding of the science. Remember, blind tests are king. The scores attempt to model/predict blind test results. So of course, listen blind when you can, or listen sighted with a healthy understanding of the measurements and where the flaws lie.

For making purchasing decisions without the luxury of listening, measurements are the most reliable source of information should the reader know how to interpret them.

 

[infinitesymphony]

Hopefully we're going to see more closer-to-full-range speakers tested that do not necessarily require a subwoofer. These floorstanding speakers are -18 dB at 41 Hz (fundamental frequency of low E on bass guitar). As @bobbooo pointed out, the bass response in these is no better than the Revel M16, Pioneer SP-BS22-LR, and other bookshelf speakers.

 

[BYRTT]

Hopefully we're going to see more closer-to-full-range speakers tested that do not necessarily require a subwoofer. These floorstanding speakers are -18 dB at 41 Hz (fundamental frequency of low E on bass guitar). As @bobbooo pointed out, the bass response in these is no better than the Revel M16, Pioneer SP-BS22-LR, and other bookshelf speakers.

In on axis overlaid (thin orange) below that 41Hz should had been a 21Hz with reference manufacture published curve for F35, for now think we cant know if manufacture published data is a golden sample that is impossible produce in high rates on a real world production line or if ASR sample is presenting a bad examble a avarage one or is kind of broken, but in suppose their fabric have some QC sample is probably meant as good enough pass in it ended shipped to amirm.

 

[amirm]

Here is an overlay of the F35 and M16 (in room/ungated):

The F35 (in teal/blue and named "Fundamental") has a lot more bass as frequencies go lower than M16 (in red). @MZKM, is the scoring just picking up the initial 6 dB drop and ignore lower bass response than that?

 

[Sal1950]

I do wish manufacturers would offer the speakers in the least expensive fit & finish it is possible to make them in. I'd like to be able to purchase SOTA performance without paying for furniture grade looks. Back in 1978 I was able to buy my new La Scala's unfinished in raw marine grade plywood and then I finished them myself. You can't do anything like that with them anymore either.
Same goes for cars. Back in the day you could order a new base model Chevy with the highest hp rated bigblock, 4 sp, 410 rear end etc, and if you were really serious you could check the boxes for things like heater and radio delete to save both money and weight. Today if you walk into Chevy and want the top performing engine, etc; package on a Camero you also have to get the very top of the line interior, wheels, appearance packages too.
The get into your wallet coming and going.

So you have to buy a bike. There is no things like line interior, heater and radio. Less pollution. Better health.
Well, I know, I know...

Are you talking about bicycles or motorcycles?
Now if motorcycles I can totally agree. Back in the very early 1970's I learned I could build a light weight Harley that could break into the 12's at the strip, kick the a-ss of any street-able door slammer and get comparably great gas mileage to boot. That was the beginning of a 50 year love affair with high performance motorcycling.
Faster bikes, looser women, and more money, that was the credo. LOL

 

[Haint]

Not sure if it's of any value or interest, but I took a sweep of an F35, M16, and S16 in the same spot of my room (tape line) without moving anything save for swapping in the 24" stand for the bookshelves. They sit ~5 feet from the front wall and 4.25 feet from the side wall, though there is a 6" deep rockwool absorber straddling the front corner, and another on the front wall directly behind the speaker. Mic a little over 8' from the tweeter.

1/6th Smoothing | Red F35 | Blue M16 | Green S16 | No EQ, No Subs

I'm assuming the 500Hz peaking on the bookshelves is related to height differences of whatever woofer(s) is producing that frequency on the tower, relative to the bookshelf. I have hot swapped the F35 and M16 before for listening tests and didn't really notice any difference, though it was in Stereo and EQ'ed up to around 500 or 600Hz.

 

[BYRTT]

Here is an overlay of the F35 and M16 (in room/ungated):
View attachment 54490

The F35 (in teal/blue and named "Fundamental") has a lot more bass as frequencies go lower than M16 (in red). @MZKM, is the scoring just picking up the initial 6 dB drop and ignore lower bass response than that?

Thanks that is informative but aaaah a bit thin isn't, think one could hope for more low end reach now they upped effective cone area, in below M16 is overlaid F35 well only on axis/power response/DI plus a blinking manufacture low end curve that would have rocked some points, its unfortunate so many speaker designs look have less or other low end extension than what their brochure or designer thought they had, but lets then hope they learn a bit from these revealings and get their tools and methods dialed in to tune right in accordance their target slope or simulated models.

 

[QMuse]

Well, I eq my L/R speakers differently , which results in them being eq'd the same if that makes sense. I don't want my L/R to sound different.
They should sound the same. Without eq in my room, they don't, since their frequency response is different from L to R at low frequencies. I don't eq at all above 500 Hz as the room doesn't affect the response much if at all above this point.

I agree with everything you said except for the "above 500Hz" part. Take a look at Amir's in room measurement of Kef R3

and tell me that above 500 room doesnt' affect it:

The truth is that room affects speaker's response up to the 900Hz and that can be seen from this graph as it is very typical.

 

[QMuse]

Not sure if it's of any value or interest, but I took a sweep of an F35, M16, and S16 in the same spot of my room (tape line) without moving anything save for swapping in the 24" stand for the bookshelves. They sit ~5 feet from the front wall and 4.25 feet from the side wall, though there is a 6" deep rockwool absorber straddling the front corner, and another on the front wall directly behind the speaker. Mic a little over 8' from the tweeter.

1/6th Smoothing | Red F35 | Blue M16 | Green S16

View attachment 54494View attachment 54495View attachment 54496View attachment 54497

I'm assuming the 500Hz peaking on the bookshelves is related to height differences of whatever woofer(s) is producing that frequency on the tower, relative to the bookshelf. I have hot swapped the F35 and M16 before for listening tests and didn't really notice any difference, though it was in Stereo and EQ'ed up to around 500 or 600Hz.

Well, you have nicely proven that effect room has on speaker is related to the point of space in a room. All speakers when put at that point would suffer the same effect.

For Amir this effect makes job easier as he has to create room EQ filters for only 1 or 2 spots where he's putting the speakers while listening to them.

 

[tuga]

"Perhaps more important to me personally, the preference scores correlate quite closely with my interpretations of the data.

I play this little game now of guessing preference ratings based on the measurements, and I've gotten pretty good at it. And with one exception (the Harbeth), they're pretty in line with measurements I'd consider being great, good, decent, meh, or bad. "

The funny thing is that though I think Amir's impressions roughly follow the formula, I often find myself disagreeing with him on some of the interpretations (and I think my interpretations follow the formula more closely, actually). That's the beauty of the measurements though; we have the same data to interpret as we see fit.can interpret them as we see fit

The crux of the matter is in your first sentence: "Perhaps more important to me personally, the preference scores correlate quite closely with my interpretations of the data."

As for the Harbeth, one can argue that many people disagree with your assessment that their Spinorama performance is not representative of their success appeal (see below).

Sales numbers are a poor way of assessing quality of course

But we are discussing preference, not quality.
I think that here, in my view, is where the problem lies. Preference is what people like and choose, not what people should be buying.

I also find it unacceptable that speakers with very different SPL and LF extension capabilities all be put in the same baskest. The smallest Genelec may perform admirably in all parameters but it just can't be used without subs in typical domestic circunstances and thus cannot be compared with a large standmount or small floorstander with larger cone surface and cabinet volume.

Finally, even though the rather limited research in regard to driver + cabinet resonances and harmonic + intermodulation distortion might indicate a very crude audibility threshold perhaps too much emphasis is being given to directivity comparatively.

Either way, I really don't think that the preference rating is of any use; fortunately I'm able to interpret the measurements and draw my own conclusions.

 

[tuga]

Thanks that is informative but aaaah a bit thin isn't, think one could hope for more low end reach now they upped effective cone area, in below M16 is overlaid F35 well only on axis/power response/DI plus a blinking manufacture low end curve that would have rocked some points, its unfortunate so many speaker designs look have less or other low end extension than what their brochure or designer thought they had, but lets then hope they learn a bit from these revealings and get their tools and methods dialed in to tune right in accordance their target slope or simulated models.
View attachment 54501

Could this unprecedented and unexpected dip at 30Hz somehow be related with the fact that these are the first floorstanders to be measured?

 

[BYRTT]

Could this unprecedented and unexpected dip at 30Hz somehow be related to the fact that these are the first floorstanders to be measured?

Good quistion and maybe never say never that said with a ton of knowledge in acoustics Klippel NFS operator is amirm himself therefor don't think so and also its all the ported designs where tune goes wrong so it could be method and tools plus available time they use is by mistake out of sync to reality and in far field being indoor its probable impossible get precise curve of port plus tranducer sum, NFS analyze 360deg around and for C52 we had a sealed design that looked a textbook 2nd order roll of some Q in the eight or nenties IIRC so that NFS looks know its stuff down there, but lets see down the road in that for example its said Neumann is investigating their KH 80 numbers from ASR site including get a Klippel NFS analyze and they probably will feedback down the road, its just that we have to wait because it probably takes serious long time investigate deep so that all possible loose ends for subject get covered. Therefor so far myself believe in amirm's analyze for F35 at low end reach and also few post back @Haint shared PIR of F35 verse M16 indoor in same location and that sweep looks confirm the Klippel haircut is right and port tuning is not inline what manufacture publish or think it is.

 

[tuga]

Good quistion and maybe never say never that said with a ton of knowledge in acoustics Klippel NFS operator is amirm himself therefor don't think so and also its all the ported designs where tune goes wrong so it could be method and tools plus available time they use is by mistake out of sync to reality and in far field being indoor its probable impossible get precise curve of port plus tranducer sum, NFS analyze 360deg around and for C52 we had a sealed design that looked a textbook 2nd order roll of of some Q in the eight or nenties IIRC so that NFS looks know its stuff down there, but lets see down the road in that for example its said Neumann is invetigating their KH 80 numbers from ASR site including get a Klippel NFS analyze and they probably will feedback down the road, its just that we have to wait because it probably takes serious long time investigate deep so that all possible loose ends for subject get covered. Therefor so far myself believe in amirm's analyze for F35 at low end reach and also few post back @Haint shared PIR of F35 verse M16 indoor in same location and that sweep looks confirm the Klippel haircut is right and port tuning is not inline what manufacture publish or think it is.

I agree that the manufacturer specified Low Frequency Extension as 55Hz, 46Hz, 35Hz (-3 dB, -6 dB, -10 dB) is far from what can be seen in Amir's measurement.
It's the fact the there's a quite significant dip at 30Hz and a disparity between measurement and specs that raises my concern regarding a potential issue with the measuring process.

I think that it would be helpful if Amir could perform isolated measurements of the port to get an idea of the port tuning and of the mid-woofer to get an idea of the decay / breakup modes (for all speakers which allow this to be done).

 

[vavan]

I am curious, if these measure so well, what would be the point of going with the new beryllium tweeters?

there's not only new be tweeters, also different (perhaps better?) drivers and crossovers varying from say f208 to f228be

 

[Bear123]

For comparison, here is my F36 full range with no eq, in room. I'll run another sweep and make sure I didn't have DEQ on.

 

[tuga]

there's not only new be tweeters, also different (perhaps better?) drivers and crossovers varying from say f208 to f228be

The decay of the midrange driver looks significantly different and according to Stereophile "unlike the plain aluminum midrange and woofer cones of the F208 or the titanium cones of the Studio2, the F228Be has diaphragms made of Revel's new Deep Ceramic Composite (DCC), in which aluminum also plays a role"

F208 - 5.25" (133mm) aluminum-cone midrange unit with cast frame

F228 - 5.25" (130mm) DCC (Deep Ceramic Composite) aluminum-cone midrange



Interestingly those weird ridges above 1kHz are also present in the CSD of the Vivid Kaya 45:

Kaya 45