KEF R3 vs. Philharmonic BMR Grudge Rematch to the Probably not Death Thread

[Steve Dallas]

As none of you recall, I tested the roadshow BMRs that Dennis was kind enough to ship all around the country last year and posted a lengthy thread about the experience. And, I ended up liking them enough to buy a pair. They even unseated my KEF R3s, which I sold shortly after receiving the BMRs. For some reason, I missed the R3s since then. I had no idea why other than the narrower directivity of the R3s works better in some rooms, and I have a terrible room with no opportunity to treat the side walls due to a large window on the left side and a large doorway pocket on the right side.

KEF recently put the R series on sale again, and I purchased a replacement pair of R3s. This thread is about a rematch between the two speakers. I will use far fewer words this time.

The measurement method was the same as before, which is using a UMIK-1 and REW to employ the Moving Mic Method (MMM) in an approximately 18" cube around the headspace of the MLP, which is my office chair sitting about 1/3 distance from the back wall, which is tastefully treated with broadband absorption to control flutter echo. Each measurements includes approximately 60 samples. I made no effort to be consistent with colors in any of the graphs. Sorry about that.

Both pairs of speakers were placed atop Monoprice Monolith 24" stands that are filled with 16Lbs baking soda each. The bottom plates are treated with Dynamat Extreme, and Sorbothane discs adorn the top plates. The speakers are ~72" apart and ~90" from the MLP.

Signal path is as follows:

Source: Homebrew Windows 10 PC
Players: Foobar 2000 or Amazon Music HD
DRC: Dirac Live 3 Standalone
DAC: Schiit Modius USB
Amp: Peachtree Nova 300

 

[Steve Dallas]

Philharmonic BMR Measurements

Left BMR Uncorrected

Right BMR Uncorrected

Left BMR with Dirac DRC to 600Hz

Right BMR with Dirac DRC to 600Hz

The speakers were set up 22" from the front wall with 1.5" toe-in. That places the SBIR frequency at ~150Hz.

The room is obviously dominating bass energy to around 250Hz and influencing up to 1KHz. Interestingly, Dirac is able to correct the SBIR region, but is not able to maintain bass SPL to the target below that, even though I have boosted it significantly in Dirac's target curve. I have no idea why. I was so surprised by this that I ran the Dirac process again and ended up with the same result. More investigation needed. (Port too close to the wall, therefore changing the tuning frequency?) IN ANY CASE, I DO NOT ATTRIBUTE THIS TO THE SPEAKER.

All in all, this is a very good result, and it sounds great, if a bit diffuse, in this room.

 

[Keened]

It's a common audio myth that baking soda reduces resonance; like all good myths there is a kernel of truth to it. You actually need to use baking powder, the slow reaction of the acid and base products create miniature bubbles throughout the mixture which disrupt the resonances that would otherwise travel through it.

Of course one must replace the baking powder every few months as the potency decays.

 

[Steve Dallas]

KEF R3 Measurements

I have several sequential R3 measurements to post, but to keep like measurements near each other in the thread, I will skip ahead and start with the same uncorrected and corrected graphs as what you see above.

Left KEF R3 Uncorrected

Right KEF R3 Uncorrected

Left KEF R3 with Dirac DRC to 1000Hz

Right KEF R3 with Dirac DRC to 1000Hz

The speakers were set up 21" from the front wall with 1.0" toe-in for these measurements. That places the SBIR frequency at ~160Hz

The room-influenced frequency range up to 1KHz looks similar between the two speakers, although the BMR manages ~6dB better bass extension than the R3, and it is audible in material which extends that low.

For reasons unknown to me, Dirac does a much better of correcting bass frequencies and is able to maintain low F energy better with the R3s.

The R3s are brighter overall above 6KHz. Unsurprisingly, they sound brighter. My hearing extends to 14.8KHz, so I do not mind a little help above 15KHz. I do not find the brightness in the delta region offensive nor fatiguing, even during extended listening. Of course, if one studies spectral data for recorded music in that range, one discovers there is not much information up there relative to midrange frequencies. Or perhaps it is because I grew up during the era of smile EQ curves and am accustomed to a bit of extra brightness...

Overall, this is a near textbook result. No wonder I missed these speakers in this room.

 

[ryanosaur]

FWIW, @Steve Dallas , I enjoyed reading your thoughts with your thread being one of the few shared out by people from the Roadshow v2.
Prior to my becoming a Philharmonic owner, I did audition the older KEF R900 and Q950. (And for kicks, elsewhere, the R11) I found the KEFs to be quite enjoyable, though something about them triggered a listening fatigue experience thus reducing said enjoyment. But I digress.
I look forward to seeing your experience rehashed!
Cheers!

 

[Steve Dallas]

KEF R3 Shadow Flare Issue

@hardisj wrote about the shadow flare issue in his excellent review.

My samples also exhibited this problem after unboxing. I had to press in the shadow flares to fix this problem.

Once pressed in, that region flattened out nicely.

 

[Steve Dallas]

KEF R3 Toe-In and Imaging

I took measurements with the R3s toed in 0" through 2" in 0.5" increments. To my surprise the change in mid through high frequency response changed less than 1dB.

Somehow I did not save the 1.5 and 2" increments, but they were not substantially different than the 1" increment. Here comparison measurements between 0" and 1". These measurements are not perfect, because I slid 24x48x5" panels behind the speakers at some point as I went off script for some reason. Ignore the deltas between 100 and 200Hz.


KEF R3 Left Uncorrected

KEF R3 Right Uncorrected

As I increased the toe-in, the stereo image sounded best at 1.5" and I returned the speakers to that position. "Best" is where Johnny Cash ignores personal space and dances on the bridge of your nose, in case you did not know.

If you guys really want to twist my rubber arm, I could be convinced to take the 1.5 and 2" measurements again tomorrow.

 

[Steve Dallas]

KEF R3 Speaker Boundary Interference Response Experiment

I am using this test as an opportunity to re-evaluate the acoustic treatment in this room. As part of that I am looking at the front wall treatment. Dirac does a good job of compensating for SBIR, but it is possible it could be improved further by employing better broadband absorption.

To test this, I measured the R3s with no sound panels, with two 24x24x5" panels, and with five 24x24x5" panels. Unfortunately, I cannot center the panels behind the speakers due to the necessary position of the console, so they are offset to the outside. (See photo below.)

Owens Corning 703 has an Noise Reduction Coefficient (NRC) of ~1.1 at 150, so it will be effective at my SBIR of 160Hz. Here is a little graph I made:

This graph illustrates the effectiveness of no panels vs. two panels vs. five panels:

KEF R3 Right Uncorrected with 5" Panels

The purple plot is zero panels. The blue plot is two panels. The green plot is 5 panels. It appears the two panels covering 24x48" area are sufficient in this application, although results would likely be different if I could center the panels behind the speakers.

This is how the panels were positioned:

(Please forgive the perspective weirdness of my new phone's camera; I have not found time to figure it all out. Those screws are exactly level. Trust me.)

 

[Steve Dallas]

It's a common audio myth that baking soda reduces resonance; like all good myths there is a kernel of truth to it. You actually need to use baking powder, the slow reaction of the acid and base products create miniature bubbles throughout the mixture which disrupt the resonances that would otherwise travel through it.

Of course one must replace the baking powder every few months as the potency decays.

Any inert filler that prevents the columns from ringing is effective at reducing resonance. These stands are absolutely dead as configured.

I did not realize this is a "common myth," as I did not know anyone else was using baking soda. I just happened to have several 12Lbs bags of it on hand for my pool and decided to fill the stands with it. I suppose nothing is new under the sun.

Anyway, I do not see how this is germane, as there is no common resonance visible in the measurements between the speakers.

 

[Steve Dallas]

My Measured In-Room Response vs. Klippel NFS Predicted In-Room Response

These graphs are not well-level-matched. I simply eyeballed them to line up areas of agreement. All-in-all, the NFS does a pretty good job here. I do not show the predicted low energy in the 1 to 1KHz region, nor the brightness between 4 and 10KHz in my room, but otherwise, the NFS is pretty accurate down to ~200Hz with its PIR.

KEF R3 Left Uncorrected

KEF R3 Right Uncorrected

 

[Steve Dallas]

BMR vs. R3 Frequency Response - Uncorrected

I made no real attempt to level match these measurements. I simply overlaid the right speaker plots and matched SPL at 1KHz, then used the same level for the left speaker plots. There is nothing scientific about what you see here.

One key difference in the measurement methodology is the R3s once again [accidentally] had two sound panels behind them, which shows in the SBIR range between 100 and 160Hz. Feel free to ignore the FR differences in that range.

Otherwise, make of these what you will:

BMR vs. R3 Left Uncorrected

BMR vs. R3 Right Uncorrected

Both speakers are pretty much anechoically flat. My theory on the differences above 1KHz in my measurements is that the BMR tweeter's very wide dispersion, and the omni-directional nature of the BMR midrange create substantial reflections in this room which result in the unevenness we see in non-gated measurements. If I have time, I will set up the BMRs again and take some gated measurements in an attempt to prove that out.

That unevenness offends the eyes much more than the ears.

 

[bkdc]

Ignoring measurements, I am pretty sure that with subjective listening, the BMR will be preferred over the R3 by most ears. Or maybe I'm projecting my own strong personal preferences. Measurements do not convey the clarity and sense as if a singer is in the room with you instead of behind a veil. But alas, this forum wants objective measurements. The best measurement would be a blind level matched preference poll instead of room response curves.

 

[Steve Dallas]

FWIW, @Steve Dallas , I enjoyed reading your thoughts with your thread being one of the few shared out by people from the Roadshow v2.
Prior to my becoming a Philharmonic owner, I did audition the older KEF R900 and Q950. (And for kicks, elsewhere, the R11) I found the KEFs to be quite enjoyable, though something about them triggered a listening fatigue experience thus reducing said enjoyment. But I digress.
I look forward to seeing your experience rehashed!
Cheers!

Thank you for your kind words. I am willing to bet your fatigue originated from the extra brightness above 6KHz, which all KEF speakers I have measured show. Some people are fine with it and others find it bothersome over time. Salud!

 

[Mnyb]

Thank you for your kind words. I am willing to bet your fatigue originated from the extra brightness above 6KHz, which all KEF speakers I have measured show. Some people are fine with it and others find it bothersome over time. Salud!

Can the Dirac system compensate for the brightness if you want it to ?

 

[Steve Dallas]

Can the Dirac system compensate for the brightness if you want it to ?

Yes. You can use it full range if you prefer. I can take those measurements tomorrow. It only works well with controlled-directivity speakers, but both speakers in question fit that description.

 

[Mnyb]

Yes. You can use it full range if you prefer. I can take those measurements tomorrow. It only works well with controlled-directivity speakers, but both speakers in question fit that description.

Oh soo that’s how it works one can not just manually tweak some part of the curve . Dirac makes a fit to a chosen curve ? Forgive my ignorance on how Dirac works .

 

[ryanosaur]

Thank you for your kind words. I am willing to bet your fatigue originated from the extra brightness above 6KHz, which all KEF speakers I have measured show. Some people are fine with it and others find it bothersome over time. Salud!

It was pointed out to me that KEFs are best used perpendicular to the front wall, or at least firing past you at a good 15º off axis. My audition was aimed to the dome by the guy at the shop I visited. I got full brunt of the Uni-Qs

That said, previous conversations with Dennis had left me with some choices to make regarding the BMRs and Phil3s I have. He intended for these to be perpendicular, but somewhere, I had also seen him talk about 15º off axis listening.
For myself, without measuerments, I found toeing his Speakers in to fire just past and cross behind me to be the best experience.

 

[Steve Dallas]

Oh soo that’s how it works one can not just manually tweak some part of the curve . Dirac makes a fit to a chosen curve ? Forgive my ignorance on how Dirac works .

That's not exactly what I meant.

You can use any number of control points to adjust the target curve. You can also use curtains to control how much of the curve Dirac corrects.

For example, here is my R3 project with my target curve applied and correction limited to 1KHz:

Here is the same with the right curtain set to full range with the treble pulled down by the right-most control point:

Here is a more complex target curve for the BMRs limited to 1.5KHz:

(This one was part of early experimentation with Dirac and was not actually used other than to measure its effect.)

You can also save up to 8 presets in the processor and switch between them with a single mouse click to audition different profiles:

When I said it only works well with controlled-directivity speakers, I meant using DRC on frequencies above the room-dominated threshold is unpredictable unless directivity is well-controlled (smooth). This is because you are adjusting direct sound, which is then reflected off nearby surfaces and mixed back into the sound field. Correcting speakers with directivity errors results in some frequencies being corrected more than others in the reflections, which leads to unpredictable results.

 

[Keened]

Any inert filler that prevents the columns from ringing is effective at reducing resonance. These stands are absolutely dead as configured.

I did not realize this is a "common myth," as I did not know anyone else was using baking soda. I just happened to have several 12Lbs bags of it on hand for my pool and decided to fill the stands with it. I suppose nothing is new under the sun.

Anyway, I do not see how this is germane, as there is no common resonance visible in the measurements between the speakers.

I was just making a cheeky comment because I've never heard of someone using baking soda as a filler before (only sand and shot), so it was odd to me to have so much baking soda on hand. But that's because I've never owned a pool I guess

Excellent data, thank you for posting all of this.

Regarding Dirac, are you using the 2.0 version: is this the Studio version and not the Live version? I tried adjusting the curve when I had the RZ50 in the Live version and it didn't seem like I could add additional points, only move around the ones that existed.

 

[sdiver68]

Love this @Steve Dallas!

Also love to see you add your F206 to the fray. I know people will protest but I bought mine more to address the desire to have nice looking stands integrated than an absolute desire to have towers vs monitors.