This is a review and detailed measurements of the KEF R3 stand-mount (bookshelf) speaker. It is on kind loan from a member who sent a pair to me at great cost (they come two in a box). The R3 costs US $1,999 for a pair.
The R3 is an example of superb industrial design by KEF:
It oozes elegance and comes in a number of colors to please anyone. It is also highly differentiated by its coaxial design and gorgeous woofer.
The back side looks plain but you are not going to be looking at that:
The sharp corners make for a nice design but likely not good for diffraction effects (little speakers playing on their own at ever corner).
We have the same knobs that allow the woofer to be separated from the rest to bi-wire/bi-amp the speaker. I only tested the R3 as one unit.
Measurements that you are about to see were performed using the Klippel Near-field Scanner (NFS). This is a robotic measurement system that analyzes the speaker all around and is able (using advanced mathematics and dual scan) to subtract room reflections (so where I measure it doesn't matter). It also measures the speaker at close distance ("near-field") which sharply reduces the impact of room noise. Both of these factors enable testing in ordinary rooms yet results that can be more accurate than anechoic chamber. In a nutshell, the measurements show the actual sound coming out of the speaker independent of the room.
All measurements are reference to tweeter axis with the grill removed. Frequency resolution is 0.7 Hz (yes, less than 1 Hz) and plots are at 20 points/octave. Spatial 3-D resolution is 1 degree.
Over 800 points around the speaker were measured (from 20 to 20 kHz) which resulted in well under 1% error in identification of the sound field to almost 20 kHz where error increased a bit (likely not visible on graphs).
Spinorama Audio Measurements
Acoustic measurements can be grouped in a way that can be perceptually analyzed to determine how good a speaker can be used. This so called spinorama shows us just about everything we need to know about the speaker with respect to tonality and some flaws:
They eye averages this out to pretty flat on-axis response. While not high in amplitude, there is a broad increase in level between 3 and 10 kHz. This will hit a lot more musical notes so it may make the speaker a bit bright relative to a speaker that has some choppiness but narrow peaks.
Sensitivity is on the money with respect to company spec if you look at mid frequencies. Deep bass though is lower at around 80 dB so amplification requirements will be significant.
We can use statistical model of listening spaces to determine, using 3-D measurements we have around the speaker, what you will hear in a real room (above is anechoic or "free field" data). That gives us this likely response:
Due to well behaved off-axis response, the curve more or less shows what the on-axis did which is good. The sloping down is expected and required. Because of good "directivity" (off-axis versus on-axis), EQ should be effective and a touch of that may improve subjective performance in upper range. Otherwise the slight suck-out in 1000 to 2000 may be too audible in how it will send that range into the background a bit.
Impedance dips to 3 ohms which is quite low, emphasizing the point I made earlier about needing a good amplifier to drive these speakers:
40 Hz happens to be highest peak in the spectrum of a set of files in my music library so having the lowest impedance there is unfortunately. Another dip exists at 150 Hz.
The distortion graphs is still work in progress as I continue to fight the Klippel software to generate proper SPL measures. It seems that in this graph, despite what I reported earlier, SPLs are not trustworthy. The ratios are. So let's look at that while giving the speaker 10 volts to push it more than 2.83 volts that I use for spinorama:
As noted, this measurement takes out the room using clever filtering in Klippel software (uses the free-field response to compensate). This results in much smoother low frequencies with no need to use gating, and response down to 20 Hz. I truncated at 50 Hz however as allowing it to go below that, shoots the distortion to nearly 100% as the speaker is incapable of producing usable output that low. An effect that I could hear in the sweeps.
There is no sudden peaks or other anomalies that stand out. Very low frequency dominates (as does with all speakers tested) which due to our high threshold of hearing and masking, likely is not very audible.
I let you all study the waterfall and see if you can correlated the little wiggles in the spinorama with the graph here:
And here is our directivity shots:
We see that 0 degree doesn't land at the center of the the driver indicating that the acoustic axis is a bit lower. Or I did not quite center the mic on the tweeter (a very delicate job given the lack of protection on the driver). I went through the painful process of compensating for this (trial and error with every cycle taking 15 minutes) but it made no difference at all in the spinorama results. Not sure if this is right or wrong. Will have to ponder more on the data and/or ask Klippel.
Here is the vertical slice:
It is as pretty as single coaxial drivers due to woofer handling lower bass now.
Speaker Listening Tests
I setup the KEF R3 in the same far-field setting I have used for my other high-fi speakers in my main system. It is stand mounted with tweeter roughly at ear level with me sitting some 12 feet away. The speaker was driven by a 1,000 watt monoblock amplifier (into 4 ohm) so power availability not an issue. Fancy audiophile cable was used so no worries there.
First the good news: the type of buzzing distortion I thought I heard with KEF Q100 was not there. It was replaced by very clear response together with strong deep bass when required. Power handling was now excellent as I could turn up the speaker as much as I needed and despite only one speaker playing. The sound was clean.
Alas, once again subjective feeling was low. My standard routine is to cycle through my reference clips that I have selected during all my normal listening to sound superb on my Revel Salon 2 Speakers. Sadly hardly any of them sounded all that good here. Yes, the highs were there. The lows at times were there. But overall experience was unexciting and unengaging for lack of a better word.
To make sure my mind has not gone crazy, I replaced the R3 with Revel M16 which I recently reviewed. Wow, all that gorgeous came right back! There are two things I clearly detect:
1. Warm, fantastic mid to upper bass. I can't emphasize enough how much difference this makes and how it impacts my subjective reviews.
2. There is a "loosness" to notes that is hard to describe but notes are separated and delightfully clean and pleasant. I get this same sense when I EQ a speaker for a room.
EDIT: later testing showed that the room mode at about 105 Hz was impacting the tonality of the speaker. Once I filtered that, the sound was very good. I have since made this part of the protocol.
Discussion and Conclusions
The objective measurements will nail the Olive score no doubt. And they present for the third time a conflict with my subjective listening impressions. Of course my subjective evaluation is much less reliable. I like us to allow some allowance for them to sink in though. Olive's latest tests shows people like to hear more bass than originally though. So could this be behind my preference for speakers like Revel M16? Here is its predicted-in room response:
I am really starting to think the 100 to 200 Hz region plays a much stronger role than we think in subjective sound a speaker produces. The other factor is not letting the higher frequencies dominate the mid-range. As I noted in the review, broad deviations in the measurements, despite their low level, may have a much larger subjective difference.
At some point we will have to reconcile these differences, either setting me straight on my subjective evaluations being wrong, or us not knowing all that Harman knows about good speaker sound. Let's remember that they won't release a new speaker unless it passes double blind listening tests against its competitor. No other score allows them to skip this test. Components are tweaked until they achieve this. So one wonders if this is not released to public.
For now, objective measurements are superb enough to give a thumbs up to KEF R3 and hence the choice of panther.
Personal note:
This pair of speaker cost US $211 to ship to me with insurance. This presented huge hardship for the owner to fund but he was adamant in doing his bit to help with our research into speaker technology. I will be paying for shipping back but like to also help him what he has paid to get them to me. If you like to support him the same way, and there is no pressure, just start a conversation with me and I will collect the funds to give him.
The R3 is an example of superb industrial design by KEF:
It oozes elegance and comes in a number of colors to please anyone. It is also highly differentiated by its coaxial design and gorgeous woofer.
The back side looks plain but you are not going to be looking at that:
The sharp corners make for a nice design but likely not good for diffraction effects (little speakers playing on their own at ever corner).
We have the same knobs that allow the woofer to be separated from the rest to bi-wire/bi-amp the speaker. I only tested the R3 as one unit.
Measurements that you are about to see were performed using the Klippel Near-field Scanner (NFS). This is a robotic measurement system that analyzes the speaker all around and is able (using advanced mathematics and dual scan) to subtract room reflections (so where I measure it doesn't matter). It also measures the speaker at close distance ("near-field") which sharply reduces the impact of room noise. Both of these factors enable testing in ordinary rooms yet results that can be more accurate than anechoic chamber. In a nutshell, the measurements show the actual sound coming out of the speaker independent of the room.
All measurements are reference to tweeter axis with the grill removed. Frequency resolution is 0.7 Hz (yes, less than 1 Hz) and plots are at 20 points/octave. Spatial 3-D resolution is 1 degree.
Over 800 points around the speaker were measured (from 20 to 20 kHz) which resulted in well under 1% error in identification of the sound field to almost 20 kHz where error increased a bit (likely not visible on graphs).
Spinorama Audio Measurements
Acoustic measurements can be grouped in a way that can be perceptually analyzed to determine how good a speaker can be used. This so called spinorama shows us just about everything we need to know about the speaker with respect to tonality and some flaws:
They eye averages this out to pretty flat on-axis response. While not high in amplitude, there is a broad increase in level between 3 and 10 kHz. This will hit a lot more musical notes so it may make the speaker a bit bright relative to a speaker that has some choppiness but narrow peaks.
Sensitivity is on the money with respect to company spec if you look at mid frequencies. Deep bass though is lower at around 80 dB so amplification requirements will be significant.
We can use statistical model of listening spaces to determine, using 3-D measurements we have around the speaker, what you will hear in a real room (above is anechoic or "free field" data). That gives us this likely response:
Due to well behaved off-axis response, the curve more or less shows what the on-axis did which is good. The sloping down is expected and required. Because of good "directivity" (off-axis versus on-axis), EQ should be effective and a touch of that may improve subjective performance in upper range. Otherwise the slight suck-out in 1000 to 2000 may be too audible in how it will send that range into the background a bit.
Impedance dips to 3 ohms which is quite low, emphasizing the point I made earlier about needing a good amplifier to drive these speakers:
40 Hz happens to be highest peak in the spectrum of a set of files in my music library so having the lowest impedance there is unfortunately. Another dip exists at 150 Hz.
The distortion graphs is still work in progress as I continue to fight the Klippel software to generate proper SPL measures. It seems that in this graph, despite what I reported earlier, SPLs are not trustworthy. The ratios are. So let's look at that while giving the speaker 10 volts to push it more than 2.83 volts that I use for spinorama:
As noted, this measurement takes out the room using clever filtering in Klippel software (uses the free-field response to compensate). This results in much smoother low frequencies with no need to use gating, and response down to 20 Hz. I truncated at 50 Hz however as allowing it to go below that, shoots the distortion to nearly 100% as the speaker is incapable of producing usable output that low. An effect that I could hear in the sweeps.
There is no sudden peaks or other anomalies that stand out. Very low frequency dominates (as does with all speakers tested) which due to our high threshold of hearing and masking, likely is not very audible.
I let you all study the waterfall and see if you can correlated the little wiggles in the spinorama with the graph here:
And here is our directivity shots:
We see that 0 degree doesn't land at the center of the the driver indicating that the acoustic axis is a bit lower. Or I did not quite center the mic on the tweeter (a very delicate job given the lack of protection on the driver). I went through the painful process of compensating for this (trial and error with every cycle taking 15 minutes) but it made no difference at all in the spinorama results. Not sure if this is right or wrong. Will have to ponder more on the data and/or ask Klippel.
Here is the vertical slice:
It is as pretty as single coaxial drivers due to woofer handling lower bass now.
Speaker Listening Tests
I setup the KEF R3 in the same far-field setting I have used for my other high-fi speakers in my main system. It is stand mounted with tweeter roughly at ear level with me sitting some 12 feet away. The speaker was driven by a 1,000 watt monoblock amplifier (into 4 ohm) so power availability not an issue. Fancy audiophile cable was used so no worries there.
First the good news: the type of buzzing distortion I thought I heard with KEF Q100 was not there. It was replaced by very clear response together with strong deep bass when required. Power handling was now excellent as I could turn up the speaker as much as I needed and despite only one speaker playing. The sound was clean.
Alas, once again subjective feeling was low. My standard routine is to cycle through my reference clips that I have selected during all my normal listening to sound superb on my Revel Salon 2 Speakers. Sadly hardly any of them sounded all that good here. Yes, the highs were there. The lows at times were there. But overall experience was unexciting and unengaging for lack of a better word.
To make sure my mind has not gone crazy, I replaced the R3 with Revel M16 which I recently reviewed. Wow, all that gorgeous came right back! There are two things I clearly detect:
1. Warm, fantastic mid to upper bass. I can't emphasize enough how much difference this makes and how it impacts my subjective reviews.
2. There is a "loosness" to notes that is hard to describe but notes are separated and delightfully clean and pleasant. I get this same sense when I EQ a speaker for a room.
EDIT: later testing showed that the room mode at about 105 Hz was impacting the tonality of the speaker. Once I filtered that, the sound was very good. I have since made this part of the protocol.
Discussion and Conclusions
The objective measurements will nail the Olive score no doubt. And they present for the third time a conflict with my subjective listening impressions. Of course my subjective evaluation is much less reliable. I like us to allow some allowance for them to sink in though. Olive's latest tests shows people like to hear more bass than originally though. So could this be behind my preference for speakers like Revel M16? Here is its predicted-in room response:
I am really starting to think the 100 to 200 Hz region plays a much stronger role than we think in subjective sound a speaker produces. The other factor is not letting the higher frequencies dominate the mid-range. As I noted in the review, broad deviations in the measurements, despite their low level, may have a much larger subjective difference.
At some point we will have to reconcile these differences, either setting me straight on my subjective evaluations being wrong, or us not knowing all that Harman knows about good speaker sound. Let's remember that they won't release a new speaker unless it passes double blind listening tests against its competitor. No other score allows them to skip this test. Components are tweaked until they achieve this. So one wonders if this is not released to public.
For now, objective measurements are superb enough to give a thumbs up to KEF R3 and hence the choice of panther.
Personal note:
This pair of speaker cost US $211 to ship to me with insurance. This presented huge hardship for the owner to fund but he was adamant in doing his bit to help with our research into speaker technology. I will be paying for shipping back but like to also help him what he has paid to get them to me. If you like to support him the same way, and there is no pressure, just start a conversation with me and I will collect the funds to give him.
Last edited: