KEF R11 Measurements from Soundstage/NRC

B

Beave

Audioholic Chief
The measurements from the NRC chamber don't account for rear ports on speakers.

So what they show is the output from the woofers only.

The port output should fill in that dip as expected.
 
D

D Murphy

Full Audioholic
It's a little more complicated than that. The anechoic measurements will capture at least some of the rear port's output, but that output above 50 Hz will wrap around the speaker and arrive out of phase with the bottom of the woofer's output, which causes the dip that you see in the R11's measurements. It's particularly a problem when there's strong output radiating from the rear. Check out the otherwise superb NRC measurements of the Dutch and Dutch monitors, which have two 8" woofers on the back. Look familiar?
https://www.soundstagenetwork.com/index.php?option=com_content&view=article&id=2073:nrc-measurements-dutch-dutch-8c-loudspeakers&catid=77:loudspeaker-measurements&Itemid=153

I had a long back and forth with the NRC engineer concerning why my BMR bass measurements were so screwy. He never could really explain all of the problems, and I'm not totally convinced that he understood everything that was going on between 50 Hz and 100 Hz, but it definitely had to do with phase issues. In any event, The NRC chamber isn't large enough to allow high accuracy below about 80 Hz. But above that point, the measurements for the R11 are certainly admirable, particularly the distortion products at 90 dB, which are about the lowest I've seen.
 
B

Beave

Audioholic Chief
Yeah, you're right - it's more complicated than what I wrote.

I've never read exactly what the issues are, but I do know that Dr. Toole has written that their bass measurements are more reliable for sealed or front ported speakers, and less reliable for rear ported speakers.

He has also mentioned that the chamber is probably long overdue for a recalibration of their bass measurements. Harman recently did some recalibration for bass results from their chambers.

I think you're right that the issues stem from port output wrapping around and adding to - or canceling out - some of the output from the front woofers. It probably depends on frequency, cabinet size of the speaker (depth in particular, but perhaps width to a lesser extent?), size and location(s) of ports, etc.
 
S

shadyJ

Speaker of the House
Staff member
It's a little more complicated than that. The anechoic measurements will capture at least some of the rear port's output, but that output above 50 Hz will wrap around the speaker and arrive out of phase with the bottom of the woofer's output, which causes the dip that you see in the R11's measurements. It's particularly a problem when there's strong output radiating from the rear. Check out the otherwise superb NRC measurements of the Dutch and Dutch monitors, which have two 8" woofers on the back. Look familiar?
https://www.soundstagenetwork.com/index.php?option=com_content&view=article&id=2073:nrc-measurements-dutch-dutch-8c-loudspeakers&catid=77:loudspeaker-measurements&Itemid=153

I had a long back and forth with the NRC engineer concerning why my BMR bass measurements were so screwy. He never could really explain all of the problems, and I'm not totally convinced that he understood everything that was going on between 50 Hz and 100 Hz, but it definitely had to do with phase issues. In any event, The NRC chamber isn't large enough to allow high accuracy below about 80 Hz. But above that point, the measurements for the R11 are certainly admirable, particularly the distortion products at 90 dB, which are about the lowest I've seen.
I don't know how the NRC handles low frequencies exactly. I know other testers will either do a groundplane measurements and splice it in with anechoic, or close mic measurements of bass drivers and port and then splice it with anechoic, or just use a transfer function on anechoic low-frequency response derived from groundplane testing. One of the reasons why I don't splice the groundplane low-frequency response in with the free-air measurements is because the testing methods don't really fully reconcile with each other.
 
D

D Murphy

Full Audioholic
Yeah, you're right - it's more complicated than what I wrote.

I've never read exactly what the issues are, but I do know that Dr. Toole has written that their bass measurements are more reliable for sealed or front ported speakers, and less reliable for rear ported speakers.

He has also mentioned that the chamber is probably long overdue for a recalibration of their bass measurements. Harman recently did some recalibration for bass results from their chambers.

I think you're right that the issues stem from port output wrapping around and adding to - or canceling out - some of the output from the front woofers. It probably depends on frequency, cabinet size of the speaker (depth in particular, but perhaps width to a lesser extent?), size and location(s) of ports, etc.
I agree with you on all counts. The inaccuracy of the NRC's bass measurements is really frustrating. You can see how much difference the measurement method makes when you compare the NRC's direct frontal measurement of the Dutch and Dutch with Stereophile's measurement of that speaker, where the port output is measured nearfield and then spliced to the woofer's output. The 60 Hz dip completely disappears.
https://www.stereophile.com/content/dutch-dutch-8c-active-loudspeaker-system-measurements
 
S

shadyJ

Speaker of the House
Staff member
The Soundstage/NRC measurements for the KEF R11 model are available now:

KEF R11 measurements
Nor surprising that these speakers are a bit heavy in the bass driver band seeing as how it has four of them. It measures well and should sound pretty good, perhaps a tad warm. Looks like a very good speaker, but it should be for $5k/pair.
 
S

shadyJ

Speaker of the House
Staff member
I agree with you on all counts. The inaccuracy of the NRC's bass measurements is really frustrating. You can see how much difference the measurement method makes when you compare the NRC's direct frontal measurement of the Dutch and Dutch with Stereophile's measurement of that speaker, where the port output is measured nearfield and then spliced to the woofer's output. The 60 Hz dip completely disappears.
https://www.stereophile.com/content/dutch-dutch-8c-active-loudspeaker-system-measurements
Stereophile isn't perfectly capturing the low-frequency response with respect to that of the rest of the band, but they are open about the shortcoming in their method.
 
AcuDefTechGuy

AcuDefTechGuy

Audioholic Jedi
Listening window 200Hz-10kHz Response is +/-2.5dB. Off-axis looks pretty good. Distortion looks low.

Looks pretty good overall to me.
 
D

D Murphy

Full Audioholic
Stereophile isn't perfectly capturing the low-frequency response with respect to that of the rest of the band, but they are open about the shortcoming in their method.
The only problem they acknowledge is an apparent elevation in the bass response that doesn't actually exist--it's just an artifact of the different measuring distances--near field for the port, 1+ meters for the whole speaker above 200 Hz. I've never understood why they can't just scale the port measurements down to account for that disparity.
 
S

shadyJ

Speaker of the House
Staff member
The only problem they acknowledge is an apparent elevation in the bass response that doesn't actually exist--it's just an artifact of the different measuring distances--near field for the port, 1+ meters for the whole speaker above 200 Hz. I've never understood why they can't just scale the port measurements down to account for that disparity.
One problem is how do you match near-field port output with the rest of the speaker output? The measurement methods are too different and can not be perfectly mated. There is no way to reliably scale the port output down to match the measurements of the 1 meter measurements, because there is no sure way to know how by how much to scale it down.
 
D

D Murphy

Full Audioholic
One problem is how do you match near-field port output with the rest of the speaker output? The measurement methods are too different and can not be perfectly mated. There is no way to reliably scale the port output down to match the measurements of the 1 meter measurements, because there is no sure way to know how by how much to scale it down.
Since we're essentially working with anechoic measurements (which a nearfield basically is),and not worrying about room effects, I don't think the scaling problem is all that difficult. It's just a function of the measuring distance. I think the bigger question is whether the splicing is done correctly, because there are a lot of assumptions you have to make. When I design speakers, I don't dice and splice. My measurements are taken at a little over one meter for each driver, and my measuring software (Praxis) transitions from an anechoic windowing to a longer room window starting at around 300 Hz. Those room measurements of the woofer also include the port's contribution that is heard from the front of the speaker. I've come to know what my room does to bass response, and I've developed a pretty good grasp of bass profile will work best in the field. That sounds loosey goosey, but there are serious problems with port splicing and ground plane measurements when it comes to knowing how the speaker will actually sound in a typical room.
 
Last edited:
KenM10759

KenM10759

Audioholic Samurai
I have auditioned the KEF R11 speakers a few times now. If I hadn't recently escaped a high-pressure job in favor of a no-pressure one with lower pay, I'd be listening at home by now.

They are to my ears, incredibly close to the Reference 3 for 1/3rd the price.
 
S

sfdoddsy

Audiophyte
The only problem they acknowledge is an apparent elevation in the bass response that doesn't actually exist--it's just an artifact of the different measuring distances--near field for the port, 1+ meters for the whole speaker above 200 Hz. I've never understood why they can't just scale the port measurements down to account for that disparity.
I’m 1000% with you Dennis.

Every single measurement Stereophile does carries the proviso that the bass measurements are not accurate due to their measurement technique.

Fine. The measurement technique they use is due to them doing it in JA’s lounge or where ever.

But if you have a consistent error that you introduce, and it is in every single measurement, just scale for it!

I love that Stereophile do measurements.

I hate that you get better idea of bass performance from the 1/3 average in-room test they sometimes do.

As for the R11 measurements, assuming measuringin the NRC chamber allows consistency with previous measurements (bass ports aside),those. U, ers don’t seem to be any improvement over the previously measured KEFs.

The Ref 201 retains its crown.
 
Last edited:
A

aarons915

Enthusiast
As for the R11 measurements, assuming measuringin the NRC chamber allows consistency with previous measurements (bass ports aside),those. U, ers don’t seem to be any improvement over the previously measured KEFs.

The Ref 201 retains its crown.
I'd say the KEF Reference 1 hold the crown actually and not just for KEF but pretty much any speaker I've seen in the NRC measurements. The R11 actually measures pretty close to the Reference series for a much cheaper price, which is pretty impressive. The R3 at $2k measures even better if you check the R series whitepaper.The new R series are a bit laid back but since the On-axis and off-axis sound is so balanced, they're the perfect speaker to shape however you want with EQ.
 
Swerd

Swerd

Audioholic Warlord
In any event, The NRC chamber isn't large enough to allow high accuracy below about 80 Hz
Stereophile isn't perfectly capturing the low-frequency response with respect to that of the rest of the band, but they are open about the shortcoming in their method.
The only problem they acknowledge is an apparent elevation in the bass response that doesn't actually exist--it's just an artifact of the different measuring distances--near field for the port, 1+ meters for the whole speaker above 200 Hz. I've never understood why they can't just scale the port measurements down to account for that disparity.
I think the take-home message is that neither the NRC nor JA's measurements published in Stereophile are perfect. They seem to be the best speaker measurements available, but maybe there is no single method of measuring bass response that can't be fooled or introduce a measurement artifact.

I also agree with Dennis that Stereophile might be better off if they simply "scale down the port measurements to account for that disparity", instead of publishing frequency response curves with misleading bass. They always print their explanation of the built-in error in their method, but too many readers may ignore that statement, or don't understand it's impact.

It may be as simple as determining an average error and subtracting it. There may be different 'fudge factors' depending on whether the speakers are sealed or ported, and whether the port is in the front or rear, but I think it could be done. If it can't be so easily done, it would be of interest to hear why not, from JA himself.
 
Last edited:
newsletter

  • RBHsound.com
  • BlueJeansCable.com
  • SVS Sound Subwoofers
  • Experience the Martin Logan Montis
Top