KEF Reference 5 loudspeaker Measurements

Sidebar 3: Measurements

I used DRA Labs' MLSSA system and a calibrated DPA 4006 microphone to measure the KEF Reference 5's frequency response in the farfield, and an Earthworks QTC-40 for the nearfield and in-room responses. The Reference 5's specified sensitivity is 90dB/2.83V/m; my estimate was slightly lower, at 89.1dB(B)/2.83V/m. Its impedance is specified as 8 ohms with a minimum magnitude of 3.2 ohms. Fig.1 shows how the impedance and electrical phase vary with frequency. While the impedance lies above 8 ohms in the low treble, it remains between 4 and 5 ohms throughout the midrange and bass and in the top octaves. The minimum value was 3.3 ohms between 90 and 100Hz, but as the phase angle is generally benign, the Reference 5 should work well with tube amplifiers from their 4 ohm output transformer taps.


Fig.1 KEF Reference 5, electrical impedance (solid) and phase (dashed) (5 ohms/vertical div.).

The traces in fig.1 are free from the small discontinuities that would suggest the presence of panel resonances. When I investigated the enclosure's vibrational behavior with a plastic-tape accelerometer (footnote 1), I did find some modes present at various places on the sidewalls (fig.2), but these are both very low in level and sufficiently high in frequency not to have audible consequences.


Fig.2 KEF Reference 5, cumulative spectral-decay plot calculated from output of accelerometer fastened to center of sidewall level with third-highest woofer (MLS driving voltage to speaker, 7.55V; measurement bandwidth, 2kHz).

As I mentioned in the review, Jack Oclee-Brown eventually decided on using the long ports for the top two woofers and the short ports for the bottom two woofers. The short port gives a slightly higher tuning frequency than the long—45 vs 36Hz—but in fig.3, the blue trace shows the sum of both ports' outputs. It peaks between 27 and 70Hz, and although some midrange output can be seen, this is both low in level and will be ameliorated by the fact that the ports are on the speaker's rear panel. The red trace is the sum of the woofer outputs; it covers the range of 60–300Hz, and the upper-frequency rolloff appears to be around 12dB/octave and free from any spurious response spikes. The midrange unit rolls in with closer to a third-order acoustic slope and the Reference 5's upper-frequency response is relatively smooth, which can also be seen in the response averaged across a 30° horizontal window centered on the tweeter axis (fig.4). As is usual with a reflex design, the overall low-frequency response rolls off rapidly below the port's tuning frequency.


Fig.3 KEF Reference 5, acoustic crossover on tweeter axis at 50", corrected for microphone response, with nearfield responses of midrange unit (green), woofers (red), and ports (blue), respectively plotted in the ratios of the square roots of their radiating areas below 400Hz, 2.1kHz, and 900Hz.


Fig.4 KEF Reference 5, anechoic response on tweeter axis at 50", averaged across 30° horizontal window and corrected for microphone response, with complex sum of nearfield midrange, woofer, and port responses plotted below 300Hz.

The KEF's plot of lateral dispersion, normalized to the tweeter-axis response (fig.5), reveals that the speaker smoothly and evenly becomes more directional throughout the treble, though the Reference 5 has a somewhat more restricted dispersion than the Blade Two. In the vertical plane, however, the Reference 5's response doesn't change appreciably over quite as wide a window (fig.6).


Fig.5 KEF Reference 5, lateral response family at 50", normalized to response on tweeter axis, from back to front: differences in response 90–5° off axis, reference response, differences in response 5–90° off axis.


Fig.6 KEF Reference 5, vertical response family at 50", normalized to response on tweeter axis, from back to front: differences in response 20–5° above axis, reference response, differences in response 5–20° below axis.

Fig.7 compares the spatially averaged response of the Reference 5 (red trace) with that of the Blade Two (blue). (Using SMUGSoftware's FuzzMeasure 3.0 program and a 96kHz sample rate, I average 20 1/6-octave–smoothed spectra, individually taken for the left and right speakers, in a rectangular grid 36" wide by 18" high and centered on the positions of my ears.) As one might expect from the similar drive-unit arrays, the two speakers behave very similarly in my room, with a relatively even response from the midbass through the low treble, and with the output around 30Hz boosted by the lowest-frequency resonant mode in my room. However, the Reference 5's greater directivity above 5kHz results in a response that slopes down more in the top two octaves than the Blade Two's response, and actually resembles that of the Magico S5 Mk.II, which I reviewed in February 2017 (fig.8). While a flat treble response is not what you want to see in a graph like this—a room's furnishings are more absorbent at high frequencies than they are lower in the audioband—this graph does explain why I felt the Reference 5 to sound a little sweet.


Fig.7 KEF Reference 5, spatially averaged, 1/6-octave response in JA's listening room (red); and of KEF Blade Two (blue).


Fig.8 KEF Reference 5, spatially averaged, 1/6-octave response in JA's listening room (red); and of Magico S5 Mk.II (blue).

Turning to the time domain, the KEF's step response on its tweeter axis (fig.9) indicates that its tweeter and woofers are connected in positive acoustic polarity, the midrange unit in inverted polarity. However, the decay of the tweeter's step smoothly blends with the start of the midrange unit's step, and the decay of that unit's step smoothly blends with the rise of the woofers' step. This suggests optimal crossover design. Finally, like the Blade Two, the Reference 5's cumulative spectral-decay plot on the tweeter axis (fig.10) demonstrates superbly clean decay from the midrange up.


Fig.9 KEF Reference 5, step response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).


Fig.10 KEF Reference 5, cumulative spectral-decay plot on tweeter axis at 50" (0.15ms risetime).

As always with speakers from KEF, the Reference 5's measured performance reveals excellent speaker engineering.—John Atkinson

Footnote 1: I had bought this PVDF accelerometer from Jesse Klapholz when I was investigating speaker-panel resonances for a 1992 article in Stereophile (see). Sadly, Mr. Klapholz passed away in June 2017, aged 63.
GP Acoustics (UK) Ltd.
US distributor: GP Acoustics (US) Inc.
10 Timber Lane
Marlboro, NJ 07746
(732) 683-2356

supamark's picture

The last two photos on page 2 are not of the Reference 5. The last one is the Reference 3, the one above... dunno what it is but it ain't the Referene 5.

Also, the treble rolloff compared to the Blade 2 was interesting and unexpected.

John Atkinson's picture
supamark wrote:
The last two photos on page 2 are not of the Reference 5. The last one is the Reference 3, the one above... dunno what it is but it ain't the Reference 5.

Thanks. Replaced the final image with one more appropriate; the one above it is a closeup of the loudspeaker's outrigger base.

John Atkinson
Editor, Stereophile

supamark's picture

but I think it's a closeup of the Reference 1's stand (see link):

Also, enjoyed the review, I've been curious about the differences between this and the blade 2.

Your story about the HVAC system noise reminded me of the time I recorded a visiting guitarist at The University of Texas in like '91, don't remember his name but he was quite good. I had originally gone with Neumann KM86's in crossed fig. 8 to get plenty of the room (was very nice - the reverb had a fully enveloping syrupy quality to it sitting in the audience, since ruined when they tore out the pipe organ) but forgot to take into account that it was cold/flu/allergy season - the coughing and sneezing was so bad that it was drowning out the guitarist so... at intermission (figuring the 1st half was unlistenable anyway) I ran out, moved the mics back a bit and switched them to cardiod. The artist was probably both annoyed and relieved (never heard back about it) but the 2nd half turned out quite nicely.

SMc's picture

PS UT eventually replaced the organ in the recital hall.

supamark's picture

The organ in Bates Recital Hall (interesting, adjustable acoustics in Bates) or the one in the old Music Building (the one that had the beautiful reverb)?

SMc's picture

An Aeolian-Skinner organ was installed in the old music building's Jessen Auditorium, replacing the one removed. The huge Visser-Rowland tracker in the Music Recital Hall is still going strong!

John Atkinson's picture
John Atkinson wrote:
the one above it is a closeup of the loudspeaker's outrigger base.

And I was wrong. So I have replaced that one with a cutaway picture of the Reference 5's enclosure.

John Atkinson
Editor, Stereophile

supamark's picture

ain't nobody perfect.

Edit: lookin' through, like the new photos/progression.

Ortofan's picture

... wave-guide/phase-plug an Altec-Lansing invention?

Axiom05's picture

Just curious, what is the magnitude of your resonance at ~30 Hz without smoothing? Do you have any issues with clipping at that frequency when taking your measurements with FuzzMeasure, if so, how do you deal with it to assure that the rest of the spectrum is not affected (maybe it doesn't matter)?


blang11's picture

I struggle to remember the last speaker review I read in Stereophile that described the sound as a bit too sweet. I recall the majority are either neutral (yay!) or overly bright or the tweeter is simply a few db too high in level. As someone who has recently moved into a room that's a bit bright, I now recognize that a bit of sweetness can be just what the doctor ordered, especially for those who recordings aren't impeccable. Great review and video interview!

foxhall's picture

I've not yet seen them in person but they look stunning in the pictures. The drivers are a work of art alone.

tonykaz's picture

These things are Beautiful.

As long as they sing nearly as well as they look they'll be big hits ( especially with the Wife's decorator).

B&O stuff is kinda overly strange looking.

Is there a more beautiful loudspeaker ?

Tony in Michigan

ps. betcha the rest of the UK speaker outfits are a bit envious.

allhifi's picture

T.I.M.: If the REF-5's are simply an extension of the REF-1's, it's a superb loudspeaker.

Personally, I believe the Reference 3 may be the best choice -and sounding.

I urge you to download and read KEF's excellent (informative) White Paper that at once is both technically astute and easy-to-read.

Not many expensive loudspeakers (to my knowledge) have and/or reveal some basic design elements; including enclosure/baffle board material composition, bracing choices, driver fastening method (KEF REF uses 6-8 BOLTS per driver -with the critical MF/HF driver no less bolted to a steel plate) and or crossover considerations -and implementations.

This and far more is discussed in this excellent White Paper.

If the equipment preceding the KEF (Reference) is up to snuff, you will be left thrilled by its sheer accuracy, delicacy, nuance, explosive dynamics and sheer realism.

peter jasz

nirodha's picture

Hi John,
I have been using the 207/2 for a loooong time and they still do it for me. How do the 2 loudspeakers compare? Bit worried about the build quality of the ref. 5. The 207/2's are build like a tank.
Cheers and all the best wishes for 2018! :-) Wim

John Atkinson's picture
nirodha wrote:
I have been using the 207/2 for a loooong time and they still do it for me. How do the 2 loudspeakers compare?

I reviewed the R207/2 in February 2008 and my system has undergone many changes since I last listened to it. But I think you still have many years ahead of you with the R207/2s. Keep 'em!

John Atkinson
Editor, Stereophile

nirodha's picture

Thanks John, keep up the good work! Wim

JimboJumbo's picture

They are a great loudspeaker.

However, in Figure 9 where you say the results suggests optimal Xover design . . .

I am not sure about that. Agreed the transition between high frequency driver and mid frequency driver, and also the transition between mid and low frequency drivers, are all reasonably smooth. But, the high frequency driver and mid frequency driver are also completely out of phase and (whilst I can see why it has been implemented that way) I am not sure that is optimal; even though it is probably not audible.

For instance, any discontinuities between the mid frequency and high frequency driver will be exacerbated by having them out of phase; and the avoidance of acoustic scattering effects at the surrounds and other areas of discontinuity within KEF’s coaxial design has been one of the evolution drivers for a long time. Perhaps you are saying the Xover design is optimized for the type of crossover/order that is implemented?

Additionally, your comment on Figure 7 . . .

“However, the Reference 5's greater directivity above 5kHz results in a response that slopes down more in the top two octaves than the Blade Two's response, and actually resembles that of the Magico S5 Mk.II, which I reviewed” . . . .

Was initially a little confusing. However, I think what you mean (please advise) is that the reference 5’s greater dispersion (rather than directivity) comes at the expense of on axis high frequency SPL.

To me, it seems there has been a design compromise, where some on axis high frequency SPL/response has been sacrificed so that off axis high frequency SPL/response performs better.

Appreciate its probably a bit of semantics between engineers, but do I have that correct. Because if I don’t then I need to know how the red trace in Figure 7 represents greater directivity.

DaveinSM's picture

While I was also surprised at the fact that the midrange is in negative polarity to the tweeter and woofer, I would think that any negative effects might be mitigated by the fact that the tweeter is located in the center of the midrange in a concentric fashion.

So this speaker design could still be time coherent (?) but not phase coherent. Even if they drivers were in the same phase, this design still wouldn't be phase coherent because the driver crossovers aren't 1st order.

I have just gotten the Reference 1 and think they sound great with great imaging. But coming from the Thiel CS3.6, which is both time and phase coherent with 1st order crossovers, I don't hear any major differences in imaging.

However, one interesting thing I do seem to be noticing is that the KEFs are a bit more forgiving of not only listening position, but also of phase in recordings. As you may know, lots of original recordings were recorded out of phase, and reversing the phase on your preamp can often make those recordings sound better. I really noticed the difference on certain recordings with the Thiels, not nearly as much so with the KEFs.

Axiom05's picture

There appears to be something odd with Figure 6, the +5 & -5 responses are flat lines while the 0 degree line is not normalized.