Bowers & Wilkins 805 D3 loudspeaker Measurements

Sidebar 3: Measurements

I used DRA Labs' MLSSA system and a calibrated DPA 4006 microphone to measure the Bowers & Wilkins 805 D3's frequency response in the farfield, and an Earthworks QTC-40 for the nearfield and in-room responses. The B&W's voltage sensitivity is specified as 88dB/2.83V/m; my estimate was slightly higher than this, at 88.4dB(B). The nominal impedance is 8 ohms; the solid trace in fig.1 reveals that the magnitude drops below 8 ohms between 100 and 750Hz and above 8kHz, with minima of 4.6 ohms at 180Hz and 22kHz. The electrical phase angle (fig.1, dotted trace) is occasionally extreme, with combinations of 6 ohms and –35° at 118Hz and 8.2kHz. Tube amplifiers will probably work best with this speaker when used from their 4 ohm output-transformer taps.


Fig.1 Bowers & Wilkins 805 D3, electrical impedance (solid) and phase (dashed) (2 ohms/vertical div.).

Although it probably can't be seen at the scale this graph is printed in the magazine, there is a very slight discontinuity around 900Hz in the impedance traces. However, when I investigated the enclosure's vibrational behavior with a simple plastic-tape accelerometer I found no major problems in this region, though I did find a moderately strong mode on the curved sidewalls at 520Hz, and two lower-level modes at higher frequencies (fig.2).


Fig.2 Bowers & Wilkins 805 D3, cumulative spectral-decay plot calculated from output of accelerometer fastened to center of sidewall (MLS driving voltage to speaker, 7.55V; measurement bandwidth, 2kHz).

The impedance-magnitude trace suggests that the large, flared port on the front baffle is tuned to 50Hz or so. The minimum-motion notch in the woofer's nearfield output (fig.3, blue trace) occurs at 48Hz, and the port's response (red) peaks between 35 and 90Hz in classic fashion. (I haven't plotted the port's output above 500Hz, as the measurement at higher frequencies was contaminated by the woofer's output.) The woofer's response is relatively even (though with a couple of small peaks and dips visible) up to 3.3kHz, where it crosses over to the tweeter (green trace). The crossover filter slopes appear to be low-order, as the complementary rolloffs of the two drive-units are initially gentle. The tweeter is balanced up to 5dB too high in level on its axis, but the primary dome resonance, and the usual dip below that resonance by a tweeter with a stiff, pistonic diaphragm, are both above the audioband.


Fig.3 Bowers & Wilkins 805 D3, acoustic crossover on tweeter axis at 50", corrected for microphone response, with nearfield responses of woofer (blue) and port (red), respectively plotted in the ratios of the square roots of their radiating areas below 300 and 500Hz.

The elevated tweeter output in the top two audio octaves can be seen in the 805 D3's farfield output, averaged across a 30° horizontal window (fig.4), and a couple of small suckouts are visible in the upper midrange and low treble. The apparent rise in the upper bass in this graph is an artifact of the nearfield measurement technique.


Fig.4 Bowers & Wilkins 805 D3, anechoic response on tweeter axis at 50", averaged across 30° horizontal window and corrected for microphone response, with complex sum of nearfield woofer and port responses plotted below 300Hz.

The plot of the B&W's lateral dispersion, normalized to the tweeter-axis response (fig.5), has impressively even contour lines—other than a well-controlled reduction in level, the speaker's balance doesn't change in an aggressive manner to its sides up to 8kHz or so, when the tweeter's increased directivity makes its presence known. Vertically, the use of low-order crossover filters means that a large suckout develops in the crossover region 5° below and 10° above the tweeter axis (fig.6).


Fig.5 Bowers & Wilkins 805 D3, 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 Bowers & Wilkins 805 D3, vertical response family at 50", normalized to response on tweeter axis, from back to front: differences in response 45–5° above axis, reference response, differences in response 5–45° below axis.

Fig.7 shows the B&W 805 D3's spatially averaged response in my room. (I average 20 1/6-octave–smoothed spectra, individually taken for the left and right speakers using SMUGSoftware's FuzzMeasure 3.0 program and a 96kHz sample rate, in a rectangular grid 36" wide by 18" high and centered on the positions of my ears. This mostly eliminates the room acoustic's effects.) The balance is not as smooth as that of the Aerial 5T, which I also reviewed for this issue, and there is both a slight lack of energy in the presence region and a slightly boosted mid-treble.


Fig.7 Bowers & Wilkins 805 D3, spatially averaged, 1/6-octave response in JA's listening room.

The red trace in fig.8 is again the spatially averaged response of the 805 D3 in my room. The excess of mid-treble energy is apparent when compared both with the BBC LS3/5a (blue trace) and the KEF LS50 (green), both measured under identical conditions. However, the B&W lacks the LS3/5a's small peak between 1 and 2kHz, which adds a touch of nasality to that vintage speaker's sound. The LS50's in-room response shelves down in the top octaves compared with the B&W; in that respect the KEF's in-room behavior resembles that of the Aerial 5T. Both speakers sound less "brilliant" than the B&W, but I believe this is actually more neutral in-room behavior, given the increased absorptivity of the room furnishings in the high treble. The 805 D3 has a little more upper bass than the LS3/5a, but, like the KEF, its ported alignment results in a faster rolloff below the midbass.


Fig.8 Bowers & Wilkins 805 D3, spatially averaged, 1/6-octave response in JA's listening room (red); of KEF LS50 (green); and of BBC LS3/5a (blue).

Turning to the time domain, the Bowers & Wilkins' step response on the tweeter axis (fig.9) reveals that both drive-units are connected in positive acoustic polarity, but with the tweeter's output leading that of the woofer. In fact, the very slight discontinuity just before the 4ms mark suggests that the drive-units blend best if the listener's ears are slightly below the tweeter axis—although, as fig.5 showed, if you sit much lower, a suckout develops in the crossover region. The cumulative spectral-decay or waterfall plot on the tweeter axis (fig.10) reveals a superb lack of delayed energy across the audioband.


Fig.9 Bowers & Wilkins 805 D3, step response on tweeter axis at 50" (5ms time window, 30kHz bandwidth).


Fig.10 Bowers & Wilkins 805 D3, cumulative spectral-decay plot on tweeter axis at 50" (0.15ms risetime).

The B&W 805 D3's measured performance indicates that it has a somewhat "tailored" frequency response in the treble, but in all other respects there is nothing amiss.—John Atkinson

B&W Group Ltd.
US: B&W Group North America
54 Concord Street
North Reading, MA 01864
(978) 664-2870

tonykaz's picture

This will fascinating to read.

I just had access to a pair of Pro-Ac Tabletts & a pair of old Magnapan MG2s. I discovered both are still superb and exciting.

How much driver improvement is a darn good question for some brave soul to tackle, I give you full marks for suggesting it. I'll buy a second year's print subscription if you can answer it using your typically engineering based methods. Who else "could" handle this conundrum?

By the way, I was stunned by you calling-out the 21bit DAC design as being obsolete. Once again, I applaud you!

Tony in Michigan

ps. I'm about to go European and embrace all music delivery systems by buying a Linn DS server. I might even invest in a LP12 ( for old time's sake ) which I'll probably display in a sealed, nitrogen filled glass case. I'm hunting for an Ittok and Rosewood Red Koetsu ( not, but maybe a cute red nosed Asak that doesn't have to work ).

mrkaic's picture

I keep reading about discontinuities in John Atkinson's (extremely good) articles. What is shown in Fig. 1 is not a discontinuity, the function appears to be continuous. Rather, the kink at 900 Hz could be a discontinuity in the 1st derivative of the impedance function wrt frequency. It is an important difference.

jmsent's picture

There certainly is a discontinuity in the impedance curve, which would normally not have such a "kink" unless there's a resonance somewhere in the system. Are you suggesting it's a measurement artifact? Having been involved in the driver industry for many years, (now retired) I'd lay odds that this kink is due to an edge reflection in the woofer, due to the use of a low loss rubber surround. It's a common occurrence that will often not be easily visible in a frequency response measurement. Given that JA didn't detect any mechanical resonances around that frequency, it's likely that it originates in the driver itself.

mrkaic's picture

Calling that kink at 900 Hz a discontinuity is not proper mathematical terminology. Loosely speaking, a discontinuity means that the curve is cut -- there is a "gap". This clearly is not the case here. But the slope (the 1st derivative) is not continuous -- has different values if you approach the kink point from the left and from the right.

hb72's picture

--> "inflection point"

John Atkinson's picture
mrkaic wrote:
Calling that kink at 900 Hz a discontinuity is not proper mathematical terminology. Loosely speaking, a discontinuity means that the curve is cut -- there is a "gap". This clearly is not the case here.

I think you are being misled by the relatively sparse data points in the impedance graph. If I had taken reading at, say 1Hz, intervals, then you would have seen a clearly defined discontinuity, equivalent to the Hilbert transform of a peak.

John Atkinson
Editor, Stereophile

mrkaic's picture


1. First, many thanks for your splendid articles. I buy Stereophile just to see your measurements -- they are the only thing worth reading in this magazine and a splendid antidote to poisonous subjectivist propaganda.

2. A speaker is a linear electric motor with a bunch of circuitry inside (+ the box, of course). Mathematically it is described as a system of coupled second order differential equations. The solutions to such equations are typically continuous -- have no breaks. That is one of the reasons why I'm wondering about the terminology.

3. It would be great if you could show more detailed scans of (all) speakers on the website. (I know it cannot be done in print) I think I am not the only reader who would appreciate that.



Richard D. George's picture

Would it have made a difference with the treble issues if the speaker cables were bi-wired?

Would it have made a difference if the amplification was Classe Audio?

arnolf's picture

I am looking forward for the review of the 805 D4...
Seriously I have been an avid reader of Stereophile for many years and I think there is enough diversity on the speaker marker to not review the same speaker forever.

Staxguy's picture

Which stereo can reproduce the sound of the piano? The Fidelity Acoustics RFM-1 Bookshelf Loudspeaker ($4000.00 / pair), a notable speaker employing the ScanSpeak Revelator Ring Radiator Tweeter (same as a Magico M5) tweeter was not able to do so.

Is there even one stereo system which can do the job properly? The Steinway Lyingdorf Model D, by virtue of Steinway heritage may be able to. The Bösendorfer VC 7 may be one as well.

We've listened to quite a number of good loudspeakers. The composing is mind-absorbing.

The Bösendorfer VC7 Loudspeaker ($25,000.00 USD) uses resonant plates ("Acoustic Sound-Boards"), mimicking a piano's sound board. The use of two tweeters is an interesting feature. It's almost proto-Genesis (the speaker, not the band), so to say. Sony uses two tweeters too (surrounding) on it's Sony SA-NA2ES ($5,000.00 USD), and coincident models, although not on their Sony SA-AR-1 and Sony SA-AR-2 ($20,000.00 USD) Loudspeakers, too.

Reproducing the piano is a glorious extremity. The frequencies on the keys themselves (on a non-Bösendorfer Imperial Piano ($200,000.00 USD), which has lower) go from 27.5 Hz (A0) to 4,186.01 Hz (C8), standard (equal temperament) tuning. This is ignoring the over-tones, and the sound of the organ stops. An organ will go lower, to 16 Hz. There is ignoring foot-pedal work, also - the thuds and jumps. Now in terms of latant response, 60 - 100 dB.

There can be reproduction with some over-play.

The Steinway Lyngdorf Model D goes for $208,000.00 USD today (and was introduced for $150,000.00 USD in 2007).

Most pianists will prefer the Steinway Model D, which is tuned at 442 Hz, standard, although some will go for a Fazzioli F308.

We'd prefer the Bösendorpher Imperial Grand, in gold leaf, though are settling with two bottles of 311 Helles Lager by Coal Harbour Brewing Company, CHBC, with which to write this review.

There was a Fazzioli, the Brunei, which mimicked the sound of the Bosendorpher IG by presenting a darker sound, though we have not heard it. We would like one, however. Perhaps with less in-lay. :)

How you make your piano is up to you.


Allen Fant's picture

I have listened to all of the iterations of this 805 speaker over the years and it never disappoints. A true staple in the B&W catalog. Match it w/ a REL subwoofer and you will be musically rewarded.

Richard D. George's picture

Have you listened to the D3? If so, did you notice the elevated upper treble cited in this review? I did not.

Regarding REL subs, a reviewer of the 805 D3 in a different publication mentioned that it works well with a pair of REL subs. He used a pair of REL T7i's with the 805 D3's with great success.

Allen Fant's picture

Hello Richard,

no- I did not hear any elevated upper treble either.
I will have to search for the other review w/ the REL sub added.

For those whom do not like the REL brand, only Sunfire, would be my second subwoofer choice. Happy Listening!

Richard D. George's picture

Hello Allen:

Reviewer is Neil G. The mention of twin REL T series subs is in the comments section of the online version of this (unnamed) magazine.

I have four S series REL subs in three systems with loss-less Longbow wireless speaker level connections and two other T series REL subs connected speaker level with cables. Happy with all of them. The speaker level connection provides a seamless "feathering" with the main speakers.

hb72's picture

the 805d3 officially reaches down to 42Hz at +/-3dB, while the LS50 officially reaches only to 79Hz +/-3dB, BUT comparing in-room response traces and focussing on frequencies below 80Hz the LS50 seems approx. 2-5dB louder than the 805.
While this might be confirmed by listeners to some extent, perhaps in-room responce curves are not exactly levelled to same percieved volume for all 3 speakers?

Les's picture

Wow, subscribing to Stereophile just for the measurements... That's quite depressingly mind-blowing.

hb72's picture

I have to say I really like how measurements are done by JA, and I usually read them in all detail before I read the critique i.e. subjectivly percieved SQ.

Richard D. George's picture

I went back and compared Fig 3 from the measurements related to the Stereophile review of the 802 D3 to Fig 4 from the measurements related to the review of the Stereophile 805 D3. Not wildly different as both have elevated values in the upper treble.

Yet no mention of elevated / exaggerated treble in Kal's listening comments for the 802 D3. Also no mention of this in other professional reviews of the 805 D3.

Kal used bi-wired cables for the 802 D3 review.

John Atkinson's picture
Richard D. George wrote:
I went back and compared Fig 3 from the measurements related to the Stereophile review of the 802 D3 to Fig 4 from the measurements related to the review of the Stereophile 805 D3. Not wildly different as both have elevated values in the upper treble.

Yet no mention of elevated / exaggerated treble in Kal's listening comments for the 802 D3.

Perception of treble is dependent not just on the tweeter balance but also on the speaker's low-frequency extension and tuning.

John Atkinson
Editor, Stereophile

Richard D. George's picture

... I sure did not hear it and there are no mentions of this in other professional reviews.

We don't have your golden ears. This might save me money.

Art Vandelay's picture

Since the Nautilus range, all 800 series are a bit hot on axis, but that's partly because the designed listening axis is about 20 degrees off the H axis (and just below tweeter height).

Additionally, when seated 3-4 metres away and in a typical room with drapes, carpet, and a sofa, the treble is brought into balance.

That said, with room correction software and dsp tools etc so commonly used these days I can understand why B&W haven't bothered to include a tweeter level adjustment to compensate for the environment.

DougM's picture

Interesting that except for the narrow 2db peak around 1KHz, the LS3/5a is significantly smoother from 500-20Khz than either the KEF or B&W in JA's room.

SoundAdvocate's picture

Dear Mr. Atkinson:

Did you ever proceed with the comparison of the John Bowers Silver Signature to the 805 D3? As a Silver Sig owner myself, this potential comparison sure piqued my interest!

Many thanks,

brams's picture

Clearly measurements need to be interpreted carefully. How much does the elevated treble contribute to the perception of "astonishing" mid-range/treble clarity and dynamics? True resolution or psycho-acoustic trick? Ultimately, does it even matter how the subjective performance is achieved?

My opinion is that a component should measure well and sound good to be ranked highly. I suspect that we are still struggling to decide exactly what "measure well" means.