B&W 800D4 series

[richard12511]

For a tweeter, does diamond have real advantages over beryllium? I know it's stiffer, but it's also heavier. What situations would you want to go with a diamond tweeter over a beryllium tweeter?

 

[preload]

For a tweeter, does diamond have real advantages over beryllium? I know it's stiffer, but it's also heavier. What situations would you want to go with a diamond tweeter over a beryllium tweeter?

Diamonds are forever.

 

[thewas]

Diamonds are forever.

And diamonds are a audiophile's best friend.

 

[MediumRare]

Because the F1 drivers exceptional understanding is subjective based on how they feel in the motorcar as it navigates the course.

That's right. So, in contrast to the comment by @mt42, that is fully sufficient for their purpose. They don't need the math to maximize the performance of their application because the tools they have are sufficient.

 

[Deleted member 27948]

Indeed, ground plane measurements can be very accurate (for the bass), more accurate than an anechoic chamber. Hartman even uses them for that purpose. Judging by Erin’s ground plane measurement pictures, he definitely has at least 15m+(likely much more) on all sides.

Hopefully Erin ends up an 801D4 to measure, so we can see exactly how accurate the spec is.

It helps to look at the ripples on the top of an FR plot, both on-axis, and off-axis. If they are "broadly speaking" periodic, and say, the period is 200Hz, the loudspeaker is a bookshelf, then these ripples are more likely an artifact of measurement, and they point out that there is a secondary path or audio propagation which is ~1.6m longer than the direct path. For 1dB ripple, the secondary path is about -20dB re direct.

It also helps to look at the directivity diagram at low frequencies. It must be in good agreement with the theory for the size of the piston. If you see that an 8" woofer loudspeaker becomes strongly directional at 200Hz then it is more likely a measurement artifact.

It also helps to know that the guys at Clippel are much less confident in the LF precision of their system than ... too many reviewers who use their equipment. Surely, not 100%.

Generally, there are very many things that can go wrong in the process of acoustic measurements regardless if you are aware of them or not.

 

[ctrl]

My 2c: The LF FR response above is, obviously, from a truncated IR, measured by the "semi-anechoic" technique. You may ask why "obviously"? Sorry, I am not capable of explaining that to non-PhDs in a forum post.

The measurements of the 800D3 were performed in an anechoic chamber whose measurement limit is 100Hz. For the frequency range below 100Hz, near-field measurements were performed and then automatically combined with the far-field measurement in the software.
So, no truncated In-room measurement.***
*** Update: or did you mean a gated measurement, then yes.

But it is clear, errors can occur when combining near and far field measurements - e.g. mixing up measurements.

Generally, measuring such low-frequency as 20Hz requires open-field conditions in a sphere of > 15m diameter.

No it doesn't. If used correctly, this also works with the Klippel NFS or with a ground plane measurement (the nearest boundary surfaces only have to be appropriately distant) and even the combination of far-field and near-field measurement is able to achieve satisfactory results if performed correctly.

As an example, here is a simple simulation of a 15'' subwoofer in a closed 140L cabinet in VCAD and in comparison the real measured combined near field and far field measurement of the finished loudspeaker (both with baffle step correction).
Above 100Hz the 15'' subwoofer no longer behaves ideally, hence the deviation from the simulation. But below 100Hz the combined nearfield-farfield measurement agrees very well with the simulation (below 20Hz my measurement microphone is no longer calibrated).
As you can see, it works without PhD and 15m tower.

The suitable drivers are not uncommon, see https://www.focal-america.com/wp-content/uploads/2014/02/FT_access_25A4.pdf Put it in a 100+L vented box with a port tuned, say, at 20Hz, and voila!

If it's that easy, let's just do it. Let's see how close a 801D4 clone will get us to the f3 of 15Hz at [email protected] in free field (anechoic). One never believes me here without all-embracing proof

First, we simulate the two subwoofers in half space (2pi) with 100L + 100L volume and a BR port tuning of 20Hz - as specified by @mt42. With a series resistor of 0.5 Ohm (representing the resistances of cable and crossover) the drivers in parallel configuration reach a minimum of 2.5 Ohm (this is not good, but somehow we can still bring this to 3ohm when the clone is built ).



In half-space we reach the targeted [email protected].
f3 = 18Hz and a f6 = 16Hz. If we could stop here, the specification of the 801D4 (800D3) would be pretty well met. Unfortunately, if we don't want to use the 801D4 as in-wall speakers, we'll have to move on.


Now we still have to take into account the transition into the full space (into free field). So now comes the baffle step correction based on the speaker baffle size.

At first glance, this looks worse than it is. The blue curve shows the ideal expected frequency response of the driver without crossover in the loudspeaker cabinet in free field.
When the filter slope of the woofer is formed in the crossover, the capacitive part of the driver impedance forms a resonant circuit together with the series-connected coil, which leads to the formation of a hump in the frequency response around the center frequency of the resonant circuit.
People with a PhD in electrical engineering can explain this more precisely

I have drawn this roughly as an orange curve. The hump can also be more pronounced - this is only an estimate by me.

In yellow I have drawn the sound pressure at 15Hz. This is with ideal drivers and the TSP data of the Focal 25A4 subwoofer just 78dB at 2.83V.
If the rest of the speaker really plays with [email protected], then f12 = 15Hz and not f3 as stated in the specification - voila! The laws of physics still apply.

Our 801D4 clone would have an f6 of 25Hz at 90dB sensitivity. The value for f3 can only be estimated by a large margin (and depends strongly on the interaction of driver impedance and crossover coil) and should be somewhere between 35-45Hz at 90dB sensitivity.

Since almost nothing is standardized with regard to the manufacturer's specifications, B&W can elegantly pull out of the matter and say "Wait a minute, our specifications refer to a speaker that is standing in the room and not as usual on the basis of anechoic measurements".

With the help of a boundary surface, such as the floor, the speaker comes close to the values in the specification:

 

[Deleted member 27948]

Dear ctrl, I apologize if my words made you angry, it was not intentional.

Your perspective on "agree very well" is quite different from mine.

I have a DIY sub build on the older cousin, 30A4, closed box, ~200L. It was designed for a specific room where the room's (0,0,0) mode takes care of the second-degree roll-off. I think it goes too low for classical music I listen to, but I did not bother to measure it - either way, I am not going to re-design it;-)

I truly appreciate it when people disagree with me because it helps me localize my bugs & mistakes & invalid assumptions.

 

[richard12511]

The measurements of the 800D3 were performed in an anechoic chamber whose measurement limit is 100Hz. For the frequency range below 100Hz, near-field measurements were performed and then automatically combined with the far-field measurement in the software.
So, no truncated In-room measurement.

But it is clear, errors can occur when combining near and far field measurements - e.g. mixing up measurements.



No it doesn't. If used correctly, this also works with the Klippel NFS or with a ground plane measurement (the nearest boundary surfaces only have to be appropriately distant) and even the combination of far-field and near-field measurement is able to achieve satisfactory results if performed correctly.

As an example, here is a simple simulation of a 15'' subwoofer in a closed 140L cabinet in VCAD and in comparison the real measured combined near field and far field measurement of the finished loudspeaker (both with baffle step correction).
Above 100Hz the 15'' subwoofer no longer behaves ideally, hence the deviation from the simulation. But below 100Hz the combined nearfield-farfield measurement agrees very well with the simulation (below 20Hz my measurement microphone is no longer calibrated).
As you can see, it works without PhD and 15m tower.
View attachment 154805



If it's that easy, let's just do it. Let's see how close a 801D4 clone will get us to the f3 of 15Hz at [email protected] in free field (anechoic). One never believes me here without all-embracing proof

First, we simulate the two subwoofers in half space (2pi) with 100L + 100L volume and a BR port tuning of 20Hz - as specified by @mt42. With a series resistor of 0.5 Ohm (representing the resistances of cable and crossover) the drivers in parallel configuration reach a minimum of 2.5 Ohm.

View attachment 154816

In half-space we reach the targeted [email protected].
f3 = 18Hz and a f6 = 16Hz. If we could stop here, the specification of the 801D4 (800D3) would be pretty well met. Unfortunately, if we don't want to use the 801D4 as wall in speakers, we'll have to move on.


Now we still have to take into account the transition into the full space (into the free field). So now comes the baffle step correction based on the speaker baffle size.
View attachment 154839
At first glance, this looks worse than it is. The blue curve shows the ideal expected frequency response of the driver without crossover in the loudspeaker cabinet in free field.
When the filter slope of the woofer is formed in the crossover, the capacitive part of the driver impedance forms a resonant circuit together with the series-connected coil, which leads to the formation of a hump in the frequency response around the center frequency of the resonant circuit.
People with a PhD in electrical engineering can explain this more precisely

I have drawn this roughly as an orange curve. The hump can also be more pronounced - this is only an estimate by me.
View attachment 154850
In yellow I have drawn the sound pressure at 15Hz. This is with ideal drivers and the TSP data of the Focal 25A4 subwoofer just 78dB at 2.83V.
If the rest of the speaker really plays with [email protected], then f12 = 15Hz and not f3 as stated in the specification - voila! The laws of physics still apply.
Our 801D4 copy would have an f6 of 25Hz at 90dB sensitivity. The value for f3 can only be estimated by a large margin (and depends strongly on the interaction of driver impedance and crossover coil) and should be somewhere between 35-45Hz at 90dB sensitivity.

Since almost nothing is standardized with regard to the manufacturer's specifications, B&W can elegantly pull out of the matter and say "Wait a minute, our specifications refer to a speaker that is standing in the room and not as usual on the basis of anechoic measurements".

With the help of a boundary surface, such as the floor, the speaker comes close to the values in the specification:
View attachment 154852

Really appreciate your in depth explanations.

 

[Deleted member 27948]

Dear ctrl, I've read your full review of 800D3.

Could you add more details on how you measured near-field LF response? How was the spk positioned? angled forward? laying down sideways on the floor? What was mechanical coupling? Where was the mic? at the floor? between woofers at X sm?

How did you make sure that the bottom port response and the woofers' response combine at the same ratio as for far-field?

Suppose that their corresponding distances to the mic were different... and the port response (the long decaying sine wave) at your particular setup was ~6...8dB below what it should have been at far-field (re woofers'). If you would rise it back, you could have ~70Hz bump down to the flat and the elbow at 30Hz would be 6...8dB up, to the same level as 100Hz.

BTW, Your room is not too bad except that the foam seems to be too dense (which you seem to know already) and reflective at high freqs.

 

[ctrl]

Dear ctrl, I've read your full review of 800D3.

I have not written a review of the 800D3. There my bad English might have led to a misconception.

Could you add more details on how you measured near-field LF response? How was the spk positioned? angled forward? laying down sideways on the floor? What was mechanical coupling? Where was the mic? at the floor? between woofers at X sm?

For this purpose, I always recommend the Arta Handbook by Heinrich Weber. Basically everything you need to know is explained there. For the examples, the Arta measurement software is (mostly) used, but the topics covered can be transferred to any other software.
The terms used are also precisely defined there (near field, far field, free field,...).
Example:

Your question is not easy to answer, because you ask for details about the "ground plane measurement" technique (allows anechoic measurements), but in my example (15'' subwoofer) I used a combination of near field (mic a centimeter from the driver) and far field measurement (a gated in room measurement).

How did you make sure that the bottom port response and the woofers' response combine at the same ratio as for far-field?

Also this "technique" is all in detail in the Arta Handbook.


How this is implemented using REW and VCAD has been shown by @napilopez in the thread "How to make quasi-anechoic speaker measurements/spinoramas with REW and VituixCAD".

With large loudspeakers where the BR ports and woofers are located far away and have large differences in delay at the listening position, it can be useful to take this phase shift into account when combining near-field and far-field measurements - e.g. 2-way loudspeaker or with high crossover frequency of the woofer, in these cases, the phase shift could have an effect when adding the higher frequency components..

 

[Deleted member 27948]

I have not written a review of the 800D3. There my bad English might have led to a misconception.


For this purpose, I always recommend the Arta Handbook by Heinrich Weber. Basically everything you need to know is explained there. For the examples, the Arta measurement software is (mostly) used, but the topics covered can be transferred to any other software.
The terms used are also precisely defined there (near field, far field, free field,...).
Example: View attachment 154936

Your question is not easy to answer, because you ask for details about the "ground plane measurement" technique (allows anechoic measurements), but in my example (15'' subwoofer) I used a combination of near field (mic a centimeter from the driver) and far field measurement (a gated in room measurement).


Also this "technique" is all in detail in the Arta Handbook.
View attachment 154944

How this is implemented using REW and VCAD has been shown by @napilopez in the thread "How to make quasi-anechoic speaker measurements/spinoramas with REW and VituixCAD".

With large loudspeakers where the BR ports and woofers are located far away and have large differences in delay at the listening position, it can be useful to take this phase shift into account when combining near-field and far-field measurements - e.g. 2-way loudspeaker or with high crossover frequency of the woofer, in these cases, the phase shift could have an effect when adding the higher frequency components..

Your English is good enough for me. I am not a native talker either, it's my third language. Ok...

did you ever know that we live in a conservative 3-D Euclidian space? if we lived in 1D space the sound would never become softer, will be able to listen to it at any distance at the same loudness. If we lived in 2D space, then a pulse emitted by a point source would decay accordingly to the law conservation energy, 3dB per each doubling of distance. As we live in the 3D space, we have to live with sound decaying by 6dB per each doubling of the sound distance. If your mic is 10sm away from the woofer, and 50sm away from the port, then this port IR you may not add the port appropriately because it's 5 times softer.

I write all processing scripts myself and borrow from MATLAB Toolboxes/ Community whenever possible. I do not use any specific commercial signal generator/analysis/presentation SW products. I'd not know what they are doing inside ...and I don't want to know.

Agree, your IR did not have its port portion "truncated", per se, but windowed down.

 

[Art Vandelay]

Just seen 801 D4 measurement in new HiFi News Mag

Very average FR, to put it lightly

These are made for people that want to show off and existing people upgrading that are happy with the house sound

Not made for well balanced / neutral playback

Interesting to see the step change in amplitude around 1kHz, which is possibly deliberate - and designed to compensate for the change in directivity / beamwidth at the same frequency. The basic response curve is quite similar but distortion is considerably lower at low and mid frequencies, so I'm not surprised that the reviewer was highly impressed with the (subjective) performance and overall improvements.

As always, the devil is in the detail, which isn't fully revealed nor necessarily highly accurate, so we'll need to wait for a few more reviews and lab tests to fill in some gaps and achieve a higher level of confidence.

I'll have the opportunity to hear the speakers soon too, in a well designed listening room and driven by high quality amplification, so that will answer my ultimate questions re neutrality, transparency and resolution etc..

 

[witwald]

Harking back to the B&W DM6 of 1975 vintage (see below), why is the measured response apparently so much smoother, falling within ±1dB limits between 70Hz and 16kHz, than those of the 801 D4 and 800 D3? Has there been a major change in design philosophy?

 

[witwald]

As a point of reference to the newer Bowers & Wilkins loudspeaker models, below is a distortion frequency response plot pertaining to the B&W DM6. The DM6 is now approximately 46 years old. Its distortion figures seem to be quite good when compared to the present series. And in the old days, B&W were quite happy to publish some measured data in their brochures to give customers some insights into what their designers had accomplished. Do they do that these days?

 

[witwald]

This is a post meant to communicate the leak of new D4 like, there is already a thread where people started fighting over the D3 series and older, one side saying it has some high frequency peaking, and the other side saying they have listened by them self and find them the best thing out there.

Is it that surprising that some people are finding them the best thing out there? It would be mildly surprising if that weren't the case, irrespective of the fact that B&W designs from 45 years ago were probably performing equally well, if not overall better. The applause and support happens with most new products these days, if the reviews are anything to go by.

 

[witwald]

800D3
Dimensions: HxWxD 1.2m x 0.41m x 0.61m - about 1m height for the woofers.
Between 100 and 70Hz the 800D3 has a slight hump, a bass boost, which is at 92dB sound pressure level, but sensitivity is about [email protected].
As mentioned before, f3=37Hz and f6=29Hz (a second source, audio.de 04 /2017, measured the same f6=29Hz). The BR tuning frequency is about 26Hz (see impedance plot).
View attachment 154695
Source: fidelity-online.de

Thanks for sharing the link to the measurements of the Bowers & Wilkins 800 D3. Therein, the test chamber is described as being "constructed as an anechoic half-space with an absolutely reverberant concrete floor and enables free field conditions from approximately 100 Hz upwards."

Can anyone provide an explanation as to how a test chamber with an "absolutely reverberant concrete floor" can enable "free field conditions from approximately 100 Hz upwards"? Reverberant and anechoic would seem to be diametrically opposed. Admittedly, the sound pressure response measurements look to be quite reasonable, but are they affected by measurement errors caused by the sound-reflective floor?

Source: fidelity-online.de

 

[Rufus T. Firefly]

Thanks for sharing the link to the measurements of the Bowers & Wilkins 800 D3. Therein, the test chamber is described as being "constructed as an anechoic half-space with an absolutely reverberant concrete floor and enables free field conditions from approximately 100 Hz upwards."

Can anyone provide an explanation as to how a test chamber with an "absolutely reverberant concrete floor" can enable "free field conditions from approximately 100 Hz upwards"? Reverberant and anechoic would seem to be diametrically opposed. Admittedly, the sound pressure response measurements look to be quite reasonable, but are they affected by measurement errors caused by the sound-reflective floor?
View attachment 155309
Source: fidelity-online.de

I want to ay that half anechoic would mean that only the walls are anechoic but their claim of that being free field would still be a contradiction.

I'm punching above my weight here so take that with a grain of salt.

 

[Art Vandelay]

Harking back to the B&W DM6 of 1975 vintage (see below), why is the measured response apparently so much smoother, falling within ±1dB limits between 70Hz and 15kHz, than those of the 801 D4 and 800 D3? Has there been a major change in design philosophy?

Yes, the new philosophy is focused more on cabinet and driver resonances and distortion. There's also a more minimalist approach to crossover design. The designers obviously concluded at some point that there were factors apart from (on-axis) frequency response that were important for ultimate subjective accuracy, and some 50 years later they now have the knowledge, the technology and the capability to better address the full suite of factors that limit loudspeakers from faithfully reproducing a live performance.

 

[Art Vandelay]

Thanks for sharing the link to the measurements of the Bowers & Wilkins 800 D3. Therein, the test chamber is described as being "constructed as an anechoic half-space with an absolutely reverberant concrete floor and enables free field conditions from approximately 100 Hz upwards."

Can anyone provide an explanation as to how a test chamber with an "absolutely reverberant concrete floor" can enable "free field conditions from approximately 100 Hz upwards"?

Removing the roof and ceiling would do it, but I suspect they're employing lots of sound absorbent material.

 

[thewas]

Can anyone provide an explanation as to how a test chamber with an "absolutely reverberant concrete floor" can enable "free field conditions from approximately 100 Hz upwards"? Reverberant and anechoic would seem to be diametrically opposed. Admittedly, the sound pressure response measurements look to be quite reasonable, but are they affected by measurement errors caused by the sound-reflective floor?

Because they use the ground plane measurement method where the floor reflection doesn't play a role:

Source of the photo: https://www.fidelity-online.de/kef-reference-1-messungen/

The size of the room, distance of mic and gating limits the measurements to the written 100 Hz.