Buchardt comments on the design and data presented
To comment on the results showed by Amir, it is for a start important to know that every speaker design comes with a lot of compromises. The true art of designing a loudspeaker is to engineer ways so that the amount of compromises becomes as few as possible to meet the desired target. Therefore, let us start by describing the target of the Buchardt S400:
Design targets are:
⦁ Compact design with extended LF response
⦁ Avoid port designs for compression and noise reasons
⦁ Produce state-of-the-art stereo image
⦁ Produce as large a sweet-spot as possible
⦁ Good power handling (avoid bottom out at too low levels)
Given these design-targets, the choice of PR design is obvious, as it allows quite low tuning and few artefacts in a compact design.
State-of-the-art stereo image requires especially horizontal dispersion to be well controlled. Objectively seen, this is done by optimizing both sound power and on-axis/listening window responses at the same time. The only way to optimize all at the same time is to assure system DI is smooth and even. In practice this means at x-over frequency, we must match DI from woofer with DI from tweeter. A waveguide design has been made, that does the job of increasing tweeter DI at x-over frequency to match with that from the midwoofer.
The large sweet spot is assured by designing dispersion to be as even as possible over a wide frequency range.
The power-handling is assured by making sure to drivers can handle high power with no risk of bottoming out. If bottoming happens, suspension is designed in a way that make sure soft parts are catching the driver to avoid any damage to the mechanical parts.
A Klippel NFS system is used for evaluating each step of the design-process, to assure targets are held.
Correction to the use of Klippel fitting error as a marker for determining correct acoustic axis
View attachment 61004
In our opinion it is NOT the right use of the NFS fitting-system, to judge where the correct acoustic axis for the product is. According to NFS measurement manual – in order to get the best possible fitting for the system, you should choose the driver playing the highest frequencies as your starting point for the spherical harmonics fitting. This will in most cases be the tweeter. So - from our perspective this result will always be the case – and you will always get bad result when a different point is chosen. This just says something about how the Klippel fitting function is implemented, nothing about the actual speaker. To judge the acoustical listening axis, the vertical contour plot will give you the answer, if reference axis is chosen properly in Klippel hardware setup.
We know the S400 has a slight downwards beaming, which is why we recommend quite tall stands for the speakers, and why the front-baffle is slightly tilted backwards. In our presented measurements, we have compensated for this – and shows results on the optimum acoustic axis.
518Hz peak-and-dip
The rootcause of this is the PR on rearside interfering with the front woofer at this frequency because of acoustic leakage from PR cone on back. This leakage frequency falling together with the distance between front- and rear drivers causes this peak-dip.
BUT – we cannot reproduce this to the extent shown in above CEA2034 plot presented by Amir. Please see our ON-axis response below at 2m distance. We have talked to Klippel directly about this – and they cannot rule out that this may in fact be caused by their calculation algorithm, that the effect is much worse in the plot presented by Amir. We have double checked several speakers about this – and did not find the problem to the extend presented.
View attachment 61005
Internal damping is not causing this – our internal damping has been accurately adjusted to optimize the acoustics of the speaker. No more – and no less.
We see mentioned as a root-cause that there is no bracing in this enclosure. This is an intended act for two reasons:
1 The internal bracing would take up internal volume (In a very compact speaker it matters), which we could use to extend our LF instead. Because we cannot close the bracing near rear wall, we were forced to use more material than in a normal bracing.
2 The benefit of the internal bracing in terms of damping acoustic vibrations from the cabinet is below what can be measured. We checked this initially with an accelerometer on the cabinet sides, to be sure we made the right compromise – and the effect of the bracing was non-existing.
2.4kHz DI peak
This peak is caused by the vertical loping at the x-over frequency (SP dip). This will always happen with cone/dome speakers when they are separated with a distance between them. Our distance in this case is however slightly larger than normal due to the large waveguide diameter. This is a compromise done well considered because we believe the cost is not significant compared to the reward. Benefit is our waveguide can help us get the smoothest possible horizontal directivity (leading to superb stereo image) and large sweet spot. It also helps us a lot vertically, except for the loping at x-over frequency. Please let us show measurements done at 2m distance for vertical contour plot:
View attachment 61006
We find Amir’s contour plots performed at 66cm in nearfield not very useful.
View attachment 61007
Crossover frequency change.
Further, it has been noticed, that the SP dip indicates acoustic x-over frequency is slightly higher than the initial posted 2kHz. We did a minor change to the x-over after our first batch, because we found the 19mm dome in some extreme cases could be thermally damaged. After the change we do not recall a single tweeter has been replaced in field. The change was done silently because we evaluated the performance change very well and found it insignificant. We did not see a reason to disturb market with this update, as it was way below what could justify a MKII update. However, we did see some comments on why we do not use the waveguide fully down to its optimum frequency range. This is the reason. BUT – we also need to comment on the contour plots shown by Amir in the review, as they do not justify the acoustic design. We can see that these plots are evaluated at 66cm distance. This means practically nearfield evaluation. If the evaluation for horizontal contour plot is done at 2m distance, the performance is still very smooth.
View attachment 61008
View attachment 61009