• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Gundry Dip in loudspeakers

DSJR

Major Contributor
Joined
Jan 27, 2020
Messages
3,312
Likes
4,425
Location
Suffolk Coastal, UK
Apologies if this has been discussed or posted before, but I stumbled on this just now and being UK based and occasional defender of the BBC style of speaker design (cough) and seeing lower kHz dips in so many modern speakers reviewed here, I thought some of you might find the linked thread interesting -

https://hydrogenaud.io/index.php?topic=75195.0
 

zeppzeppzepp

Member
Joined
Jul 10, 2020
Messages
68
Likes
31
Dip in frequency of present range is surely better than boost for most of recordings.
More likely a historic reason than technical issue.
 
OP
DSJR

DSJR

Major Contributor
Joined
Jan 27, 2020
Messages
3,312
Likes
4,425
Location
Suffolk Coastal, UK
What I got was the fact the BBC engineers (and other broadcasting editing people) worked close to the speakers and reducing the 'glare frequencies' just a little helped them do their work better. Mind you, the LS5/8 and 5/9 had their presence dips lower in frequency - from around 500 to 2kHz, thereby pushing far more of the soundfield back behind the plane of the speakers.

I know of two scandinavian makers which in some of their models, boost these frequencies - to make them appear more 'detailed?' I find the tones cold as ice and fatiguing if not *very* carefully matched.
 

youngho

Senior Member
Joined
Apr 21, 2019
Messages
486
Likes
799
Also important to point out that the dip often occurs in the off-axis response, like here: https://www.stereophile.com/content/stirling-broadcast-bbc-ls36-loudspeaker-measurements, somewhat here https://www.stereophile.com/content/harbeth-m401-loudspeaker-measurements. I've wondered whether this off-axis dip was analogous to spandrels in an evolutionary sense, simply a by-product of the directivity of the drivers, which were chosen for the on-axis frequency ranges, not specifically in order to produce this off-axis dip.
 

ctrl

Major Contributor
Forum Donor
Joined
Jan 24, 2020
Messages
1,616
Likes
6,086
Location
.de, DE, DEU
Apologies if this has been discussed or posted before,...
This is discussed in every audio forum around the globe. But most often it is called BBC dip. A search here in the forum also immediately provides a few hits ;)

There are arguments about the name, the origin, ... whatever. For "modern"/recent loudspeakers this does not really matter.

What is relevant is that there are situations where a dip in certain frequency ranges makes perfect sense and the approach to establish a frequency response (FR) and listening window (LW) as flat as possible does not make sense.

This is the case when the loudspeaker shows a strong, especially in the horizontal, widening of the radiation in certain frequency ranges and loudspeaker designers who do not design a crossover based on axial frequency response alone (almost all should be), consider the resulting change in sound power (and ER and PIR). This then leads to the dip in "modern" speakers.

This is very often the case due to edge diffraction. This strongly depends on the width of the loudspeaker and the crossover frequency. Edge diffraction results in a dip in the axial frequency response (more precisely in on-axis dips and humps, but mostly only one dip is really relevant) and a more or less strong widening in the radiation outside the axial frequency response.
Since most speakers have a width of 0.15-0.4m, the effect occurs in the range 1700 - 4500 Hz.

This phenomenon is particularly severe when a "normal" tweeter is placed in the center of the baffle and a low crossover frequency is selected.
Here is an example with [email protected].
1620397303307.png
A simple crossover without additional parametric EQ or notch filter gives us a wide dip in the frequency range 2.3-4kHz.
1620397284848.png

What you should not do in any case is to get the frequency response as smooth as possible, because then this happens (FR normalized to the axis frequency response):
1620397328182.png
The sound power in the 2-4 kHz range is much too dominant.

In fact, it is very likely that you will have to deepen the dip in the frequency response to avoid a too harsh sound at higher sound pressure levels, since the vertical FRs also show a widening in the radiation in the same frequency range:
1620402387369.png

Therefore, it is not always a bad sign when an attempt is made to compensate for design weaknesses via the axis frequency response (and thus usually also the LW) - and a linear on-axis FR/LW is dispensed with for this purpose.
 

zeppzeppzepp

Member
Joined
Jul 10, 2020
Messages
68
Likes
31
Yes, quite understandable.
And for most of the existing recordings, we don't need to fill up the dip to get better sound, it just won't be better.
 
OP
DSJR

DSJR

Major Contributor
Joined
Jan 27, 2020
Messages
3,312
Likes
4,425
Location
Suffolk Coastal, UK
For my sins, I belong to a members only Spendor group and one of the posters there is Derek Hughes, son of Spencer (the Spen in Spendor) and designer of many of their post 1983 models after his father's passing up until the buy out in the early noughties.

Derek has posted an interesting explanation as another reason for the crossover-ish dip but hopefully it's also related to things the Klippel shows easily enough. Back in the 70's, no such complex computerised measuring systems were available, but hopefully his explanation might help with another insight. Of course, my own small listening room is very heavily damped and furnished and with my ears as they now are, the last thing I need is a suckout in the lower kHz region ;)

Maybe the highly characterful LS5/9 (used often in editing suites) was altered from flat for different reasons as their dip is lower - in the upper midrange...

Apparently also, Mr Gundry would be horrified at his name being put to this response characteristic!

I quote -

"Although Spen never gave it the 'BBC dip' name, he did explain the origin of the idea to me once.
All Research designed speakers were balanced using a combination of Anechoic testing and listening in the well treated Kingswood listening room which had low RT across the band and did not suffer for first reflections off the side walls.
A question was raised one day as to what the speakers (probably the LS5/5) would sound like in the more reverberant environment of the lab, and it was found they were significantly more bright/hard in sound, which could be improved by reducing the level of the tweeter just above the crossover where the off axis response widens out.
This dip was never applied to the actual BBC designs as far as I know, because they were for use in acoustically treated control rooms.
It can certainly be beneficial to the overall sound if this area is reduced in the usually more 'live' environment of a domestic living room, but the effect will vary for every different situation.
It can be seen in a lively environment if one has access to test gear with a time gating window. The differences in the lower tweeter area are easy to see when comparing the gated/non gated responses."
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,281
Location
Oxford, England
This is taken from a piece that BBC engineer H. D. Harwood (the man who started Harbeth) wrote for Wireless World in 1976 titled "Some Factors In Loudspeaker Design" (I think that the whole piece can be downloaded from Linkwitz's website)

gh9BHbM.jpg
 

Blockader

Active Member
Joined
Nov 17, 2021
Messages
282
Likes
675
Location
Denmark
This is discussed in every audio forum around the globe. But most often it is called BBC dip. A search here in the forum also immediately provides a few hits ;)

There are arguments about the name, the origin, ... whatever. For "modern"/recent loudspeakers this does not really matter.

What is relevant is that there are situations where a dip in certain frequency ranges makes perfect sense and the approach to establish a frequency response (FR) and listening window (LW) as flat as possible does not make sense.

This is the case when the loudspeaker shows a strong, especially in the horizontal, widening of the radiation in certain frequency ranges and loudspeaker designers who do not design a crossover based on axial frequency response alone (almost all should be), consider the resulting change in sound power (and ER and PIR). This then leads to the dip in "modern" speakers.

This is very often the case due to edge diffraction. This strongly depends on the width of the loudspeaker and the crossover frequency. Edge diffraction results in a dip in the axial frequency response (more precisely in on-axis dips and humps, but mostly only one dip is really relevant) and a more or less strong widening in the radiation outside the axial frequency response.
Since most speakers have a width of 0.15-0.4m, the effect occurs in the range 1700 - 4500 Hz.

This phenomenon is particularly severe when a "normal" tweeter is placed in the center of the baffle and a low crossover frequency is selected.
Here is an example with [email protected].
View attachment 128508
A simple crossover without additional parametric EQ or notch filter gives us a wide dip in the frequency range 2.3-4kHz.
View attachment 128507

What you should not do in any case is to get the frequency response as smooth as possible, because then this happens (FR normalized to the axis frequency response):
View attachment 128509
The sound power in the 2-4 kHz range is much too dominant.

In fact, it is very likely that you will have to deepen the dip in the frequency response to avoid a too harsh sound at higher sound pressure levels, since the vertical FRs also show a widening in the radiation in the same frequency range:
View attachment 128514

Therefore, it is not always a bad sign when an attempt is made to compensate for design weaknesses via the axis frequency response (and thus usually also the LW) - and a linear on-axis FR/LW is dispensed with for this purpose.
Underrated post. Thanks a lot for such clear explanation.
 
Top Bottom