• 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!

Build of Bagby Mandolin speaker

MrPeabody

Addicted to Fun and Learning
Joined
Dec 19, 2020
Messages
657
Likes
942
Location
USA
Two suggestions have been offered for what is causing the dip at 3 kHz. Both are plausible.

The first suggestion that was made is that it is due to the measurement point being at the wrong height. The wavelength at the frequency of the dip is 4.5”. The graph of frequency response and phase on the Meniscus site indicates that the two drivers are offset in acoustic phase by approximately 360 degrees over a wide frequency range to either side of the crossover frequency. In essence they are acoustically coherent, so if the dip is associated with phase cancellation between the two drivers, one-half of the 4.5” wavelength would need to equal the difference in the distance from the mic to one of the drivers vs. the distance to the other driver. As the distance from the speaker to the microphone increases, the difference between these two distances converges to a specific value, however when the microphone distance is not great, the difference between the two distances depends significantly on the microphone distance. If the measurements were taken at tweeter level rather than equidistant from the two drivers, it is easily possible that the difference between the two distances would have been 2.25”, in which case the two wavefronts would have been out of phase for wavelength equal to 4.5” and frequency equal to 3 kHz. However, if this were the case, there would likely have been a stronger dip than what is seen, depending on the relative strength of the two acoustic waves.

As for baffle edge diffraction, I want to point out first that this can’t be ruled out simply on the basis of the response not being changed significantly by the application of 7/8” edge rounding. This would be correct reasoning if and only if it were known with certainty that this amount of edge rounding would be highly effective at suppressing diffraction at the wavelength in question. This amount of rounding would likely be somewhat effective at 4.5” wavelength, but not sufficiently effective to suppress diffraction to the extent needed to eliminate the dip.

If the dip at 3 kHz is due primarily to edge diffraction, there should also be a pronounced peak at 1.5 kHz, and a mild peak at 4.5 kHz, and a very mild dip at 6 kHz. There is evidence of a peak at 4.5 kHz and evidence of a dip at 6 kHz. But no evidence of a peak at 1.5 kHz. I estimated the width of your baffle to be about 8.5”, which implies that the peak should be found at 1.6 kHz and the dip at 3.2 kHz. The lack of a peak at 1.6 kHz is explained by a couple of things, both of which have to do with the woofer. And it is the woofer that would primarily be affected at this frequency. There will also be a mild ripple associated with the top edge of the baffle, and the distance from the center of the woofer to the top edge looks to be almost exactly the same as the width of the baffle. This means that the first dip in the ripple associated with the top edge will coincide with the frequency of the peak for the stronger side-edge ripple, at 1.6 kHz. But, this dip wouldn’t be strong enough to completely nullify the peak at 1.6 kHz all by itself, because the top edge isn’t very wide and the bottom edge isn’t at the same distance as the top edge. The other contributing factor, that explains why there is no peak at 1.6 kHz, is that there is a mild but sharp dip in the woofer’s natural response right at this frequency. All of this considered, there is nothing that would contradict the hypothesis that the dip at 3 kHz is mainly a consequence of baffle edge diffraction, which is likewise a strong contributing factor in the peak at 4.5 kHz and the mild but wide dip at 6 kHz.

But this raises the question of why the response looks different from the response shown on the Meniscus site. It might be that in the speaker they (or Bagby) measured, the tweeter was mounted to one side of the center of the baffle. I’m not familiar with this speaker, however I did a quick web search and saw a bunch of images where the tweeter is mounted horizontally off center. If it were mounted 1/3 of the way from one edge, this should mostly eliminate the dip at 3 kHz.

The off-axis measurements will reveal whether baffle edge diffraction is the correct explanation. If this is the correct explanation, the horizontally off-axis response will exhibit peaks at the same frequencies where the on-axis response exhibits dips, and vice versa. The horizontally off-axis response should exhibit a peak at 3 kHz and a dip at 4.5 kHz. At 3 kHz, the on-axis and off-axis response curves with be squeezed closer together, and at 4.5 kHz, they will be spread slightly further apart, compared to the overall separation of the on-axis and off-axis response curves in the vicinity of these frequencies. This will be the tell-tale sign that will reveal whether baffle edge diffraction is or isn’t the principal cause of the response deviations at 3 kHz, 4.5 kHz, and 6 kHz.

In physical science, the gold standard for any hypothesis is with making a prediction that hasn’t previously been predicted through any alternate hypothesis, and for this prediction to be confirmed after it has been made. If we do not see the squeezing and spreading of the on-axis and off-axis response curves at these frequencies as predicted, then the hypothesis that these response anomalies are caused by baffle edge diffraction will be junk, no matter if there is anything else that anyone else thinks is evidence of baffle edge diffraction. If baffle edge diffraction is the cause, then the squeezing and spreading of the on-axis and off-axis response curves will be evident. If this effect is not evident, then baffle edge diffraction is not the correct hypothesis.
 

forlau

Member
Joined
May 11, 2021
Messages
18
Likes
9
While the design did not call for it, to reduce any speculation (and still had my 7/8" roundover bit in the router), here is the difference between the rounded edge and not...

View attachment 129862
Red trace is before (not rounded) and green is after (7/8" round). Clearly at this point, diffraction is NOT a significant contributor to the 3k Hz dip as you can see the response barely changed.

A round over with 7/8 inch (=0.022225 m) results in a frequency change in: f=c/lambda = 344/0.022225 = 15478Hz. To come close to 3kHz, you should go for a 4.5inch (0.1146 m) round over :)
 

tonvo

New Member
Joined
Jan 5, 2022
Messages
4
Likes
0
Ok, for those who want to build a simple cabinet, this one follows my same basic design. Just for fun, here are a couple of pics. One with the brace...

View attachment 84335

and one with the bottom...View attachment 84336


I did not bother to draw layout as am not using full panels to build this one, but here is cut list (all .75“ thick except 1 inch thick front baffle):

back: 7 x 13.5
sides: 10.5 x 13.5
front: 8.5 x 15
brace: 7 x 9.75
top/bottom: 8.5 x 10.5
I don't see the brace on the original drawings. What are the dimensions of the 4 holes?
 
OP
Rick Sykora

Rick Sykora

Major Contributor
Forum Donor
Joined
Jan 14, 2020
Messages
3,511
Likes
7,011
Location
Stow, Ohio USA
I don't see the brace on the original drawings. What are the dimensions of the 4 holes?

Nor did I, but the holes should be large enough to avoid restricting airflow...

In my case, pretty sure I went with 2.5" round holes as I wanted enough brace to connect across the middle of the sides.
 
Top Bottom