Well, resonances tend not to shift in frequency in the time domain. The ridges in your measurement plot do so. So unless something screwed up the measurement my opinion was more of an observation 😉
Consider constructing your new cab in a triangular (only in 2 dimensions, keep the height limited to the net driver size) asymmetrical shape. That is easy and pretty effective.
I agree with the reasoning, but, when reading about the burst decay waterfall plot implemented in REW, John Mulcahy makes note of those upward shifting resonances are indeed an artifact of processing. He states "Note, however, that an artefact of the period axis is to skew the tail of the decay slightly towards higher frequencies rather than maintaining the symmetry about the resonance's centre frequency that would be seen in a time-based plot." It sure would be great if John could do some more work in this area and perhaps overcome this? I wonder if it is because its processing the spectrum in successive slices in time, but the sine sweep is upward sweeping in time?
A cumulative spectral decay waterfall plot doesn't have this artifact but the spectral decay is also frequency dependent and higher frequencies have shorter and shorter decays, so an artifact of a CSD is that all high frequencies will show no resonances. The burst decay plot was actually created to solve that. So we just have to use the tools we have and understand the assumptions/limitations of each.
Yeah, I rather like the idea of that. So you're describing a triangular (non-equilateral) shape from the plan view (top view), and driver located off center on one face. I sure would like to find an enclosure simulator that would use CFD to find the enclosure resonances of irregular shaped geometries.
Dunno about that. I respect Mulcahy very much, but:
This is an ARTA burst decay of a metal cone speaker. The resonance ridge stays at 9kHz neatly. In your BD plots there are quite big frequency shifts present.
[Edit]neither the plots in the REW manual show major shifts, so I stick to my observation.
Correct. A triangle could be solved by using the wave equation. But my gut feeling says that minimizing the enclosure size will solve your problems. The driver doesn’t have to be that far from center either. 1/3 could work fine.
Interesting. I just ran a sweep with ARTA and am not seeing the upward sweeping tails here. So, this must be an issue with REW.
Non parallel walls will do absolutely nothing to mitigate standing waves.
https://www.acculution.com/single-post/037-misconceptions-in-acoustics
https://www.acculution.com/single-post/037-misconceptions-in-acoustics
Last edited:
They sure get mitigated. They never will disappear, that is true. Like I wrote, solve the wave equation. Ore use BEM. But non-parallel walls will spread modes in location, reduce levels and spread them in frequency. So in general, yes, put that one non-parallel wall in your enclosure. But: the final goal is to get them out of the way of the cone.
The link is both interesting and may lead to confused readers. Statements like “pressure on one side can be in phase with pressure at the other side of the cone” don’t help, while they are certainly true at certain moments in time.
You will create more complex modal patterns for mid and high frequencies, through scattering, that much is true.
I should have read the thread starter’s requirements more carefully, for lower frequencies standing waves persist in irregular shapes and are not a stand alone solution.
I should have read the thread starter’s requirements more carefully, for lower frequencies standing waves persist in irregular shapes and are not a stand alone solution.
At what distance from baffle measured?
To me it looks close, then you would not see baffle diffraction or if you lower the floor to -40 or so ypu might see hints.
JanRSmit, correct, it was measured near-field, 2.5" from the cone. But then again the sweep with REW was taken from the same distance.
You'll get the same number of modes in an irregularly shaped box of similar volume. It's one thing to spread them, but you'll still get delayed versions of the original coming back.
Did someone say damping material wasn't a significant helper?
Did someone say damping material wasn't a significant helper?
EDIT: Just wanted to correct myself. The tails of increasing resolution do appear when viewing a Burst Decay preferring time resolution. When performing a Burst Decay preferring frequency resolution there is no such artifact. I believe that is what John Mulcahy was referring to.
The rew bd goes much deeper some 50dB. The first arta is just 30dB, your last one to 40dB. Apple with pears
No, but keep in mind I learned this in late '50s - early '60s drafting all manner of architectural structures + IIRC it's in one of my much newer audio books.
Update: I have created a new enclosure. I designed using a triangular shape as viewed from above and driver height, 4.25", producing a 3.0L volume.
I ran the same burst decay analysis on it. Here are the before and afters. I also find I prefer sonograms to the 3D waterfall plots. Yeah, the waterfalls look "cool" but ultimately I find the sonogram easier to visually process.
Before:
After:
I found the biggest improvement in the 1K to 3khz region, where before it would take 9-15 periods for the resonance to reach 30dB down, and after only 6 periods to reach the same period. I am not really concerned with the 5K+ response as this midrange will be crossed over around 2.5khz.
What I don't know is how to attribute the changes in the resonances, whether its the new enclosure with non-parallel walls or its due to the driver being mounted in a different sized baffle (quite a bit smaller, 5.75" x 11", the original baffle is part of a tower which measure 41" x 14"). I am unsure how to measure baffle diffraction separately from enclosure resonances.
Note that the impulse response gating window, which is necessary to obtain quasi-anechoic results, also results in the large region of quiet in the lower left quadrant of the sonogram. There is an inherent tradeoff between low frequency fidelity and measuring the speaker vs measuring the room, the larger the window the lower frequency visible but it measures more of the room. Both enclosures were stuffed with a light amount of polyfill.
Ultimately however, my original hypothesis was disproven, that there were standing waves in the enclosure which produced the uneven frequency response, (peak at 2.5k, the dip at 4k and the peak at 6k and to a lesser extent the dip at 1k). I observe roughly the same characteristic frequency response out of both enclosures, and it isn't likely that they have the same resonances present, therefore the frequency response cannot be attributed to the resonances.
I ran the same burst decay analysis on it. Here are the before and afters. I also find I prefer sonograms to the 3D waterfall plots. Yeah, the waterfalls look "cool" but ultimately I find the sonogram easier to visually process.
Before:
After:
I found the biggest improvement in the 1K to 3khz region, where before it would take 9-15 periods for the resonance to reach 30dB down, and after only 6 periods to reach the same period. I am not really concerned with the 5K+ response as this midrange will be crossed over around 2.5khz.
What I don't know is how to attribute the changes in the resonances, whether its the new enclosure with non-parallel walls or its due to the driver being mounted in a different sized baffle (quite a bit smaller, 5.75" x 11", the original baffle is part of a tower which measure 41" x 14"). I am unsure how to measure baffle diffraction separately from enclosure resonances.
Note that the impulse response gating window, which is necessary to obtain quasi-anechoic results, also results in the large region of quiet in the lower left quadrant of the sonogram. There is an inherent tradeoff between low frequency fidelity and measuring the speaker vs measuring the room, the larger the window the lower frequency visible but it measures more of the room. Both enclosures were stuffed with a light amount of polyfill.
Ultimately however, my original hypothesis was disproven, that there were standing waves in the enclosure which produced the uneven frequency response, (peak at 2.5k, the dip at 4k and the peak at 6k and to a lesser extent the dip at 1k). I observe roughly the same characteristic frequency response out of both enclosures, and it isn't likely that they have the same resonances present, therefore the frequency response cannot be attributed to the resonances.
For enclosure or driver resonances (pistonic cone movement) measure just a few mm (0,1 times the radius of the cone) from the cone. Adjust level in order not to overload the microphone and apply long time window > 50ms.
For diffraction measure at least 3 times baffle width from driver and repeat at 1m or, outside, at 2m. Use normal windowing, excluding first boundary reflection. 4 to 8ms are typical. Shorter gates require more critical settings of the window shaping, read the manual for that I guess. But try to keep things constant. One of the tricks of the trade is to keep constant good measurement conditions and it is good practice to write down settings.
The stuff between 2,5 and 6k was probably diffraction, some pointed that out already. Do you mind to post pictures of your enclosure?
For diffraction measure at least 3 times baffle width from driver and repeat at 1m or, outside, at 2m. Use normal windowing, excluding first boundary reflection. 4 to 8ms are typical. Shorter gates require more critical settings of the window shaping, read the manual for that I guess. But try to keep things constant. One of the tricks of the trade is to keep constant good measurement conditions and it is good practice to write down settings.
The stuff between 2,5 and 6k was probably diffraction, some pointed that out already. Do you mind to post pictures of your enclosure?