I was wondering what your subjective opinion was.Here is the driver response and the contribution from the box, you'll have to decide if this is a problem or not:
I'm having a little trouble getting my head around the relative volumes of your measurements. Particularly, the bare driver reading is measured how- the driver is just mounted on an open baffle in free space and the mic is directly in front of the driver? Of course you would have to do this before mounting the driver on the boxes. Was the smaller silver-ish boxy cover for the driver installed on the back of this baffle for that testing? The reason why I'm so puzzled by this is that the relative volume of your measurements of the various box strategies is only 10 dB below the bare driver in many of the measurements. At these relative volumes I would think such a speaker would have significant and audible radiation out the sides of the cabinet that would appear in polar plots of the speaker's output. I would have expected the peaks of the 3/4" mdf with a brace to be at least 30 dB below the signal in order for cabinet resonances to not be obvious and annoying, and yet 3/4" mdf cabinets with simple bracing are quite common in the 0.4 to 1.0 cu ft speaker size range.
You are correct. For "bare driver" response it was mounted outside the box, actually in a little box sitting on top of the test box. Then it was mounted with the little rear enclosure to the test box so it was only radiating into the box. The relative levels are correct, within +/- 2 dB or so. Yes the sound is quite audible and probably would show up in a polar. It is pretty narrowband though. Even with the braces the sound radiated is quite high. You can test this yourself pretty easily.
ok, but then looking at the polar below from measurement of an Elac DBR-62 on a Klippel, and assuming the Elac is a reasonably well damped design with peaks ~20 dB below the signal, the measurements you've made show radiation from the side panels should appear in the 600-800 Hz range at around ~40 dB at 180 degrees off axis. In the polar it actually shows signal at 50 dB or greater in this range, so there could be sound coming from the cabinet but it is getting "swamped" a little by the main signal, or not. At that frequency I am not sure if that wavelength is short enough to "wrap around" the cabinet from the front of the driver - I need to check the wave length I suppose. But more to the point, these measurements would seem to suggest that except for the most exotically built speakers in the 20k+ range that use virtually every tool- special cabinet materials, CLD techniques, driver isolation, and maximum stuffing quality, most speakers produce clearly audible resonances from their cabinets when driven to moderately high volumes, and charts like John Atkinson's old "tape on accelerometer" measurements massively under-suggest the amount of resonance most speakers produce? I would expect designers to be screaming from the trees about how important this is, but also I would expect analyses of measurements by the likes of ASR to include better quantification of this behavior since I would expect it to significantly color the sound even in front of the speaker in many environments. What am I missing because I am seriously mystified by this (and definitely not questioning your work!)?
ASR notes a cabinet resonance right in the 500-1000hz range that is typical for a bookshelf-sized speaker in that ELAC.
That seems to support my testing that these resonances can be significant, and show up with the right kind of measurement. As for the comment that designers would be screaming from the treetops, well, they do in the more expensive models. They probably keep quiet on the less expensive models because the solution would add considerable cost to resolve. I'm not sure if that answered your question...
That seems to support my testing that these resonances can be significant, and show up with the right kind of measurement. As for the comment that designers would be screaming from the treetops, well, they do in the more expensive models. They probably keep quiet on the less expensive models because the solution would add considerable cost to resolve. I'm not sure if that answered your question...
It’s possible that it’s not be the cabinet,
that kind of narrow band resonance, at that frequency, could be caused by interference between the driver and resonator (port or passive radiator) that shares the same volume.
Unfortunately, it’s usually beyond the scope of ASR or EAC to troubleshoot/solve these kinds of problems.
Further investigations are needed.
that kind of narrow band resonance, at that frequency, could be caused by interference between the driver and resonator (port or passive radiator) that shares the same volume.
Unfortunately, it’s usually beyond the scope of ASR or EAC to troubleshoot/solve these kinds of problems.
Further investigations are needed.
sidenote, play multitone signal with a speaker, it seems to trigger box panels quite nicely. Now it's easy to scope with touch which panel(s) resonate and at which location. Then, use the woofer as mic and tap those spots to see frequency spectrum. This is easiest test to get worst offender(s) pop out without extra tools other than connecting the driver to mic input.
Beyond that, it's easy to imagine how bad panel resonances could be, just imagine all panels as transducers. Huge Sd, beaming, edge diffraction, all sortsa nasty stuff just like with real transducers, but even worse. If someone is building their dream system it is very important to make any enclosure/structure problem free.
Beyond that, it's easy to imagine how bad panel resonances could be, just imagine all panels as transducers. Huge Sd, beaming, edge diffraction, all sortsa nasty stuff just like with real transducers, but even worse. If someone is building their dream system it is very important to make any enclosure/structure problem free.
The impedence shows two cancellation issues either side of a small resonance which makes it stand out more
Jeez, such huge blip in impedance I've only seen with box mode or port problems. As it is single blip it's one resonance at particular wavelength. Panels don't seem to make so much into impedance plot, or do they?
Resonance seems to be centered around 750Hz, which is half wavelength of 1500Hz which is about 23cm long. Port seems to be at worst possible location although don't know where the inlet is. If it was worst possible port inlet location I'd expect port issues show more than one blip starting from height related wavelengths, longer than 23cm. Box outer dimension depth is 27.48cm according to product data, so the problemo is likely reflection from back of the box, between driver and back panel? Perhaps port has something to do with it as it looks quite bad, either leasks the box mode or is pipe mode in the port or something. Impedance measurement with blocked port could help.
Resonance seems to be centered around 750Hz, which is half wavelength of 1500Hz which is about 23cm long. Port seems to be at worst possible location although don't know where the inlet is. If it was worst possible port inlet location I'd expect port issues show more than one blip starting from height related wavelengths, longer than 23cm. Box outer dimension depth is 27.48cm according to product data, so the problemo is likely reflection from back of the box, between driver and back panel? Perhaps port has something to do with it as it looks quite bad, either leasks the box mode or is pipe mode in the port or something. Impedance measurement with blocked port could help.
Last edited:
I've always thought they wouldn't show up in impedance, but others have claimed they can and I'm pretty sure I've seen some posts here showing that. The horizontal response of the ELAC shows exactly the same behavior as a prototype I had Klippeled that was a sealed box with no bracing. In the final box this resonance was eliminated due to my CLD bracing and Resonix tiles. ELAC responses, followed by my prototype:
Last edited:
Do you have Power or DI plot for the prototype box? Elac DI seems quite stratight as in your final box, so radiation to various directions ought to be similar around the 500-1000Hz region, which could be just EQ:d I think. Your prorotype data shows measurements to various angles bundle up, which ought to make power response peak there, DI dip, which does seem to differ from Elac data. Fun stuff trying to interpret data 