I'm looking for more practical info (measurements of whole cabinets) about this topic.
I've seen 3 good articles or webpages:
BBC RD 1977/3 Factors in the design of loudspeaker cabinets
Ludwig's page:
Loudspeaker construction
Csaba's page:
Measuring the sound radiation of speaker cabinet walls
...and that's it. There is lots more academic info, and some measurements of single panels (which reach ~similar conclusions to the above), but I don't see other measurements of complete cabinets. Anyone got any more?
I realise these threads can get bogged down in subjective blah, so I'd prefer any commentary to be related to a linked set of measurements (that is: no anecdotes about sighted / subjective tests).
I really like Ludwig's very simple approach (pictured).
It has occurred to me that this same method could be used to generate comparative data for complete speakers boxes (as long as they have a flat face) -
1) put a small (computer) speaker on the floor, facing sideways (relative to the mic)
2 ) play impulses and pink noise through the small speaker; record the in-room result
3) lay the larger speaker box face down over the small speaker
4) repeat 2 (same volume and mic position)
It'd be nice to see a few tweaks measured this way - braced vs unbraced, damped vs undamped, etc, to get a less subjective / waffly idea about what these tweaks actually do.
[EDIT] some extra info:
LS50 white paper (KEF document, some helpful measurements)
patent US7270215 (CLD in bracing)
Damping factor values:
damping factor values - audio qualia
I've seen 3 good articles or webpages:
BBC RD 1977/3 Factors in the design of loudspeaker cabinets
Ludwig's page:
Loudspeaker construction
Csaba's page:
Measuring the sound radiation of speaker cabinet walls
...and that's it. There is lots more academic info, and some measurements of single panels (which reach ~similar conclusions to the above), but I don't see other measurements of complete cabinets. Anyone got any more?
I realise these threads can get bogged down in subjective blah, so I'd prefer any commentary to be related to a linked set of measurements (that is: no anecdotes about sighted / subjective tests).
I really like Ludwig's very simple approach (pictured).
It has occurred to me that this same method could be used to generate comparative data for complete speakers boxes (as long as they have a flat face) -
1) put a small (computer) speaker on the floor, facing sideways (relative to the mic)
2 ) play impulses and pink noise through the small speaker; record the in-room result
3) lay the larger speaker box face down over the small speaker
4) repeat 2 (same volume and mic position)
It'd be nice to see a few tweaks measured this way - braced vs unbraced, damped vs undamped, etc, to get a less subjective / waffly idea about what these tweaks actually do.
[EDIT] some extra info:
LS50 white paper (KEF document, some helpful measurements)
patent US7270215 (CLD in bracing)
Damping factor values:
damping factor values - audio qualia
Attachments
Last edited:
Even more enlightening is to measure (or just listen for it) the amount of sound that comes back out through the cone! Mount a driver in a box (e.g. on the baffle). Put a small speaker inside and play music on it. Enjoy the sound coming out through the thin paper cone! That's why boxed (closed box, vented, box, etc) speakers often sound very bad. It's why I am currently only pursuing open baffle designs...
Its why venteds can sound bad but a well stuffed, reasonably large closed box has a fighting chance. A nice tline has even better chance
A driver in the box isnt realistic, but run a noise signal, then abruptly end it and listen for hangover from the cabinet
Again: I'm mostly keen to see more articles / measurements like my examples, if y'all can link any.
...but Ima address the replies, cos I can't help myself
CharlieLaub: valid sometimes, but not what I'm after here.
Hearinspace: that's actually similar to Ludwig's test (did you read that link? It rocks). His contact mic results did show interesting things (the bit under Panel Bracing Experiments surprised me) & that he did improve his cabinet. However, the results seem exaggeratedly good, when compared to the tests using a free air mic. My concern is that contact mic tests might make a tweak seem EPIC when it is not actually audible at the listening position.
Crash test dummies don't have realistic blood.
The point is to isolate the factor that I'm asking about: sound output from speaker cabinet walls.
Using the cabinet normally with its own driver(s) would add confounding factors:
- woofer mounting technique
- rear wave radiating back out through the diaphragm(s)
- driver properties (e.g. a too large or high Qts woofer on a 'perfect' box may perform badly on the test you propose)
...but Ima address the replies, cos I can't help myself
CharlieLaub: valid sometimes, but not what I'm after here.
Hearinspace: that's actually similar to Ludwig's test (did you read that link? It rocks). His contact mic results did show interesting things (the bit under Panel Bracing Experiments surprised me) & that he did improve his cabinet. However, the results seem exaggeratedly good, when compared to the tests using a free air mic. My concern is that contact mic tests might make a tweak seem EPIC when it is not actually audible at the listening position.
Crash test dummies don't have realistic blood.
The point is to isolate the factor that I'm asking about: sound output from speaker cabinet walls.
Using the cabinet normally with its own driver(s) would add confounding factors:
- woofer mounting technique
- rear wave radiating back out through the diaphragm(s)
- driver properties (e.g. a too large or high Qts woofer on a 'perfect' box may perform badly on the test you propose)
Cabinet materials have been a major head-ache for loudspeaker manufacturers for many years.
Even when there were no other materials than particle board manufacturers have been trying to find out exactly how thin Sheets of particle board they could get way with.
Measuring cabinet resonancs (that is: resonances in the walls) cannot be done using a microphone, as the contribution from the walls will be buried in the sound from the main driver, which is the exciter of the resonances.
Instead an accelerometer should be used. Even a cheap piezo-device can give an idea of how severe the resonances are and at which frequencies they occur.
But be carefull: at some positions the wall can move a lot, and at other they don't move at all. This will take time.
Steen
Even when there were no other materials than particle board manufacturers have been trying to find out exactly how thin Sheets of particle board they could get way with.
Measuring cabinet resonancs (that is: resonances in the walls) cannot be done using a microphone, as the contribution from the walls will be buried in the sound from the main driver, which is the exciter of the resonances.
Instead an accelerometer should be used. Even a cheap piezo-device can give an idea of how severe the resonances are and at which frequencies they occur.
But be carefull: at some positions the wall can move a lot, and at other they don't move at all. This will take time.
Steen
The approach used in industry is to measure the motion over the complete surface of the cabinet and then to use this to simulate the sound at the listening position. This gives the information that is wanted about which parts of the surface are contributing to the most audible resonances.
Expensive scanning vibrometers can do this quickly. If all that is available is a point vibrometer then one would take a lot of point measurements on a grid over the complete surface. Phase is required as well as magnitude. This is tedious but easy enough. Not aware of it being done with a stick on accelerometer but it should work. Even more tedious though.
Point vibrometers cost a few thousand and most groups involved in measuring vibration will have a few around to borrow or hire. They are fairly similar to microphones to use. Software to get the sound at the listening position is freely available on the web but not in a point and click form. Alternatively one could write a script in a matlab-like toolbox if one possesses the relevant knowledge.
Expensive scanning vibrometers can do this quickly. If all that is available is a point vibrometer then one would take a lot of point measurements on a grid over the complete surface. Phase is required as well as magnitude. This is tedious but easy enough. Not aware of it being done with a stick on accelerometer but it should work. Even more tedious though.
Point vibrometers cost a few thousand and most groups involved in measuring vibration will have a few around to borrow or hire. They are fairly similar to microphones to use. Software to get the sound at the listening position is freely available on the web but not in a point and click form. Alternatively one could write a script in a matlab-like toolbox if one possesses the relevant knowledge.
A speaker in cabinet will get one clean shot at the cone on the baffle without even one wall reflection to absorb the wave. As such, its a gross exaggeration of the sound that would "leak" out of the cone, for sounds generated by the cone itself.
Acoustic Camera might be an option, too. Don't know though whether the focussing and S/N is up to the task, given that the SPL from a panel might be several ten's of dB's down from the global SPL.
I have a buddy working at the company, will ask him about this application....
I have a buddy working at the company, will ask him about this application....
If that is the one in the Adlershof science park then I have some familiarity with their cameras from a few years ago. No they are not suitable in standard form because they will not be able to see the wall vibration in the presence of the 40dB or so stronger radiation from the drivers. They will simply show some contour lines around the drivers. Acoustic cameras tend to be more useful at showing the location of the loudest sources particularly if they move around. The resolution is not high.
Having said that, if the location of the drivers is known and the radiation constant it should be possible to progressively cancel the signal in the software. However, I am unsure if the original ability to see quiet sources would remain or whether it would improve and to the extent of seeing radiation that is 40dB quieter. Even if it did the information would be easier and higher resolution using a vibrometer.
I think that many of these tests are flawed because they ignore the fact that the speaker in a real situation is also playing at the same time that the box is radiating. The speakers radiation will dwarf that of the box making the box only audible test not very reliable.
To wit, a student of mine did this test many years ago. He took a speaker and buried it in sand but with the baffle showing and he compared this to the same speaker in a free field the difference being the box radiation. He was not able to detect a difference.
I have tried many times to make a valid measurement like this and have never come up with anything meaningful.
The sound coming through the cone test is completely invalid as in a real situation because the speaker would be playing and cone movement that is not in the signal would be strongly opposed by the amplifier. This is what the impedance, i.e. back emf is all about.
To wit, a student of mine did this test many years ago. He took a speaker and buried it in sand but with the baffle showing and he compared this to the same speaker in a free field the difference being the box radiation. He was not able to detect a difference.
I have tried many times to make a valid measurement like this and have never come up with anything meaningful.
The sound coming through the cone test is completely invalid as in a real situation because the speaker would be playing and cone movement that is not in the signal would be strongly opposed by the amplifier. This is what the impedance, i.e. back emf is all about.
Long ago at Bell Labs, I had a colleague who was a well-known percussionist, Max Neuhaus, recently deceased.
He created music using contact mics on speaker cabinets and included the feedback (called "howlback" of "foldback" or something today?) in the music. Yes, you can hear the box!
I think, musically speaking, it was absolutely atrocious. I may have forgotten to throw away some of his recordings. But then that's how i feel about any "music" that isn't acoustic such as Mozart, or new instruments from Harry Partch or Uakti.
First problem is that music-instrument contact mics are not neutral and clean test instruments, like the good membrane mics that all of buy from Parts Express for cheap.
B.
Don't you think that the high gain in this loop would make your comment inaccurate? Yes, you can hear the box if you amplify it enough.
Don't you think that the high gain in this loop would make your comment inaccurate? Yes, you can hear the box if you amplify it enough.
I guess you didn't notice that that was the very point I was making.
Just wanted to a historical note about how nice acoustic instruments can be, even new ones. Mozart wrote a piece for the "Glass Armonica" invented by Ben Franklin. But the sound you get from sticking a contact mic on an instrument designed as an acoustic instrument or the sound of the sides of a speaker box are not "musical" to my ears.
B.
Last edited:
Sound coming through the cone is not what I was asking about. The links and the test proposal in link on have no external cone. They are tests just of the box.
The BBC report is the only one I referenced with the speaker set up normally, but their measurement technique seemed quite effective at picking up panel noise and discriminating against the output of the cone.
Seems like a good test. What's the context? Was this a 'normal' box, or one of yours?
As I understand it: your boxes are fibreglass shells with a thick non-resonant lining, and you also damped them using CLD and bracing that you've described as complex and based on a lifetime of study & experience.
I assume you didn't go to all that trouble just for bragging rights. You must have believed that a lesser box would have a detectable difference.
The BBC report seems to contain valid measurements.
It describes a link between measurments and audibility, indicating that around 500Hz is the critical band.
Do you mean you measured the same thing, but didn't find it meaningful? Or that you were unable to replicate their measurements?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.