We'll have to agree to disagree. I've made it work in practise, so i'm happy.
dave
Member
Joined 2009
Paid Member
Lots of good stuff here:
BBC RD - Publications - BBC R&D Reports
The particular paper:
http://downloads.bbc.co.uk/rd/pubs/reports/1977-03.pdf
David S.
Cabinet vibration curves that I have seen usually have pretty strong resonant energy up to 2kHz or so. Certainly through a woofers range in a typical 2 way. I suspect that a mid or tweeter drives the cabinet much less strongly.
In your case, can you get your resonances all above 300? Possibly although it being a large box suggests not. In the case of a compact subwoofer with steep and low crossover points you might be able to raise all resonances out of band. For any two way I think it is highly unlikely, and dealing with the resonances via damping is much more likely to work than through box stiffness.
90% of audiophiles resist the notion mightyly that high stiffness is not the ultimate answer to cabinet construction, but research shows that to be the case.
David S.
So what about materials like marble which some use for cabinets? Is there enough energy to excite ringing? How could that be damped down?
Most of his vibration spectrums showed fairly flat resonant energy to about 2000 Hz. That might have been the rolloff point of his woofer so it wouldn't preclude higher resonances with full range drive. See figure 13 in the link above.
Certainly no evidence of an f to the 4th power roll off of energy.
David S.
There is always enough energy to excite ringing. It is a linear system so the energy intensity isn't really relevant, there is no threshold.
Marble is fairly high Q so it should ring pretty good.
The thing to remember about cabinet resonances is that at resonance the cabinet is essentially transparent. If there is no loss then the mass reactance and the stiffness reactances exactly cancel and the sound gets out, but only for a narrow band of frequencies. I.e. it rings from the excitiation.
David S.
Purely a mathematical construct (Linkwitz's will likely have some measures behind it). Based on the energy vrs frequency curves (i have to refind the reference (either Olson or Beranek).
likliehood = 1/f^2 and 1/f^4.
The extra 2 powers (purely speculative on my part, but spurred by the reading of the 2 authours above) come from the decrease in energy distribution above 300 Hz, and the increase in relative size v w/l so greater damping exists.
dave
David, you're suggesting that a good way to build a cabinet is to use plywood (say 3/4") and damp the panels internally using something like roofing felt. Then for the back wave, use appropriate thickness of fiberglass or similar material to absorb the back wave. Is this a good way to go?
But wouldn't a quick test with a stethoscope reveal such ringing? If you do the test, all you usually hear is a faint sound. So how is that faint sound going to affect the loud music playing on top? It doesn't seem like it should does it?
Such is the nature of resonances, they hang on, due to their Q, until after the orriginal sound dies away. That makes them audible.
Read the Harwood paper. Jump to section 10 if you want to, where he starts discussing real cabinets (mostly BBC monitors). He found that measurements near the cabinet side tended to show a variety of resonances that peak to about 25 to 30 dB below the woofer sound. After evaluating a lot of cabinets and determining which resonances he could hear and which he couldn't he concluded that resonances that peaked to less than 30 dB below the direct sound were inaudible. Resonances peaking to higher than that were perceived as coloration. Also, for frequencies below 400 the required level became progressivly louder. So lower frequency resonances were generally less audible.
Don't forget that a near mic measurement or probe sensor underplays the large surface area that the cabinet represents.
David S.
Read the Harwood paper. Jump to section 10 if you want to, where he starts discussing real cabinets (mostly BBC monitors). He found that measurements near the cabinet side tended to show a variety of resonances that peak to about 25 to 30 dB below the woofer sound. After evaluating a lot of cabinets and determining which resonances he could hear and which he couldn't he concluded that resonances that peaked to less than 30 dB below the direct sound were inaudible. Resonances peaking to higher than that were perceived as coloration. Also, for frequencies below 400 the required level became progressivly louder. So lower frequency resonances were generally less audible.
Don't forget that a near mic measurement or probe sensor underplays the large surface area that the cabinet represents.
David S.
I need some more explanation of your energy vs. frequency notion. Are you talking about average power distirbutions of typical music? Those tend to peak in the midrange and roll off gently at the extremes. The peak level distributions are nearly flat for most music.
Or are you talking about probabilities of finding energy in a fixed bandwidth, say a 100 Hz window? That would drop as frequency goes up since a fixed bandwidth shrinks when compared to a fixed percentage bandwidth (like, say, a 1/3 Octave band).
When viewed as we hear, with logarithmic frequency scale and fixed percentage bandwidths, the measurable resonances of loudspeaker cabinets have fairly constant bandwidth or Q. As such the probability of "hitting one" does not vary by Octave. It certainly doesn't drop by frequency squared to frequency to the fourth.
Harwoods own measurements back that up. Look at his figures 13 and 14 And you can see fairly constant level peaks from 150 Hz to 3000 Hz. These were taken with a log frequency sweep which would give the same curve as pink noise analysis. If the probability of exciting a resonance were dropping you woud see it.
David S
Member
Joined 2009
Paid Member
Wow, he's saying 30dB (below axial output) coloration is inaudible, possibly even more than 30dB can be tolerated once the frequency is in the bass region.
Firstly: These seem like very high levels when you think about amplifier distortion being unacceptable to most designers when it's at the 40dB below fundamental. I assume that such high levels of box colouration are inaudible primarily because they are at the same frequency as that of the excitation - rather than at a harmonic as would be the case with amplifier distortion.
Secondly: It suggests that these levels of coloration are only inaudible when comparing a 'bad' box speaker with a 'not bad' box speaker. The empirical evidence reported on the internet gives examples of people who previously did not find their speakers to be coloured but then listened to an open baffle and on going back to their box speaker discovered that they sounded coloured. If this is the case, the threshold for audibility is very likely lower than reported in the paper and we don't have a method to bring the colourations down to a low enough level ?
Last edited:
I am talking about the energy delivered by sound at the same amplitude. It is inversely proportional to the square of the frequency. First pointed out by Svante, and a devil to find a reference to, i did eventually find a confirming equation. I need to find it again. In 1 of 3 books i have set aside by Olson or Beranek.
dave
dave
Member
Joined 2009
Paid Member
This looks like a relevant and interesting link: Loudspeaker construction
and I hadn't seen this before - uses sand even for a small box: onez_index
I'm now quite interested in the use of sand - it seems to have a lot of adherents. I can see why it's not commercially popular, but for DIY it looks well worth exploring further for control of panel vibrations.
and I hadn't seen this before - uses sand even for a small box: onez_index
I'm now quite interested in the use of sand - it seems to have a lot of adherents. I can see why it's not commercially popular, but for DIY it looks well worth exploring further for control of panel vibrations.
Last edited:
When you say sound at the same amplitude I assume you mean same SPL? Sound power ( the rate of energy flow) is proportional to pressure squared (just as electrical Watts are V squared over R). Frequency isn't a factor if the load is resistive.
We can talk about particular cases with non-flat resistance, such as a driver with a rising air load impedance where flat acceleration means flat power, but velocity falls at one rate and displacement falls at another rate. Still, flat pressure equals flat power.
David S.
I dug into where i got the linkwitz chart. It us a chart i made of some of the table in section N, here. Issues in speaker design - 2,
from that page:
dave
from that page:
dave