Measuring the sound radiation of speaker cabinet walls
Last edited: Nov. 25, 2017
Measuring sound radiation of speaker cabinet walls is a real challenge. The conventional frequency response measurement methods fail, so we have to look for other ways. Unfortunately, most of these methods require professional tools, and the majority of DIY loudspeaker builders are discouraged from measuring sound transmission loss. Around 2013 I came up with a new idea how to measure sound insulation of speaker cabinets with a very simple method, using a closed back speaker only.
Several attempts and methods are known for measuring the sound radiation of speaker cabinets. Some of them are based on vibration measurement data, while others use a microphone for the measurement. Note that the panel vibration doesn't represent the acoustic output directly so additional calculations have to be performed.
Here is a brief description of the best-known methods:
Measuring panel vibration with an accelerometer.
Cons: Coupling to the panel has to be very stiff and resonance free, besides the vibration analysis is carried out at one point only and not on the whole surface. Suitable for low frequency measurements only.
Close field measurement with a bidirectional (figure of eight) microphone.
This is a relatively simple method, which is described in detail in a White Paper from 1977 (see Factors in the design of loudspeaker cabinets by H.D. Harwood and R. Mathews). The essence of the measurement is that microphones with figure of eight characteristics show about 20dB smaller sensitivity for sounds arriving from the side than arriving from the microphone axis. The drawback of this method that the microphone has to be placed very close to the wall, so it is not possible to perform distant (far field) measurements. Another problem is the proximity effect, which is well known for a point source, but not for direct radiators and large panels.
Extra cabinet method.
The cabinet under test is built into a wall, or a larger cabinet. Only one side of the cabinet is facing to the microphone. Another variation of this method when the cabinet under test is built around a smaller enclosure.
Measuring panel vibrations with scanning laser Doppler vibrometer, and calculating the sound transmission loss with FEA software.
Modern non contact (and expensive) method, which is based on finite element analysis and boundary element analysis to predict the acoustical output of a panel from panel surface acceleration.
I came up with a very simple method how to measure speaker cabinet sound insulation. I choose a tweeter with very low resonance frequency, and during the sound insulation measurements I've turned it inwards the cabinet. Theoretically any kind of loudspeaker with a low resonance frequency and with closed chamber suits, if it is possible to achieve 90dB/m sound pressure from 100Hz upwards without distortion. Another criteria is that the cabinet should not alter the response of the speaker, more closely the T/S parameters, when the speaker faces inwards. Fortunately, this criteria is true if we use closed back tweeters or midranges.
The test cabinet and the Vifa DX25TG05-04
For the tests I used a Vifa DX25TG05-04 tweeter (Fs: 650Hz, Qts: 0,67). The test signal was a sine sweep with 2V RMS. I made the recordings by ARTA software.
When I measured the reference signal the tweeter was turned outwards (as a normal speaker in a normal cabinet). I made the reference signal measurement from a distance of 5cm and 20cm from the tweeter's faceplate and on axis. When I measured the side walls I turned the tweeter inwards the cabinet. I covered the tweeter with a flowerpot which was filled with sound absorbing material. The sound radiation measurements of the cabinet walls made from a distance of 5cm and 20cm from the center of the panels.
Finally I corrected the response between 200Hz and 1000Hz because the tweeter response shows a gradual decrease below 1000Hz (see lower figure). The results are appreciable above 300Hz - it's enough because the side panel resonances are higher in frequency.
Frequency correction between 200Hz and 1000Hz
The external dimensions of the test cabinet: 27cm * 27cm * 39cm. The larger sides of the cabinet are made of MDF, particle board, birch plywood and OSB sheets with thickness of 18mm (I measured these only). The net volume of the cabinet is 19,3 litres. Unfortunately, before the measurements I added side bracing to the particle board, so I do not report measurement results for this.
The air closed into the enclosure shows standing wave resonances at 485Hz, 970Hz, 1455Hz in the longitudinal direction, and at 1470Hz in lateral direction. These resonances (especially the one at 1470Hz) are very strong due to the bad geometry, in a well-designed cabinet they would be smaller by 6-10dB.
The test cabinet
The real advantage of this method, that we can measure the reference signal and the sound transmitted through the panels with the same tools. What we have to do only is to flip the speaker and that's all. We don't need computer simulation software and acceleration sensor or an extra cabinet, only a tweeter or a closed back midrange with low fs. The drawback of this method is that the mechanical vibrations generated by the woofer are missing from the measurements. (Side note: I think that the resonance at 200 Hz comes from the magnet vibrations of the Vifa DX)
In a few words on the graphs. The light green is the reference, the blue and the black are the sound radiation measurements of the panels. The blue is measured with empty cabinet, the black is measured with cabinet filled 60% with sound absorbing material. The difference between the reference and the sound radiation is the sound transmission loss (sound insulation) in decibels.
OSB measured from 5cm
OSB measured from 20cm
MDF measured from 5cm
MDF measured from 20cm
Plywood measured from 5cm
Plywood measured from 20cm
And finally, two graphs for comparing MDF with plywood. The MDF has a stronger panel resonance at 640 Hz.
Plywood vs. MDF from 5cm, empty cabinet
Plywood vs. MDF from 5cm, cabinet stuffed with polyester fill in ~60%
It is visible from the measurements that the sound absorbing material plays a very important role in reducing the sound pressure level inside the enclosure and hence decreasing sound radiation from the cabinet. Due to the foam the intensity of the sound transmitted through the walls decreased by 10dB on average. The sound insulation for the plywood with filling is between 30-40dB, which is not so bad.
There's only one question left: What is the best wood for speaker enclosures? The difference is not so great between wood types as commonly believed. Plywood has better sound transmission loss than MDF in the resonance region, but it's hard to recognize the difference in subjective tests.