Lowering Mechanical Noise Floor in Speakers Pt 2
In Part 1 of this series devoted to loudspeaker panel vibration we took a look at the maths & theory underpinning the mechanics of panel vibration. Here in Part II, we'll take an empirical look at panel vibration, investigating with both accelerometer & microphone. The mechanical & acoustical performance of a loudspeaker cabinet's panel will be assessed at various stages of construction, with an eye to noting any improvements along the way. This report will conclude with a simple before/after comparison of the acoustical output of one of the test cabinet's panels, followed by a subjective assessment of the system's performance.
There are two modes of mechanical damping commonly employed in loudspeaker construction: extensional and shearing.
Figure 1: Free Later Damping Figure 2: Constrained Layers Damping (Extensional Mode) (Shearing Mode)
Extensional mode (free substrate layer) is an approach whereby a second, viscoelastic material absorbs kinetic (vibrational) energy from, in this case, a vibrating plywood panel substrate, converting it into heat (Fig 1).
The effectiveness of this approach depends upon the degree of panel flexure as well as the mechanical and physical properties of both the damping and substrate layers. Additionally, it's useful to know the mechanical properties of the adhesive used: if the mechanical impedance of the adhesive is low enough, the damping pad will function as a constraining layer. Whether this is actually desirable or not can only be determined by further measurement.
Shearing mode (constrained layer damping) as seen in Fig.2 , is an alternate approach whereby a vibrating panel is constrained by a second panel of sufficiently high mechanical impedance. In this mode, the two panels, one vibrating, the other constraining, are joined by an adhesive viscoelastic damping layer. Energy is dissipated through tensional forces of sheer deformation. Intuitively, it might appear the thicker the adhesive damping layer, the more damping. In practice, however, the opposite holds true: the thinner the layer, the more effective constrained layer damping (CLD).
So which method is best? In my experience (and all else being equal) CLD has generally provided for a higher degree of panel damping when compared to extensional mode damping, as applied to plywood cabinets constructed for mid-bass or even LF purposes. However, CLD is more costly and less forgiving where it comes to the selection of the appropriate materials. If the budget allows for it, experimenting with both approaches is recommended. If the budget doesn't allow for any experimentation, you'll likely find the less expensive, non-CLD approach works well, as you'll see in the following graphs.
In order to investigate the vibration-damping effectiveness of various materials, a test rig was built that allowed for quick, repeatable measurements of various, commonly available plywood panel samples along with various common, commercially available materials suitable for damping test sample vibration.
Figure 3 (Courtesy BBC)
Similar in appearance to that showing on Figure 3, the rig comprised two solid, heavy mount points to which the rectangular-shaped panel samples were clamped at each end.
The panel samples were set in motion by the impact of a 100 gram disk-shaped weight, dropped from a height of 30cms. The test panels were angled 45 degrees along their long axis. In doing so, the disk, when dropped on its edge, would impact, then fly off the panel, coming to rest off the panel, thus not interfering with the post-impact vibrational behavior of the individual wood samples. Low tech, but very repeatable and capable of producing consistent results.
Working up a series of before/after comparisons, as seen in Figures 6 & 7, didn't require absolute measurement values, therefore using an uncalibrated accelerometer is a valid approach. I could have used a calibrated accelerometer but chose instead to take the opportunity to experiment with an inexpensive accelerometer I'd seen mentioned in both online forums and print media.
The Measurement Specialties ACH-01 accelerometer chosen worked quite well and the +/-3dB 2Hz - 20 kHz typical amplitude response, along with a +/- 150g dynamic range made it a very cost-effective, suitable choice. It's available through retailers such as Digikey and makes for a worthwhile addition to any DIY Audioholics toolkit.
Figure 4 ; Figure 5
The wood chosen for this project from the half dozen test sample candidates was 19mm (.75") thick, void-free Baltic Birch. I use this (or its marine-grade derivative) quite a lot and have found it to be a consistently top-quality engineered wood product. (Figure 4.)
The damping material used in this case was a multi-ply "panel" built up from heavy, self-adhering roofing material. Sold in long rolls, the material is easily cut to fit as needed. Multiply the amount seen in Figure 5 by 5.5 to get an idea of the total amount used in the construction of both cabinets.
The half-dozen plywood test samples used were (free length) approximately 70 cm (27.6") long and just over 12 cm (4.7") wide. The free length was chosen as being a reasonable approximation of a typical loudspeaker cabinet's panel height. The width was chosen to keep mechanical impedance low and to minimize transverse resonance modes. The results of the test series was both encouraging and enlightening. The results showing in Figures 6 & 7 are taken from the Baltic Birch test sample.
Whether you use an inexpensive accelerometer, such as the ACH-01 or a more expensive calibrated accelerometer, a few caveats are in order. First, owing to the sensitivity of accelerometers suitable for measuring loudspeaker panel vibration, it's easy for extraneous vibrational noise from such things as nearby fans, motors, traffic, etc to creep in to your measurements. Second, be consistent in your use of whatever base or support your cabinet under test is situated upon. Measure your cabinet while it rests on, for example, the concrete floor of a garage, then re-measure it, this time with the cabinet perched on top of a wooden workbench and you'll be looking at very different results, invalidating even comparisons based on relative amplitude measurements.
On the left hand side of Figure 6, we see the amplitude response generated by the accelerometer fixed to the undamped panel sample. On the right hand side we see the response of the same sample but with the damping material lightly affixed to the panel. (Both impulse responses are on the same amplitude/time scale). Already a significant difference is evident. Clearly, the plywood panel, when damped, settles much more quickly than when undamped.
The damped response showing in Figure 7 is that which resulted when the damping material was directly adhered to the plywood test sample.
In Figure 7, an improved settling time can be seen when comparing the damped amplitude response showing here with that seen in Figure 6. Clearly, adhering each layer to its neighbor, then adhering the multi-ply damping panel directly to the plywood test sample is an effective approach.
Its tempting to think that when it comes to the application of a damping material, the more the better. However, there exists a diminishing returns effect and inevitably there comes a point where more is not better. From an empirical standpoint, building a damping pad of the particular material used in this project ½ to ¾ the thickness of the plywood panel substrate is a good starting point when setting out to determine optimal thickness.