The matter of laboratory evaluation of a rack is a thorny case. Although we have already seen that vibration control by such a construction is measurable and plays a role, and also that the vibrations from devices and their effect on electrical output is a real phenomenon (minor in absolute levels, but measurable by conventional methods), the field continues to stagnate in terms of measurement methodology and standardized specifications.
This time, we decided to adopt a stricter method than those used in the articles referenced above. The strategy involved a specialized excitation device (exciter), the standard accelerometer we use in speaker resonance measurements (with a sensitivity of 9.5mV/g and a ±150g measurement range), and four different excitation methods—i.e., ways of transferring vibrations to the rack. These included, initially, seismic excitation (don’t worry: “seismic” simply refers to excitation via statistically random noise), harmonic excitation (the exciter is driven by sine wave signals—sweeps or tones), excitation via an MLS signal (yielding data for various graphs we've long used in speaker testing), and finally impact excitation using a specialized non-rebound hammer—a method that reveals the impulse response profile of a structure, and in digital form is currently used in predictive maintenance of large machinery. Some of our more seasoned readers might recall the wheel-tappers at railway terminals performing a similar task on train wheels.
Among all the above, the only method that didn’t perform as expected was the MLS analysis. The reason was that the accelerometer signals were too weak, with a signal-to-noise ratio—particularly at low frequencies—not adequate to yield meaningful results (we’ll return to this at another time, for sure). Data analysis was performed using the usual analyzers we have in the lab (dScope III and Clio), along with a specialized software package for vibration evaluation—VibInspect from Revibe Energy.
The key goals of the measurements were threefold: First, to determine whether the materials and structure used in the Model 1 conceal any resonances that might impart character to its vibration absorption profile. Second and third, to examine how vibration attenuation occurs at two reference levels: when excitation happens outside the rack (i.e., before the spikes and their sockets, with the exciter placed on the base surface supporting the rack), and inside the rack—i.e., after the spikes, such as with a device generating significant mechanical noise (a power amplifier with a large transformer or a CD transport, for example).
Harmonic excitation using a sine wave sweep with the exciter outside the rack clearly showed a strong vibration-damping capacity. Compared to the excitation profile itself, accelerations on the top shelf were 22dB lower, cutting down pronounced resonances observed around 1200Hz and in the 3–5kHz region. The addition of a plinth further improved performance by nearly 6dB in the 2.5kHz region. These findings suggest that the rack does indeed isolate environmental vibrations quite effectively.
What happens when vibrations are generated within the rack? For this case, we analyzed a 440Hz tone (equivalent to the musical note A in the third octave). In this test, the exciter was placed on the bottom shelf and the accelerometer on the top. The corresponding graph shows a tone attenuation of about 3dB from one shelf to the other. This indicates that the Model 1 does reduce vibrations originating from the devices it supports.

Also quite interesting was the wide-band version of this measurement, in the 2kHz–10kHz range. Here, it is clear that the exciter produces a range of high-frequency components that are sharply attenuated and fully absent in the accelerometer's spectrum. We're talking about significant attenuation—upwards of 5dB in some cases. Macroscopically, the "top" shelf is visibly quieter than the "bottom."
Seismic excitation was performed using statistically random noise, with the exciter placed on the wooden surface where Model 1 was positioned during testing. For these measurements, the accelerometer was initially placed near the excitation point and then on the plinth of the top shelf. Since the signal is stochastic in nature (i.e., random), analysis was done using a PSD diagram (Power Spectral Density), a graph showing how acceleration is distributed across frequencies. The reference surface plot shows a clear series of components both in the low and higher frequency regions, just above 2kHz. This is the familiar excitation profile we’ve seen in other measurements and can identify as characteristic of the excitation itself.

The PSD from the shelf shows lower acceleration power in the low end and near-complete attenuation of high-frequency acceleration. This suggests excellent vibration damping from environmental sources (speakers, traffic, other machinery). As for the impact excitation, the first measurement was of the reference surface itself. Here, the accelerometer was placed very close to the impact point. The acceleration values were large but rapidly damped, with minimal resonance.

The measurement on the top shelf showed significantly lower acceleration values and slightly longer damping times. Both findings are reasonable, considering that between the point of excitation and measurement lies the... entire rack, adding mass and a small—but real—amount of elasticity.
The measurement on the top shelf’s plinth showed even greater attenuation of acceleration at the cost of slightly increased ringing. The very low acceleration levels are evident from the relative rise in sinusoidal noise. If you run the numbers, you’ll find this corresponds to a 50Hz component.

The rack’s behavior under lateral impact excitation is also quite interesting. In the first case, the excitation was aligned with the short axis of the shelf (as if hitting it from the front). One can observe oscillation at relatively high frequencies with increased accelerations that decay fairly quickly.
Impact excitation aligned with the long axis (side impact) resulted in lower acceleration and slightly slower damping with a component in the low-frequency region. Of course, neither measurement reflects real-world conditions a rack would typically face. No one expects the shelves to be kicked! Nonetheless, they show that the construction is rigid and robust, with no tendency to flex or deform under lateral force.

Finally, we examined the rack’s response to direct impact on the spacer columns. For this test, an impact was applied to one of the top assembly bolts (front left as viewed), and acceleration values were measured at the other three corresponding points (front right, rear right, rear left).
The findings were intriguing mainly because there wasn't much to find. The three measurements were nearly identical in terms of damping time, acceleration magnitude, and damping profile.

How to interpret this? One possible explanation is that the rack's propagation properties are directionally consistent, meaning that the Model 1 is a homogeneous structure that does not add any of its own characteristics.

Previous | Next | More Reviews