Dynamic tests measure the response of physical systems to inputs that are not constant with time. In contrast to static testing, dynamic testing characterizes systems in terms of quantities such as frequency and time responses. In today's real world applications, dynamic testing is a necessary supplement to analytical methods. In many applications, dynamic tests can provide a cost effective alternative to analytical techniques. CSA commonly uses innovative and unique test methods to assess structural performance, and generate solutions that are cost effective. Examples below document some techniques used by CSA Engineering in the past.

Performance Predictions of a Vibration-Sensitive
System Using Test-Acquired Data

Many systems used by the semiconductor and electronics industries are sensitive to disturbances from floor vibration from sound pressure deviations, and from their own internal moving parts. Examples are E-beam scopes, wafer profilers, photo-lithography systems, and others. Most manufacturers use a generic standard to indicate the acceptable vibration threshold for operation. Yet, meeting a generic criteria is no guarantee that a system will actually function; vibration sensitivity is very much a function of design.

In one recent project, the dynamic characteristics of a wafer production tool were determined by test and used as the basis of an analytical model that predicted error in machine performance as a function of floor vibration. The result was a powerful diagnostic and design tool. Manufacturers may use such a model to predict the error produced by anticipated vibration environments before the system is shipped and installed. The dynamic model can accommodate stationary random or transient base vibration. The system shown in Figure 1 is similar to the device to that was served by the model.

Modal Tests on an Operating Transmission

Peak transmission loads can be elevated by resonant properties of a gearbox assembly, particularly in cases where dynamic inputs are applied. In many systems, resonant properties can change as a function of speed and load across the gear train. Acquiring modal data at operating conditions can alleviate uncertainties produced by differences in speed and load from nominal.

Instrumenting components inside a transmission for a modal test can be expensive because the components to measure are rotating. However, gear speed measurements can be acquired using the gear teeth as pulse generators for an encoder. Instrumentation need only mount to the gearbox housing, not on rotating components. A signal proportional to instantaneous gear speed can be constructed by demodulating the output from encoders.

In a test performed by CSA Engineering, excitation forces were applied by superimposing white noise on the command to motor armatures. Figure 2 shows a similarly proportioned gearbox. Instantaneous gear speed was sensed by frequency demodulation of the output from rotary encoders. Results were reduced to frequency response functions, and standard modal extraction methods followed. The technique provided excellent confidence in modal parameters at a fraction of the price of the alternatives.

Predictions of Tilt Response Generated by Discrete
Force and Torque Sources

Vibration disturbance sources commonly need to coexist on the same optical bench with vibration sensitive components.Response predictions of optics aboard an optical bench are commonly performed with a finite element model and dynamic forcing functions. Direct measurements of structural sensitivities can be a cost effective alternative in cases where the structure has already been fabricated.

The disturbance sources may be motors, effectors, or other moving components. Fast beam steering systems are one example. Reactions from these steering systems are rotational in nature, and so are  critical response quantities for optics. Rotational sensors are less common than translational, and frequency response functions of rotation output to torque input were calculated from admittance functions in translation. Although the technique requires careful test and data reduction techniques to generate precise estimates, the approach is commonly a cost effective one.

Measured frequency response functions can be used to predict the response generated by discrete excitation sources without relying on analytical models. Potential error sources are reduced to measurement accuracy, with no other modeling assumptions.

Measurements were used to predict jitter of a beam at the output of an optical system. The model was created directly from measurements, optical relationships, and expected reaction forces, and it enabled prediction of system jitter due to excitation forces applied by any of the sources aboard the bench.

Direct Complex Stiffness Measurements Enable Uncommon 
Precision in Finite Element Predictions

Direct measurements of complex stiffness can improve the precision of a finite element model greatly if a structure is generally simple, yet complex at a few locations. This case is common to many truss structures. Joints of truss structures can be complex and dynamic behavior is not always accurately predicted by finite element modeling alone.

The SPace Integrated Controls Experiment (SPICE) required high fidelity from the structural dynamic model. The SPICE truss is shown in Figure 4. The experiment required a highly accurate dynamic structural model both inside and above the control bandwidth. CSA performed the complex stiffness tests of truss elements, the modal analysis, and much of the correlation between test and analysis.

The experiment demonstrated excellent fidelity of the finite element model. After tuning, the model matched 26 modes with a cross-orthogonality of .85 or better and with less than 5 % frequency error.

Contact brad.allen@csaengineering.com to discuss CSA's capabilities and your requirements.