As reported in klog 29785, the transfer function from the ground to the board on the stack is not modeled well as a triple-stage harmonic oscillator. Some large couplings from other degrees of freedom were also found in klog 29818. Here I'm trying to modify the model to explain the measured transfer function by adding some unknown coupling to the simple oscillator model.

Attached figure is the first trial of the modification. Red curve is the true (?) horizontal to horizontal response. I assumed it as a single-stage harmonic oscillator. Blue curve is the transfer function of a triple-stage harmonic oscillator, which I introduced as a coupling from another degree of freedom. Green curve is the sum of the red and blue curve. Although there are many discrepancy, this model seems to fit the measured data better than the previous triple-stage model.

I just estimated the coupling to fit the model qulitatively and I didn't care about the quantitative characterization, like the resonant frequency of each mode and the coupling factor. Next step is to estimate these values and discuss whether this model is reasonable or not. The source of the coupling is another important point. It could be introduced by the shaker, it could be explained by a rigid body model.