I performed the TF measurement locating the bottom ACC (TEAC707LFZ) on the top, to calibrate the frequency response between the two ACCs.
As a result, the discrepancy was found below 10Hz, and to be -30dB at 1Hz.
This value is inconsistent (too strong) with the Stack TF(V) results, -20dB at 1Hz. [https://klog.icrr.u-tokyo.ac.jp/osl/?r=29785]
Also, it is inconsistent with the Stack TF(H) results,

I measured the transfer functions of this shaking table setup (No stacks, only weight), locating the ACCs at the default positions.
Through the weekend, some eigenmodes shifted.

I performed the same measurement, by changing the DAC amplitude from 1V to 0.5V.
The magnitude below 10Hz becomes smaller.

Dosen't the enviromental temperature change related to the shift of resonant frequencies??

The temperature in this booth was not changed within 0.1℃ resolution.

I guess the reason was the spacer rubbers (to avoid glitches, the same series to be used in the stack) became harder by pressing.
A similar behavior was found in the Mitaka tests (https://klog.icrr.u-tokyo.ac.jp/osl/?r=29809) 

I measured the transfer functions from Z acceleration (shaking direction) to the X/Y acceleration.

Very large couplings are found.

It could be couplings from rotational degrees of freedom (Pitich, Roll, Yaw).

Is it possible to put 6 accelerometers at least to decouple these DoFs?

Indeed ideally, but we don't have enough time and resources, and strong motivation.

[R.Takahashi, Washimi]

We updated the Stack TF measurement setup.

• Changed the shaking plate from the thin aluminum plate to a thick SUS plate (31kg weight), to avoid the eigenmodes of the plate around 120 Hz.
• Changed the spacer rubbers below the shaking plate, from the spare rubbers of stacks to thin rubbers to avoid the eigenmodes being fakes of those of the stacks.
• Changed the position of the bottom accelerometer from the edge to the center of the shaking plate, to avoid the tilt modes.

In this setup, the tilt of the shaking plate was small.

We located the "Stack 3" extracted from the OMC chamber, which was the most pressed stack (see https://klog.icrr.u-tokyo.ac.jp/osl/?r=29272), on the shaking setup, and also +155kg weight was added.
The gaps were not touched but were very closed and not the same for each place. (We will check the condition after one night, tomorrow.)

We roughly measured the transfer function of it from 10 to 500 Hz as a test. This setup looks fine. More detailed measurements will be done tomorrow morning.

[R.Takahashi, Washimi]

We performed the TF measurements for the OMC Stack3 (old) with vertical shaking.

At first, we measured V -> V (it means vertical. not voltage) acceleration after one night and compared it to yesterday's result.
Some details were changed, but global changes (like frequency shift in klog29736) were not found.

After that, during rubber replacement works for Stack 1&2, we measured the TFs from V to other Dof (X, Y, pit, yaw).
Note that the pit (corresponds to X) and yaw (corresponds to Y) measurements were just putting the ACC like the picture, not taking a difference or ratio of two positions.
These data are located at JGW-T2415843.

We performed the TF measurements for the OMC Stack3 (new) with vertical shaking before and after loading the additional weight (+155kg). The gaps between each stage were large enough and no
screw heads were touched (see pictures). 
After that, the TFs from vertical to the other DoF were also measured with the weight.
These data are uploaded on JGW-T2415845.

The TF of the new Stack3 without additional weight was a simple structure and good for comparing theoretical models.
The TF of the new Stack3 with +155kg weight had some peaks over 100 Hz. They might be couplings or mixing from the other DoF, caused because the position and angle of the weights were not perfect.

We measured the TFs (vertical to vertical) for each stack, with the +155kg weight (same).

These data are uploaded on JGW-T2415848.