I compared the coefficients of the BF damper LVDT with the SF GAS LVDT. When the SF keystone was actuated by 1000count to COILOUTF, the displacement in the BF damper LVDT was about half of the SF GAS LVDT. They should be the same. 

With Takahashi-san.

Summary: we calibrated the BF Damper vertical LVDTs.

We calibrated the SF LVDT and the BF Damper vertical LVDTs. There was no known problem with the SF LVDT, only with the BF Damper ones. However, we also measured data to double-check the previously measured value (klog 19431).

The table below shows the new calibration factors that must be  written in the medm screen. Comparison with the previous values from the medm screen is also shown. The new ones for the BF Damper are 1.86 times larger than the old ones.

I wrote in the medm screen the new values only for the BF Damper Vertical LVDTs. I kept the current value for the SF LVDT  because the new value is close enough (within 10%) . I accepted the changes in the SDF (lines 7, 8 and 9 in screenshot).

Summary: I calibrated the BF horizontal LVDTs.

It is an approximate calibration because it relies on the oplev yaw readout, which I assume is reliable. In case we find it is not reliable, we can repeat the calibration if necessary.

I converted the values in µrad to µm by multiplying by 408 mm / 1000, where 408 mm is the distance of the BF Damper LVDT to the centre of rotation (per JGW-D2113320-v5).

In average, the new calibration factors are 1.8 times larger than the old ones.

The old calibration factors have a minus sign. That is likely to make it consistent with the actuation at the BF Damper (see in the worksheet the actuation values used to move the BF Damper). However, it's not consistent with the oplev. I'll check the signs before committing the new values to the medm screens and the SDF.

I checked PRM OpLev signs.
Attached figures show the matrices and filter banks between each segment signal to EUL signals of OpLev, and their functions are as follows:
Fig1: Filter banks for dewhiteing of each segment signal. gains of this filter banks should be -1 for compensating inverse of QPD circuit.
Fig2: Matrix for calculating normarized horizontal and vertical signals from each segment signal.
Fig3: Filter banks for calibration of OpLev signal. Gains of this filter bank should be calibration factors to pitch or yaw.
Fig4: Matrix for selecting QPD signals to sense longitudinal, pitch and yaw. Only Type-B suspension uses LEN_PITCH to LEN signal because their OpLev beam travels vertically.
Fig5: decoupling matrix for diagonalization.

Since Definition of segment is
SEG1: lower left
SEG2: lower right
SEG3: top right
SEG4: Top left,
the sign of OpLev yaw signal is correct: + sign is corresponding to rotation alogn anti-clockwise viewed from the top.
So, it is likely that BF LVDT signs are wrong.
We are now looking for the old calibration data of BF LVDTs for confirming the sign definition.

Original calibration of the BF LVDT for PRM was performed on the bench in June 2017 (klog). According to the logbook (picture by Hirata-san), the + direction is CCW for H. The actual sign of the BF LVDT depended on the installation situation (cabling, so on).

I implemented new calibration factors for PRM BF horizontal LVDTs (fig1).
At the same time, I flipped signs of COILOUTF gain for compensating sign flip of calibration factors (fig2).

SDF was accepted after the change (fig3).

After the change of BF horizotal LVDT calibration factors, TFs of all PR suspensions become consistent (fig1).