== Finished ==
[Sugimoto, Enomoto]

We measured the phase (frequency) noise of a 5 m fiber with a Mach-Zehnder interferomter (see the first picture).
The result can be found in the second attachment. This result shows that Δf ~ 2e-3 Hz/sqrt(Hz) in f > 10 Hz and 3e-2*(1Hz/f)^2 Hz/sqrt(Hz) below ~3 Hz. 

= How we measured =
We locked the MZI at mid-fringe. The power of the interference light was measured by a PDA100A, and the output of the PD was low-passed after the mid-fringe offset was subtracted.
LPF and offset subtracting was realized by an SR560. The cutoff frequency of the low pass was 100 Hz.
The output of the SR560 was then sent to a PZT driver. The output of the PZT driver went to the laset PZT.
The measured OLTF is shown in the third attachment.
The actuator efficiency (input of the PZT driver to MZI phase) was also measured, by sweeping the input of the PZT driver with 10 Hz triangular wave (see the fourth attachment).
The measured efficiency is: 2π/0.992 V = 6.33 rad/V. The error of this should be within ~ 5%.
Since we are interested in low frequency region, we meaured the PSD of the control signal (input of the PZT driver) to estimate the frequency noise of the fiber.

During the measurement the clean air fan was turned off, and we put a few irises to eliminate nasty scattered light.

I also attach the pictures that show the contrast and the lock acquisition.

= Notes on the result =
-- Around 100-300 Hz, there is a structure probably due to acoustic noise.
-- In high frequency, the noise is limited by the electrical noise, more explicitly the noise of the spectrum analyzer.
I know this can be lowered by band-passing and amplifying the signal, but I didn't; I don't care much about high frequency.

= Next =
Assuming this result can be scalable to the KAGRA real 50m fiber, I am going to implement this into the ALS noise budget.
If the noise could be critical, we might need to consider measuring the noise of the 50m fiber after the installation.