Today we tested the position dependency of the tiltsensors on the Narahara breadboards on each pylon.

Attached plots are the specta [urad/rtHz] on each case. Following are the explanations of the plot labels.

• Pos 1. wider angle across the sender pylon, mirror and receiver pylon.
• Pos 2. narrower angle across the sender, mirror and receiver. In this case, the laser source and sensors are located roughly center. We though this configuration might be better not to excite pylon too much.
• Pos 3. even narrower of them.

Effects of wind shields were tested:

• Heavy wind shield (WS): Wind shields prepared for iKAGRA consists of acrylic panels alumni colums. They are fixed on the breadboards by one or two dog clamps.
• Light WS: Just tin foils covered entire breadboard and the light paths towards the viewports.

Also, effects of clean booth fans were tested; on or off. If no labeled, fans were on. Curves labeled as "different laser" is another test to use a cheaper version of diode laser as a tilt sensor'ss light source.

Tiltsensor posisions
The positions of the sensors does not seem matter very much. Pos 2 seems good at 1-10 Hz but it is not clear if it is the position effect or a good timing of air flow.

Windshields
On the other hand, the heavy windshields make the pylon mechanical motion much louder above 10 Hz. The main mechanical resonant mode seems ~35 Hz, showing up both in P and Y. Akutsu san suggested that probably even heavier shield fixed tightly on the breadboards may be better not to hit the pylon (like drums and drum sticks) this much.

Pylon's other mechanical peaks 
There is a big peak in PIT at around 4.5 Hz which seems like a mechanical resonant. It is not clear if it is coming from the pylon or the mirror motion itself. You can see the second and third harmonics-ish peaks, too. If this is a mechanical motion of pylon, it is not very nice. If it's pylon, we may have to fix the pylon on the ground tighter with three clamps, instead of the two clamps like the current situation.

Conclusion
The tiltsensor, with fans OFF and with the light WS, can get a good spectrum which is likely limited by the pylon motion above 10 Hz, and by the mirror motion below 0.1 Hz. From 0.1 to 10 Hz, it is not clear if the mirror motion is seems or the pylon motion is seen.