Summary

I performed Omicron and Hveto for investigating the cause of this sudden power reduction on GreenX. The cause might connect with magnetic phenomenon around EXC.

Details

I prepared the page to summarize Omicron triggers, by combining gwsumm and Omicron.

Here is a summary of Omicron trigger for GreenX. This link is only available from control room. So, I also attache the screenshot.

Top left figure shows the timeseries of second trend. We are investigating the cause of this sudden power reduction.

Second Left figure shows the scatter plot of Omicron triggers on time-frequency map, and the color indicate SNR. As you see, Omicron can catch the sudden power reduction as high SNR triggers (SNR>=50).

Next I applied these triggers to Hveto. Scripts are located in /users/yuzu/work/0216_hveto . The aim for aplying Hveto is to identify the coincident channels with this sudden power reduction. So, I set the threshold for SNR as 50 and frequency range as 20~1000Hz.

Here is the result of Hveto. 

22 triggers of magnetmeter at EXC (K1:PEM_MAG_EXC_BOOTH_EXC_Y_OUT_DQ) were coincident with triggers on K1:TMS-X_GR_PDA1_OUT_DQ. One triggers of rest 32 triggers was coincident with ACC sensor at EXA (K1:PEM_ACC_EXA_TABLE_RX_Z_OUT_DQ).

I run Hveto by removing the magnetmeter at EXC (K1:PEM_MAG_EXC_BOOTH_EXC_Y_OUT_DQ) from the channel list. 3 triggers of ACC sensor at EXA (K1:PEM_ACC_EXA_TABLE_RX_Z_OUT_DQ) was coincident with 54 triggers of GreenX. Result is Here.

This analysis suggests that the cause of sudden power reduction or instability might connect with magnetic phenomenon around EXC. This page summarizes the Omicron triggers for that magnetmeter and shows the perodically happening glitches.

Although this analysis covers the oplev channels, Hveto didn't show coincidence between oplev channel and GreenX. I guess that that coincidence happens only when the SNR is small (such as <50). The other possibility is that I'm using Hveto wrong...

To show the covered channels, I attach the channel list which is used for coincidence analysis.

Next step

• See the timeseries at the glitch (conveniently the trigger list at the bottom of these pages are connected with Pastavi !)
• Perform the field survey with PEM specialist
• To find the other cause, relax the SNR threshold from 50. 

Thanks so much for your work. In Omicron, did you also include some physically meaningful/relevant channels such as GreenX input power or so? Considering the physics, they would be the channels worth including in the evaluation.

Thank you for the comment!

I talked with Ushiba-san and he gave me the relavant channels such as green input power, laser control, PLL. I will try the analysis again. We also found the PR3 oplev channels were not included in this analysis.
Actually this is the first time for me to touch Hveto analysis. Probably I will be mature gradually.