KAGRA Logbook

AOS (General)

tomotada.akutsu - 18:47 Monday 30 January 2017 (2407)

Leak tests for the oplev viewport windows

[Yoshida, Hoshino (Technoport), Akutsu] with helps by Sawada, and preparation by Kimura

We have measured leak speed of every remaining viewport window for oplev.

Type | Purchased (pcs) | Tested today (pcs) | Leaking (pcs)

ICF152 | 40 | 19 | 14

ICF70 | 10 | 5 | 4

ICF152

The background of the leak detector showed typically 1e-9 Pa m^3/s, but for those terribly leaking one it showed 1e-7 Pa m^3/s.

Typical ones (11pcs) showed 1e-8 ~1e-6 Pa m^3/s leak. 2pcs showed 1e-5, 1pc showed 2e-3.

5pcs (S/N 0009, 0010, 0011, 0012, 0013) are surviving in terms of this background.

ICF70

The background of the leak detector showed 1e-9 Pa m^3/s for the only surviving one.

3pcs showed 1e-6 Pa m^3/s leak. 1pc showed 1e-8.

1pc (S/N 010) is surviving in terms of this background.

Cu Gasket

One suspected that the use of usual (not yet anealed) Cu gaskets could cause of leak. Using ICF152 S/N0013, one of the surviving ones, we tested the effect of the difference of anealed Cu gasket and usual one. Beacuse we only tested single piece, the statistic is weak, but no difference was found; for each case the flange of the viewport window was tightend 7Nm, 10Nm, and 15Nm in this order, and every time the leak was measured. For the anealed one, only aflter 7Nm case it leaked, but for the other measurement shows no change from the background of 1e-9 Pa m^3/s.

Images attached to this report

Comments to this report:

tomotada.akutsu - 23:11 Monday 30 January 2017 (2411)

Due to my poor English, my last post was not easy to understand in some sense. I'd like to summarize it somehow here.

Note that the requirement to every viewport window written in the procurement specification document (in 2013) was that the leak needs to be meausred as less than 1e-10 Pa m^3/s. Today the leak detector for our measurement had a background of 1e-9 Pa m^3/s. So what I like to say is that even this easy measurement, such large number of viewport windows were identified as failed.
In this sense, the measurement setup today can detect leak larger than 1e-9 Pa m^3/s; the viewport windows identified as "surviving" in my last post means that the leak detector did not show significant changes when He blew the window bonding area.
When a viewport under test with large leakage was set to the leak detector, its indicator did not smoothly get down to 1e-9 Pa m^3/s immediately after "evacuate" button was pushed. So we can predict the bad result for such a viewport case even before we blew He.
Even though only one sample was tested, there are no significant differences in terms of leakage whichever an anealed copper gasket or a usual (not necessarily anealed) copper gasket was used to tighten the surviving viewport window.

tomotada.akutsu - 14:07 Tuesday 31 January 2017 (2413)

The surviving viewport windows for oplevs are now stocked in one of the desiccators at the side of the IMC duct.

Images attached to this comment