Meeting on the Tracker Flexure/Grid attachment issues
December 3, 2003
The meeting began with a discussion of the test results.
Tom V said that there was no obvious change in the signature run until after –3dB run in X. However, some nonlinearity seen throughout the X run, with continual softening in the random vibration. However, there was still the same signature until after -3dB, at which point a small extra lobe appeared on low side of the f0 peak. In Z there was only a slight downward shift (<<5%) in signature after each run (sine 5-50 Hz, then random up to –6dB).
Erik said that from the peaks in x and y in the z run (i.e. crosstalk) there was maybe some evidence of loosening already. He noted increased off-axis response and new peaks in the x,y accelerometers in the later Z signature runs.
Chris noted that occasional slippage will not show up in frequency plots. The drop in resonant frequency occurred only after preload was lost. He also cautioned that some softening seen in the data could be partially due to other fasteners, in addition to the flexure-mount bolts.
Erik said that they say 3 to 6 mil slipping in the static test data. The static test article always bottomed out on the shoulder well before the full load was reached. They were using a steel fixture, not aluminum.
Martin wanted to know what was the shoulder bolt material and what was the helicoil material?
There was a discussion of torque spec and preload requirements. There seems to be inconsistencies between different specs. Also, the spec vary a lot depending on whether lubrication is used. It was concluded that we should be lubricating the threads in order to have predictable preloads. (Johns said that some helicoils come with a lubed surface, but this tends to wear off with repeated use.) It was noted that we should measure how much preload we get from a given torque.
Robert asked whether, if the preload issues could be solved, and secondary locking implemented, would it be acceptable to slip +-2 mils occasionally during the vibration. The consensus was that this was not really acceptable, that we should not be relying on friction at all, and that we really need to get back to the original intent of the design, which was to have the tower supported on shear pins.
John Ku noted that bending of the bolt from the tower load could easily be enough to start losing the preload, allowing the bolts to start backing out. He backed this up with a presentation http://www-glast.slac.stanford.edu/Tracker-Hardware/ART/vibe/TKR_Anomaly-StrucAnalysis.pdf
In his presentation John pointed out that 4-sigma excursions, and even more, are likely for a 140 Hz resonance in random vibe for 2 minutes. Chris said that controllers most often will clip at 3 sigma, but whether that was the case at Alenia was not known.
When questioned, Tom B said that the grid holes are 2 mils larger in diameter than the shoulders on the bolts. For the flexure holes the difference is 4 mils.
It was resolved that we must get stronger bolts (not 70 ksi) for the next test.
Martin went through his review of the relevant Tracker drawings.
One problem that came us is that the specifications of the pins in the assembly drawing of the corner bracket+flexure is not adequate. We need to know the pin material. Tom said that it is hardened steel allow, ground to size and shape.
Tom B also confirmed that assembly of the flexure & corner bracket will include some epoxy. However, the two will be assembled permanently by the vendor of those parts and assembled onto the bottom tray already assembled.
Martin asked if the assembly fixture drawings available, since they also flow into the tolerance stackup. Tom said that he has been trying to get all the tooling drawings from Italy. However, the bottom tray tooling is presently under revision.
Martin started drawing some concepts on the board. In total 4 similar concepts were discussed. See the drawings in http://www-glast.slac.stanford.edu/Tracker-Hardware/ART/vibe/TKR_Design%20options.pdf.
Secondary locking mechanisms discussed:
A few random points that came up:
To move forward, we decided to pursue Martin’s conical bushing concept for the corner flexures.
Issues needing action:
We did not want to use the same method for the side flexures. A proposal to use a liquid bushing (a la the calorimeter) was generally received well. Tom will look into how the existing flexures could be readily modified to accommodate this. Martin to talk with Rob Black regarding injection and liquid shim.
Tom V put up a list of issues to be settled prior to the next vibe test: