Dec 2006 Warm Servicing

This servicing took place over Dec 8-18 at the Mimir Clean Room located on the first floor of the Perkins Telescope building.

Present were Dan Clemens and Brian Taylor, with consulting from Marc Buie.

 

Mimir off telescope, on support cart Mimir, with outer stainless steel cryostat cover and internal cold shield removed, in clean room

Areas and issues addressed included:

Cryogenics/Cooling

    During the period September until warm-up in late Nov, Mimir was experiencing a lack of adequate 2nd stage cooling, which prevented holding to the desired 33.5 K detector temperature. Many workarounds were tried during the fall (purging, warming the helium shed) but none proved effective. This period featured the "new" (used) DD-model CTI 1050 cold head. We sent the "old" CP 1050 off for refurbishment, and expected we would have to install it into Mimir during this service period. Instead...

    1. Test chamber put back into service
      • Added heaters and temperature sensors to the two cold stages of the (just returned from refurbishment) CP 1050 cold head, then installing the cold head in the test chamber, vacuum pumping, attaching the helium system, and running the cold head for a day.
      • The refurbished head failed to cool to proper operating temps (it reached 112 K and 18K, instead of the 41K and 13K expected).
      • We did a complete flush/purge of the helium lines and compressor and changed out the adsorber in the compressor.
      • We also did a complete flush/purge of the CP cold head (5 x 5 = 25 cycles)

     

    2. Cleaned up DD Cold Head

      • We opened the DD Cold Head and found the 2nd stage ring seal, while intact, was missing its backing split-ring spring. This was found to be located up under the first stage displacer, along with the usual ring seal cover (a full ring, not split). In August, the split ring was put outside the ring seal when installing the cold head into the SS cylinder in Mimir. This caused the split ring to slide upwards and to not keep the ring seal pushed against the SS cylinder wall. This ring seal failure seems to be the right explanation for the poor 2nd stage performance seen all fall (while the 1st stage seemed to work mostly normally).
      • We cleaned the displacers (ethanol soak, blown out with dry N2, then baked at 60C in N2 atmosphere overnight), and reinstalled the DD cold head into the SS cylinder.
      • We purge/filled with Helium (5 x 5 = 25 cycles) and ran it in the tests chamber to recertify its operation.
      • It produced good no-load temps (41K and 15K) and a reasonable load curve.

       

    3. Returned CP cold head for repair

      • With the success in the test chamber of the cleaned up DD cold head, and the similar failure of the CP cold head, we judged the CP had failed acceptance testing and so sent it back to the firm that refurbished it.

       

    4. "Lecture Bottles" removed from system

      • Brian had already found that our "lecture bottles" expansion volume systems were leaking, so they were taken out of the system during the late fall
      • While we had the DD cold head on the test chamber, we did an experiment with and without one lecture bottle (on the supply side of the cold head), finding no measureable improvement in cooling capacity with the bottle.
      • We concluded that other than noise reduction, they had no benefit and would no longer be in the helium system

       

    5. Helium Gas Operating Pressure

      • Also while the DD cold head was still on the test chamber, we experimented with reducing the pressure of the supply-side helium delivered to the cold head, looking for reductions in cooling capacity.
      • This did not occur until the gas pressure was lowered to 215 PSI, considerably lower than the 290-300 PSI we had been operating at.
      • We have decided to henceforth operate at 245+/-10 PSI to provide adequate cooling, but much better lifetimes for the parts in the helium system.

       

Filter Wheel Interferences

    The POL and FW3 filter wheels would once in a while stop rotating while Mimir was operating. This happened because the clearances between the filter wheel faces and the nearest bulkheads were very small (32 mils, by design), the wheels carry some small warp, and the filter wheel "stack" needed internal shimming to set the wheel locations properly. From the previous warm servicing, this amount was judged to be about 15 mils of additional shimming between the stack that starts with FW3 and ends with the POL wheel. Shims were designed to cover the inner portions of the axle bearing races for the FW3 and FW1 wheels. This has the effect of pushing FW3 "back" and the others forward, especially POL. The shims were 12 mils thickness each. It did prove necessary to replace the hex-head 4-40 screws that hold down the POL wheel position encoding magnet blocks with button-head 4-40 screws to remove a *new* interference introduced between these screws and the back of the collimator. Once the screws were replaced, the POL wheel turned freely with no interference.
    After rebuilding the filter wheel/axle stack, and reinstalling into Mimir, no interferences of any kind were found in the four filter wheels. All turn easily both directions and for all filter locations. The filters were reinstalled (with 3 unfinished filter cells added in FW2 and FW3 to improve balance) and the full systems tested in both horizontal and vertical orientations of Mimir. All wheels worked fine, all encoders worked fine.

     

Slit/Decker Unit Upgrades

    The slit/decker unit has not been working reliably or accurately for the past two years. The problems were: (1) inability to reliably position the slit or decker to the same location in the beam, (2) drift of the slit with hour angle by about 1 full slit width, and (3) slow maximum speeds of the slit and decker car motions.

    1. Positioning reliability improvement

      • We believed that the positioning problem likely stemmed from the nature of the deep pin-based detent unit attached to each stepper motor
      • A new detent system was designed to run on the drive screws instead of on the motors
      • Parts were fabricated, some anodized, and were installed on the slit/decker unit
      • The detent forces were tuned, and the detents sensed using reed switches.
      • These systems (one for each of the slit and decker drives) seem to produce somewhat smoother motions and allow higher step rates (by up to a factor of three or so).
      • The old motor-based detents were removed from the unit.

       

    2. Slit sag improvements

      • Examination of the drive systems identified four places where play could introduce slit wander or sag
      • The first of these places, namely the motor gear to drive screw gear mesh, was measured and found unable to contribute enough motion to explain more than 1/10th of the slit motion. Also, with the new drive screw detents, this gear mesh is no longer "down stream" of the detent, and so will no longer contribute any slit motion.
      • The second place is in the drive screw "nut" that captures the rotating drive screw and turns it into car motion. However, this nut is comprised of a set of recirculating ball bearings, with many in contact with the screw threads at any time. Pushing on the nut unit developed no more than 1 mil of relative play, and so was judged unimportant (the slit wander is some 7-10 mils).
      • The third place is in the bearings and springs used to locate each drive screw relative to one end pilon for each of the slit and decker. In fact, the decker drive screw was essentially floating free along its length, though the slit's drive screw was not. This was fixed by adding additional wave washers to the end of the decker drive screw that captures these wave washers between a flat washer and one shoulder bearing in the opposite side end pilon. After this tuning, both decker and slit drive screws were judged well located and fixed to 1 mil or better.
      • The fourth place where play was evident was in the slit and decker cars themselves. Both showed strong torsional motions - the ability to twist about the single bearing race holding their "tops" on one steel rod.
        • We measured the importance of this torsional motion by attaching a dial indicator to the baseplate, then indicating off the side of the slit car.
        • When the unit was rotated to simulate different hour angles on the telescope, the dial indicator showed linear motion of the order 7-8 mils, right in line with the wander of 1 slit width seen on the telescope.
        • The cars are held onto the two opposing steel idler rods via sets of recirculating linear bearing units, themselves held to the slit and decker cars by C-clips.
        • However, the linear bearing units had some play with respect to their cars.
        • Shim washers were made and installed between the C-clips for the linear bearings and the cars.
        • This cured the torsional motion problem enormously. We reperformed the dial indicator experiment, recording a maximum motion of 2.5 mils at the end of the car farthest from its drive screw. This should translate to under 1.25 mils of motion of the slit center.

Close up, pumping

    Mimir was closed up on Sunday, Dec 17, and pumping with the new scroll pump started. About 24 hours later, with the dewar pressure about 0.120 torr, the turbo pump was added and run in "standby" mode (low RPM) until the pressure was below 0.050 torr when full RPMs were allowed. The pressure fell to 1 micro-torr within a few hours. During all this time, the cold bulkhead heater was on, to drive that part up to 320 K (under closed loop control) to aid in getting water off the anodized surfaces. This bakeout will cease after 48 hours.