Recent Work at the Stanford Engineering Test Facility

Transcription

1 1 Recent Work at the Stanford Engineering Test Facility Tarmigan Casebolt, Dan DeBra, Matt DeGree, William East, Brian Lantz, Norna Robertson, and the SEI team March 22, 2006 Special thanks to SUS, Calum, Janeen, Caroline, Justin, Tim

2 Various Activities 2 Low noise GS-13, use a simple op-amp follower circuit Noise is ~6e-12 m/rthz at 1 Hz, 3e-13 m/rthz at 10 Hz Matt DeGree working on tickle testing, automated check of all the sensors and actuators. William East measuring thermal behavior of actuators. BSC is OK Large DC offset, measure 32 mdeg of temp rise. Scale to typical ETF operation, get 6e-4 degrees Scale to BSC, softer springs, HEPI input motion: 2e-6 degrees Quad pendulum frame interactions (Brian and Tarm)

3 Quad Pendulum Frame 3 Received a frame from Caltech Install on the Tech Demo Study the impact of frame resonances on the system Not so great Try to improve the interaction Electronic damping (OK) Passive Constrained Layer damping (Great)

4 Installation installed 12/16/05 Upside down ge sta 2y e2 g sta x stage 1 x, global x top view of frame installation G

5 Interaction Performance Tests 5 Goal: Try to understand the impact of the frame vibrations on the system performance. But: Testing in air makes performance measures difficult. So: Predict performance by: Measuring mechanical transfer function of stage, and Multiplying by calculated suppression of the isolation loop. We see that: Mechanical transfer function is worse. Control loop performance is worse.

6 Largest Coupling to rx & ry 6 TF mag (dspace center SS/dspace center drive) 10 0 Stage 2 Rotation plant, rx pend frame dummy mass freq (Hz) Coupled frame modes at 61 and 73 Hz s, stage 1 and stage 2 damped, stage 1 isolated

7 Performance Impact of resonances To predict performance: multiply passive transmission by the sensitivity 10!1 ETF Stage 2! transmission (norm(plant) * sensitivity), rx, before and after 2. Much worse at peaks G Somewhat worse between Hz with pendulum frame mag (V/V) 10!2 with dummy masses (scaled) 10!3 10!4 Note: I have equalized passive transmission at 10 Hz by dividing dummy mass by dummy masses pendulum frame created by design_txty_stg2_ on 18!Dec!2005

8 What to do? 8 SUS - Working to increase the frequency of the modes Damp the modes: Actively, using existing sensors, or using new sensors Passively with constrained layer damping

9 Active damping control loop 9 Final control stg2 V Mag (cnt/cnt) Freq (Hz)

10 Result of active damping Impact of Frame Damping on Stage 2 rotation mode rx 10 10!1 mag TF (super out/ drive in) 10!2 10!3 10!4 original frame damping created by plot_frame_damper_ on 10!Jan!2006 data _1 and _ freq (Hz)

11 Requires many irritating notches Notch alignment for stg2 V3 Mag (V/V) !1 10! Freq (Hz) created by design_stg2_framedamping_ on 20!Dec!2005 data = tf_121705_1_stg2_cen_all

12 12 Passive damping of frame mode Use vibrational motion to create to create shear in a lossy material (Dyad 601 by SoundCoat) Lossy material placed between frame and something which moves differently. FEA modeling courtesy of Tim Haylor at Rutherford Labs

13 Test setup Constrained Layer t stru stru t stru t top bracket foot 13

14 Damping strut performance 14 Impact of Frame Damping on Stage 2 rotation mode rx 10!1 mag TF (super out/ drive in) 10!2 10!3 10!4 original frame damping 10.2 cm 2 created by plot_frame_damper_031506_btl on 16!Mar!2006 data _1 and _ freq (Hz)

15 Optimizing the layer Impact of Frame Damping on Stage 2 rotation mode rx f = 60.9 Hz Q = 654 original frame damping 2.56 cm 2 frame damping 5.11 cm 2 frame damping 10.2 cm 2 mag TF (super out/ drive in) f = 61.7 Hz Q = 67.6 f = 62.3 Hz Q = 57.8 f = 63.7 Hz Q = freq (Hz) created by plot_frame_damper_ on 20 Mar 2006 data _1, _1, _1, and _1

16 Conclusions 16 System works with the quad pendulum frame. It will work better if the peaks are smaller amplitude and higher frequency. Demonstrated 2 ways to improve the damping. We prefer the constrained layer: much easier for Advance LIGO operations. Eager to help the SUS team get it working in vacuum. Making good progress dealing with the issues identified at the last set of reviews.

17 Largest Coupling to rx & ry 17 TF mag (dspace center SS/dspace center drive) 10 0 Stage 2 Rotation plant, ry pend frame dummy mass freq (Hz) Coupled frame modes at 61 and 73 Hz s, stage 1 and stage 2 damped, stage 1 isolated

18 TF mag (dspace center SS/dspace center drive) 10 0 Stage 2 Rotation plant, Y Stage 2 Rotation plant, X Smaller coupling in other DOF TF mag (dspace center SS/dspace center drive) pend frame dummy mass pend frame dummy mass freq (Hz) freq (Hz) _xyzrz, STAGE 2 ISOLATED IN rx and ry 18 STAGE 2 ISOLATED IN rx and ry

19 Improved Active Damping 19 Sensors on the frame tip give better signal for active frame damping Compare sensors for frame damping, V3 drive 10 1 TF mag (dspace readout/dspace drive) !1 10!2 local GS13 frametip 1 * freq (Hz) created by plot_aux_framedamping_sensors on 20!Jan! grams + preamp +cable + mount + can

20 Changing the undamped Frame 20 10!1 Impact of Undamped Frame on Stage 2 rotation mode rx f = 60.9 Hz Q = 684 f = 61.2 Hz Q = !2 f = 60.9 Hz Q = 285 mag TF (super out/ drive in) 10!3 10!4 dec 17 March 10 March freq (Hz) created by plot_frame_damper_ on 16!Mar!2006

21 Servo for rx ETF Stage 2, rx G basic control mag (V/V) !2 plant notches plant*10 controller/10 notch openloop*10 sens open loop sensitivity 10! phase (degree) 50 0!50 ty_stg2_ on 18!Dec!2005

22 Servo for rx 10 2 ETF Stage 2, rx G pendulum frame To notice: 1. Servo amplifies above 13 Hz. 2. Hair about 70 Hz Hz attenuation is modest: 4. Big notches 5. It works. Dummy mass loop is: 1. Better performance. 2. More robust. 3. Easier to design. mag (V/V) phase (degree) mag (V/V) ! !50 plant*10 controller/10 notch openloop*10 sens 10! !2 plant*10 controller/10 notch openloop*10 ETF Stage 2, rx dummy masses ty_stg2_ on 18!Dec!2005