Synchrotron-Light Monitors Alan Fisher PEP-II Machine Advisory Committee 2006-01-19

Topics Commissioning the LER’s x-ray monitor Turn-by-turn measurements with a rotating mirror Beam-abort diagnostics Video abort buffers Rotating mirror Streak camera Plans IP beam size Bunch-by-bunch profiles

LER Synchrotron X-Ray Monitor At the last MAC meeting, I presented the design of the SXM, which was installed and waiting for beam. Dipole radiation from Arc 7, via a modified photon stop Pinhole, scintillator, video camera Commissioned with first beam last April.

Typical SXM Image Images have expected calibration. Same size with all 4 pinholes. Size is above resolution limit. Ellipse is strongly tilted. Tilt increases from 10 with LER only at low current, to over 50 in collision. Tilt agrees with Yiton Yan’s MIA code for single-beam at low current. A coupling maximum near the x-ray source point.

Beam-Abort Movies Frame grabbers for synchrotron-light monitors have 4 channels (beam image, interferometer, alignment,…). Software saves a 9-frame movie every time it switches to the beam-image camera. Spans 300 ms with 30-Hz frames. Previous movie is overwritten, unless the beam has aborted. Then the movie is transferred to a buffer. User in Main Control can view and save the movie.

SXM, as instability is about to start. Frame -8: Stable Beam SXM, as instability is about to start.

Frame -7

Frame -6

Frame -5

Frame -4

Frame -3

Frame -2

Frame -1

Frame 0: Movie Ends, Abort Follows Final frame recorded. BaBar radiation triggered an abort afterward.

Turn-by-Turn Transverse Imaging Concept Image the transverse profile of a single bunch: Over many turns In a full ring See changes over 100 or 1000 turns: During steady running During injection, as charge is added to the bunch Investigate injection backgrounds Instability preceding a beam abort

Gated Camera Gives transverse (x and y) images of the beam. Gate can be as short as 2 ns. Can capture single-bunch images in a full ring. Can be gated repeatedly on successive turns. Image intensifier (microchannel plate) gives sufficient brightness. But…image readout is too slow. Repetition rate is a few Hz. Cannot resolve x or y evolution over consecutive turns.

Sweep Image across Gated Camera Project the transverse image onto one axis. Measure only one transverse dimension at a time. Turn the beam ellipse into a thin vertical stripe. Two cylindrical lenses image the synchrotron light with different magnifications in x and y. Fire the camera’s gate on one bunch over many consecutive turns (or every nth turn). Use a rapidly rotating mirror to sweep this projection across the camera. One camera readout then captures many turns.

July 2004: Dissected Bar-Code Scanner Quick proof-of-principle test in LER at the end of the 2004 run Aperture of gated camera Cylindrical lens for horizontal demagnification DC motor with servo Rotating octagonal mirror, 3 mm high, 13 mm across

2004 July 30: 64 Turns

What Needed Improving? Cannot be triggered to take data after injection. Motor is free running, not synchronized with PEP. Scanner issues a timing pulse on each turn of the polygon. Used to trigger the camera. Motor starts camera, rather than a PEP timing signal. Only measures one axis of the beam. The polygon is too small: 3 mm high, while camera’s photocathode is 12.4 mm high. Needs demagnification vertically, making it harder to do even more horizontally. Dissected scanner can’t be mounted rigidly.

Scanning Galvanometer Mirror on a servo motor Used for laser scanners, often in pairs for raster scans. Built for stable mounting. Can get a bigger mirror: 14.2 mm high  8.7 mm wide Mounts on VM500 motor (smallest one in photo). Driven by a ramp waveform Triggered by PEP timing.

Measuring x and y Simultaneously Split the light, then rotate one beam by 90. Image one path onto the top half of the camera, the other onto the bottom half. Two sets of stripes swept out across the camera. Different magnifications on each path. Major-axis projection of beam ellipse viewed on top half of video image, minor axis on bottom. Paths must be equal within 2 ns to catch the same bunch in the camera’s gate. New optics installed in summer and fall of 2005 (after PEP resumed operation) for both rings.

Optical Table Layout for LER Hutch Scanning mirror Cylindrical lenses Gated-camera locations Turn-by-turn imaging Bunch-by-bunch imaging (also for streak camera) In-plane periscope (no rotation) Color filter Polarizer Out-of-plane periscope (rotates 90) Flip-up mirror (bunch-by-bunch imaging or streak camera) Beamsplitter Incoming light

2005 Oct 8: Stable Beams, 100 Turns Major axis Minor axis

Injecting into an Empty LER Bucket 2005 August 5 44 turns, starting near first turn Hard to get enough light to see a small injected bunch: Viewed y axis only Summed 3 images Triggered on PEP injection fiducial Images don’t show any specific bucket, but whichever one is getting injection.

Adding a Second Injection New charge oscillating around previously injected bunch

Now the Third Injection

Measuring Beam Aborts The problem of “fast LER dI/dt” aborts, not injection backgrounds, became my objective in November. Triggering to view milliseconds before an abort: Image one bunch every nth turn over a longer time. Image every 80th turn (587 ms) 125 times = 10,000 turns (73 ms). Trigger on the ring-turn clock, not the injection fiducial. Repeat as fast as possible. 2 Hz due to camera’s readout and transfer to computer. Chance of capturing any one abort is then 73/500 = 15%. Disable camera’s trigger on an abort. Abort logic signal was available near LER hutch. HER hutch is remote, needed tunnel access to pull a 200-m cable. Abort logic should be completed this week.

2005 Dec 7: Blow-up in x and y in <5ms Correlation with BPM and transverse-feedback abort data shows: Centroid motion in BPMs for ~1 ms. Seen here in last image before blow-up. Current drops for ~3 ms while beam blows up. Abort triggered by LER current loss (dI/dt). 75 ms

2005 Dec 7: Transverse Instability

Longitudinal Axis: Streak Camera Abort spectra showed 2ns: Quadrupole motion? Borrowed streak camera from Berkeley (ALS) at Christmas. Set-up On LER table. A small focus, no cylindrical imaging. 500-ps fast time sweep vertically. Dual axis: 20-ms sweep horizontally. Image one bunch every 250 ms. 80 images across the screen, with overlap, making a stripe. Images separated by 133 ms (eight 60-Hz periods). 15% chance of capturing any one abort. But aborts since Christmas caused mostly by IP vacuum, not dI/dt. Have not yet caught a dI/dt abort, but have taken interesting images.

Dual-Axis Streak of Stable Beam 500 ps 60-Hz camera blanking Bowed envelope: 60-Hz subharmonic on trigger? 20 ms

Radiation Abort: No Longitudinal Motion

Adding Charge to a Bucket during a Fill After first injection. Dim image:  0.1 mA in bunch. Second injection enters, a bit late in phase. Image brightens. Injected charge oscillates, starts to damp. Next image, 133 ms later. Oscillation has damped.

Longitudinal Growth and Recovery 9 ms

What’s Next? Install HER abort monitoring. Cable has been run; interface board is almost ready. On-line beam sizes at IP using machine model. I have started to add this code to my SLM software. Add bunch-by-bunch capability to x-ray monitor. Presented to the MAC last year, as an upgrade option. Scan an x-ray mask through the image plane. A joint project with Caltech (if their grant is funded), with application to the ILC damping rings as well.

Bunch-by-Bunch Transverse Profiles Slots scanning across a 5s beam ellipse Modeled on a wire scanner. A rotating x-ray mask based on modified optical chopper wheel. 100-mm-thick tungsten. 3 moving slots on image plane. Form projections on x, y, and u (45) axes as slots move. Followed by a 1-ns scintillator and PMT. Fast digitizer and electronics (FPGA) sort pulses by bunch. Get profiles for fitting. Will use the “IGP” board being built as a longitudinal feedback upgrade.