1. 1 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 M. Sullivan for D. Arnett, S. DeBarger, S. Ecklund, C. Field, A. Fisher, S. Gierman, P. Grossberg, M. Petree, K. Krauter, A. Kulikov, E. Miller, K. Sonnad, N. Spencer, K. Underwood, U. Wienands Machine Advisory Committee Review October 25-27, 2006 Fast Luminosity Dither Update

2. 2 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Outline Present luminosity feedback New fast feedback –General Characteristics –Hardware –Software Progress this last year Summary

3. 3 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Present Feedback system Presently we keep the beams in collision by dithering the HEB while monitoring the luminosity signal and then move the HEB to the LEB in order to maximize the luminosity signal. We use 8 correctors in the HEB (4 for X and 4 for Y) to move the HEB beam back and forth in the X, Y and Y angle dimensions. The feedback moves the beam left then right, then calculates the peak of the luminosity with a parabolic fit and then moves the beam to the newly calculated peak. The process is repeated for the Y dimension and then for the Y angle. There is at least a 5/8 sec settle time for the corrector magnets for each beam move. This means the present system speed is dominated by the corrector magnet settle time. There are 5 magnet moves for each dimension and three dimensions so it takes at least 8 sec to complete a full go-around of the feedback. Several times over the last few years we have thought of trying to make a faster luminosity feedback system, but could not come up with a location for some air coils until we finally thought of using locations near the HER fast separator magnets. These magnets need a thin stainless steel beam pipe which is exactly what the new system needs.

4. 4 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 New Luminosity Feedback The new idea is to use air coils that can be driven at frequencies of around 100 Hz. We simultaneously excite the coils at 3 different frequencies – one for each dimension and detect the 3 signals with separate lock-in amplifiers. The lock-ins deliver an amplitude and sign for each dimension. The sign indicates which direction to move. The feedback algorithm then uses this information to decide how much to step in each dimension. The 3 steps are then added together and the magnets are moved in all 3 dimensions at once. We should immediately gain a factor of about 8 in speed and hope to improve that even more by using an improved algorithm that decides how much to move the beam.

5. 5 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Luminosity signal We have a very high rate luminosity signal that we have not been fully utilizing Clive Field says that at luminosities of near 1e34 nearly every crossing produces a hit in his detector. This corresponds to a signal rate of well over 100 MHz. Our present minimum luminosity signal is 1e30 which should still have a signal rate of something like 10 kHz The background level for the luminosity signal is less than 1e29 at low to moderate beam currents Work is going on to provide a couple of signals. The trick is to get a fast enough response time without too much noise.

6. 6 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Coil Design and Hardware The design and construction of the coils include: –Inductance calculations and measurements –Field penetration and phase shift measurements –Field uniformity calculations –Power handling capacity – air cooled The kick angles needed for the beam movements were calculated from the MAD model of the HER. This gives us initial coefficients to make closed bumps around the IP. There are two different beam pipes for the coils so calculations and measurements had to be made for both locations. There are two very good presentations by Steve Gierman and Alan Fisher on more details at these links: –http://www.slac.stanford.edu/~sullivan/Fast_Dither/2005_10_19%20Design %20Review/fast_dither_19oct05%20Gierman.pdfhttp://www.slac.stanford.edu/~sullivan/Fast_Dither/2005_10_19%20Design %20Review/fast_dither_19oct05%20Gierman.pdf –http://www.slac.stanford.edu/~sullivan/Fast_Dither/2005_10_19%20Design %20Review/Fast-DitherDesgnReview-2005-10-19%20Fisher.pdfhttp://www.slac.stanford.edu/~sullivan/Fast_Dither/2005_10_19%20Design %20Review/Fast-DitherDesgnReview-2005-10-19%20Fisher.pdf

7. 7 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Coil Hardware Coil set at 50 meters Coil set at 30 meters

8. 8 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Lock-ins From Alan Fisher’s design review presentation

9. 9 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Initial study on how to minimize the response time from the lock-in and still be able to get a clean signal

10. 10 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006

11. 11 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006

12. 12 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 More Hardware We have in hand the Lock-ins We have found and refurbished 8 power amplifiers (KEPCOs) The circuit board with the frequency additions and with the driver circuits for the power amplifiers has been designed, sent out for fabrication, and has just come back from the vendor last week. The mother board and the 8 daughter boards are being stuffed in preparation for checkout.

13. 13 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Software We have a team of people from the software group helping to put together a software package for the fast dither We will keep the present system intact and build another package for the new hardware. The present system then becomes a backup. The present SLAC fast feedback software package can readily accommodate the new fast dither feedback design We plan to use a “Newton search” algorithm to decide how far to move. We have a simulator in software to study the algorithm. We are keeping an up-to-date flow chart to make sure everyone understands how the feedback works. We have the interface panels to the control system built

14. 14 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Fast Dither SCP panel Fast Dither Lock-in panel

15. 15 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Block diagram for the fast dither FLUM M. Sullivan Jun. 7, 2006 Rev. 11 SAM 66 Algorithm Gierman, Sullivan, Wienands, Himel, Krauter, Miller, … PR02 Interface to correctors Other feedbacks HERO, LERO, etc. Bit-Bus Lock-in signals X (x,y) Y (x,y) YP (x,y) Lock-ins Freq. control EPICS Fast Coils Interface board 24 coil coefficients Power amplifiers Lumi signals SCP Coil coeffs. (24) Dither amplitudes (3) Dither windows (6) Default step sizes (3) Newton search boundary (3) Phase lock thresholds (3) Minimum lumi signal (1) PR00 DC correctors 12 magnets DACSAM 8 8 24 3224 32 24 31 Magnet job ampls. Analog status 31 333 6

16. 16 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Algorithm 1 2 Total Lumi Phase lock signals 2 2 X Y YP XRE, XIM YRE, YIM YPRE, YPIM Lumi threshold for starting feedback Default step sizes when outside Newton search zone BitBus commands to DC correctors Boundary definition for stable Newton search zone Minimum amplitude thresholds for lock in scale change 1 Inside Newton zone search zone compute magnet move Outside Newton zone XMOVE = XSTEPONZ *SIGN(XLOCK) Newton search zone boundary XMOVE  0 XSLOPE > XSLOPEMIN XMOVE YMOVE YPMOVE Fast dither step sizes Sanity checks of input signals RE >3*IM if VALUE > MINVAL If signal is too big then request scale change 6 Fast dither windows Accuator task Computation – controller and measurement task Compute XLOCK YLOCK YPLOCK Convert MOVEs to magnet delta BDES Add all 3 for one move LUM (1of 4) Actuator status Maximum amplitude thresholds for lock in scale change (WARN, BAD) 63 33 3 Scale windows 6 Zone status 6 Move windows

17. 17 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Progress since last MAC review All work on this project stopped until we had fixed the IR 2 vacuum problem Most of the names in the list on the first page were heavily involved in either building hardware to help pinpoint the source or fully engaged in efforts to collect and analyze data to help find the source We picked up again in May and got the remaining hardware built and then bench tested and debugged by near the end of June The mother board and 8 daughter boards were plugged into the system and we started to test the hardware parasitically with a first draft of the software package in July

18. 18 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Progress (part 2) We could clearly see a signal from the lock-ins when we deliberately moved the beams out of collision. The Y signal was very clear. The X signal was not as clear but could also be seen. After some initial software debugging issues got resolved, we started to run the program in the background while the accelerator was delivering luminosity to the detector. In August we were in the position of being able to switch the new software in and switch the old software out so we could see how well it worked when the accelerator was stably running at peak luminosity

19. 19 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 50120 60 708090100110 Fast Dither Frequencies YPXY 8710373 June 8, 2006 A. Fisher M. Sullivan U. Wienands We started up with these chosen frequencies

20. 20 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 50120 60 708090100110 Fast Dither Frequencies XYPY 8710373 August 28, 2006 A. Fisher M. Sullivan U. Wienands We are now using these values. The X signal and the YP signal have been switched.

21. 21 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Progress (part 3) Just near the end of the run in August we started to get the system going We have seen the feedback work in the Y plane while we were steadily running We were ready to try both planes (X and Y) at the same time when the run ended At this stage we are running the feedback in the simple state of just moving the beam in the right direction by a fixed step

22. 22 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Plans Get the feedback working in the X and Y planes simultaneously. First in the simple “step in the right direction state” Get all 3 planes working at the same time Get the “Newton search” method working – initially with steady state running Get the feedback to work while injecting from scratch – one beam or both beams Get the transition from simple “step in the right direction” to “Newton search” working We should be able to get the feedback working well enough to be able to give it to the operators very soon after the startup

23. 23 IR Fast Dither M. Sullivan MAC Review Oct. 25-27, 2006 Summary We are on the verge of using a new fast dither luminosity feedback system The new system should be about 10 times faster than the present system This shows great promise for a more accurate and quicker luminosity tune-up of the accelerator. The faster feedback should also improve machine stability by getting the beams quickly back into collision when something in the ring shifts the beam orbit(s)