Minimising disturbances to the beamline users

The most essential diagnostic in the storage ring of a third generation light source is the beam position monitoring system and its associated orbit correction system. The quality of such systems largely determines the final beam stability which is of utmost importance for many beamlines. The ESRF storage ring is equipped with 224 (32 cells x 7 BPM/cells) beam position monitors (BPMs) located at regular intervals along the ring circumference. Each BPM station is constituted of 4 insulated buttons which capture weak radio frequency (RF) signals when the electron beam is passing through. The relative amplitudes of these weak signals depend on the position of the beam. The electronics processing these signals have been in use for 17 years. They were originally developed in-house as no suitable solution was available from industry. The four weak RF signals from the BPM buttons are multiplexed, and then filtered, amplified and demodulated by a unique signal processor, including the ADC. This concept allows a number of imperfections in the signal processor to be cancelled-out. In recent years higher performance electronics have been commercially available; they make use of modern digital electronics. As part of the upgrade programme, the ESRF has decided to upgrade its BPM electronics as well as the steerers and orbit correction system. A digital electronics system named Libera-Brillance has been selected. The replacement of the old electronics by a Libera required some recabling which makes the change nearly irreversible. Following extensive tests carried out in 2008 on a few Libera systems, the decision was made to spread the replacement over a 5 month period in order to allow smooth continuous operation of the accelerator systems and the beamlines with full beam stability. One of the most critical issues was to maintain the same absolute position accuracy for every beamline before and after the BPM electronics replacement. The work took place partly during shutdown and during MDTs. As the new system gave a new positional offset for each BPM station, a precise measurement and cancellation was performed following each of the Libera electronics installation. Thirteen out of thirty-two cells were upgraded during the winter shutdown of 2008-2009. The remaining cells were upgraded during Machine Dedicated Time (MDT) in batches of one to three cells (7 to 21 BPM). This meant four hours of beam interruption during each MDT day followed by 8 hours of BPM offset measurements made with beam in the storage ring. The reproducibility of BPM positions measured before and after the upgrade as well as the very few complaints from beamlines concerning orbit changes leads us to believe that the goal of minimal disturbance to the user programme was indeed achieved.


Improved performance and functionality

The new Libera-Brilliance system yields a strongly improved performance in all modes of operation. The increased resolution and reduced noise content was most vividly demonstrated at the time when part of the storage ring had been upgraded while another part was still under the old system (see Figure 161). The system also includes functionalities like a beam position interlock (which kills the beam in case of excessive beam displacements that could damage part of the ring vacuum chambers) and various flexible buffers for data acquisition. The control of the BPMs is made through the Tango control system. It includes 224 device servers running on 4 dedicated powerful PCs running under Linux and connected to a dedicated Ethernet VLAN with 1 GBits links. Each BPM is controlled by a device server. On top of these 224 device servers is another Tango device server which collects all horizontal and vertical position data using Tango group calls allowing a synchronised measurement of all BPMs within 15 milliseconds.

Fig. 161: Intermediate situation recorded in January 2009 with 20 out of 32 cells equipped with the new BPM electronics. The graph shows the difference between two vertical orbits along the ring circumference performed at intervals of a few seconds. The BPMs on cell 8-29 are processed with the new Libera electronics and shows orbit change in the order of 1 micrometre rms. Indeed the pattern observed is typical of that observed in a real orbit drift while the noise in each BPM sensor is much lower. The signals from BPMS 29-8 are processed with the old electronics and show a much larger noise introduced in the electronics itself.

Thanks to the highly flexible and fast digital electronics processing, several flows of data are simultaneously available with different acquisition rates and averaging, resulting in different noise levels. The normal mode of acquisition is slow acquisition (SA) in which position data is output at 10 Hz, which provides a 100 ms time integration and results in the lowest noise measurement. The SA mode is the default mode of operation based on which the orbit is corrected every 30 seconds.

Another mode is the fast acquisition (FA) mode with data processed at 10 kHz with a higher noise level and a low latency (<130 microseconds). The FA mode will be used in future DC-AC orbit correction systems stabilising time varying orbit deviations in a frequency range continuous from DC to a few hundred hertz. The 10 kHz data stream will be distributed over a network of fibre-optic links, using a dedicated broad-coast protocol which is executed in each Libera unit, to a number of processors which drive in real time the 96 corrector power supplies located all along the ring circumference.

In addition to these SA and FA modes each Libera generates data turn-by-turn at the storage ring revolution frequency of 355 kHz. This data is stored in buffers, which are synchronised within all BPM stations at precisely 9.3 ns. The turn by turn mode is interesting for several reasons. It includes orbit measurement at high frequency (up to 170 kHz). The turn by turn data is essential in the case of re-commissioning of the storage ring after major interventions. It also serves in beam dynamics studies such as phase-advance and amplitude of betatron oscillations inside the magnet lattice of the ring. Figure 162 presents a typical horizontal position measured turn by turn over the first 400 turns on a BPM following the excitation of the beam by a horizontal kick. Figure 162 also presents the horizontal phase space trajectory of an electron in the middle of a straight section obtained by processing the position from two adjacent BPMs.

Fig. 162: a) BPM horizontal measurement turn-by-turn over 400 turns following an horizontal kick. b) Phase-space plot of the horizontal beam motion in the middle of high beta straight section of the ring reconstructed from the turn by turn data of two adjacent BPMs.


Improved emittance measurement

Until now, the emittance measurements had essentially relied on two independent X-ray pinhole camera systems. One installed on a soft-end dipole in cell 9 (with large energy dispersion) and one on a hard-end dipole in cell 25 (low energy dispersion). These systems produce an image of the electron beam and derive the beam emittance in both horizontal and vertical planes, and also the energy spread of the electron beam. The CCD cameras of these 2 systems have been replaced with (IEEE-1394) digital cameras which have numerous advantages with respect to the previously running analog CCD. This has resulted in an improved resolution in the measured emittance.

In addition, the storage ring is equipped with 11 independent devices which measure the vertical emittance in the centre of the first dipole in cells 3, 5, 10, 11, 14, 18, 21, 25, 26, 29 and 31. They are referred to as ‘In-Air-Xray’ (IAX) detectors, a simple, compact and dedicated imaging device which measures the vertical divergence of the bending magnet radiation at high photon energy (170 keV). At such energies the copper absorber is largely transparent and a small part of the initial bending magnet power spectrum is transmitted in the air to the detector and produces an image. These imaging devices use the same digital cameras and device-servers as the pinhole camera system and provide emittance measurements at a rate of 15 Hz. Four new units were added in 2009. Further improvements were added to all ‘IAX’ detectors including an optimum choice of scintillator material and UV imaging optics, and a micro-mechanism that allows in situ focusing.

Fig. 163: Vertical emittance processed by averaging all 15 emittance measurement devices measured over a duration of 100 minutes. The averaging over the signal from all cameras allows the detection of very small emittance fluctuations. Indeed, the source of this small emittance blow-up has been identified and eliminated.

By averaging the emittance processed from the pinhole cameras as well as the the 11 IAX detectors, it is possible to detect weak emittance fluctuations such as those presented in Figure 163 which shows a 0.5 pm emittance blow-up every 12 minutes. Using such averaging at a frequency of 1 Hz, the vertical emittance of 20 pm is measured with a resolution of 0.01 pm.



B.K. Scheidt et al., Proceedings of DIPAC-2005, p 238 (2005).
F. Epaud, Icaleps-2009, Oct. 12-16, Kobe, Japan (2009).
E. Plouviez et al., Proceedings of DIPAC-2005, p 84 (2005).
B.K. Scheidt, Proceedings of DIPAC 2003, p 125 (2005).