During the period from August 97 to July 98, the major achievement of the Machine Division was the commissioning of the third Radio Frequency (RF) unit. This had to be carried out in parallel with the continuous operation of the machine for users and without disturbing the X-ray beam delivery too much (5600 hours in 1998). The installation of the front-ends and of the insertion devices for the last ESRF beamlines to come into operation before the end of 1998 was also completed. Some of the major topics which were studied during the period are outlined below.


Machine tuning

Looking for smaller emittances

Lattice developments have focused on the implementation of changes in the 4 nm.rad, low ßz optics with a view to providing an increased brilliance. Attempts are currently being pursued in two directions :

Further decreasing the electron beam horizontal emittance down to 3 nm.rad. The strategy is to increase the detuning of the achromat and to readjust the quadrupoles of the straight section triplets in order to keep the ß functions at the source points unchanged as far as possible. The higher focusing resulting from the emittance reduction implies moving the horizontal tune above 37. In parallel to understanding the blow-up of the measured emittance (which is most likely linked to the proximity of a systematic third-order or dodecapolar resonance), the tuning of 32-fold optics with x = 3 nm is being considered.

Achieving a smaller X-ray beam size at the sample by optimising the horizontal ß function in a straight section. The tuning of a straight section to a non-zero x and to a large ßx is equivalent to focusing the beam to a virtual point downstream along the beamline. Hardware modifications of the quadrupole power supplies have been developed in-house in order to provide different currents in the downstream and upstream quadrupoles of the straight section and to obtain the required asymmetry of the optical functions. Preliminary tests of an optics providing ßx = 50 m and x = ­1.1 (as compared to ßx = 35 m and x = 0 for the regular optics) in one of the even straight sections are very promising. The effectiveness of the horizontal focusing was confirmed on the ID6 machine diagnostics beamline with a gain in beam sizes by a factor of 2 and a significant increase in spectral flux per unit surface (see Figure 138). Future efforts will be devoted to assessing the ultimate lack of symmetry of a straight section (ßx = 100 m) and to test the possibility of providing horizontal focusing in several straight sections simultaneously.

In a storage ring, the finite vertical emittance is generated by a transfer of electron motion from the horizontal plane to the vertical plane. This is why it is expressed through the coupling which is the ratio between the vertical and horizontal emittances. Some studies have been undertaken to understand better the different contributions to this coupling so as to reduce it further.

Two X-ray pinhole cameras enable the horizontal and vertical emittances to be permanently monitored during beam delivery. A coupling of 0.8% is routinely achieved during User Service Mode (USM) and at times values in the 0.6% range are recorded, corresponding to a 25.10-12m.rad vertical emittance. A further reduction to 0.3% seems possible.


Beam losses

With the installation of a few narrow gap insertion device vacuum vessels (8 mm vertical aperture), special attention has had to be paid to the possible associated generation of radiation due to scraping effects. Some studies were made to qualify various beamloss monitors, and measurements were performed by the Safety Group using ionisation chambers and neutron detectors, both inside the storage ring tunnel and on the roof. The conclusion is that the radiation at the 10 mm vessels is larger than that at the 15 mm vessels. As a result, collimators are being designed which will be installed on the ring in order to concentrate the electron losses at places where shielding can be reinforced.


A fast AC position feedback system

A fast AC position feedback system was developed during the year. The goal is to reduce the amplitude of the beam oscillations in the vertical plane in the frequency range from 0.1 Hz to 100 Hz. The principle is to make a global correction over the full machine instead of several local corrections on each ID straight section. This system relies on 16 dedicated electron Beam Position Monitors (BPM), located in 8 straight sections and 16 fast steerer magnets. It is controlled by a central Digital Signal Processing (DSP) board communicating with local DSP boards via fast data links operating at 44 kHz.

The commissioning of the system has already started and damping factors of about 2 have been recorded. This system will be put into routine operation during autumn 98.




Operation in 1997/1998

From August 1997 to July 1998, 661 shifts were dedicated to beam delivery to the users. This represents a total of 5290 hours. To this, 162 shifts for the Machine Dedicated Time (MDT) must be added, as well as 27 shifts dedicated to radiation tests for beamlines or tests of the Personal Safety System. The remaining time represents machine shutdowns. Note that the User Service Mode (USM) scheduled hours have been increased from 5359 hours in 1997 to 5664 hours in 1998.


Filling modes

The predominant mode was the 2/3 filling mode with 69% of the shifts (Figure 139). The presence of the 1/3 filling (9% of the shifts) is explained by the fact that this mode was the most convenient one for conditioning the newly installed cavities without penalizing those users requiring a high beam intensity. After three weeks of conditioning with this mode, the machine was switched to the 2/3 filling mode (the second most appropriate mode for the conditioning). At the end of October 1997 the new cavities saw the 16 bunch for the first time. However later on, two major incidents caused by the heating of the RF fingers in 16 bunch mode meant that this mode had to be abandoned temporarily. It was replaced by the 32 bunch mode (also at 90 mA). The main aim being to decrease the intensity per bunch in order to limit the heating of the RF fingers. The hybrid mode was successfully achieved in July 1998 for the first time since the new RF device was installed.

With the voltage modulation which is now possible thanks to the third RF transmitter (SRRF3), various other filling patterns such as groups of bunches are being looked into. The goal is to allow some users to carry out time resolved experiments and at the same time provide high average intensity as required by the majority of the users.

It should also be noted the fact that several pieces of equipment were optimised so as to allow the machine to be refilled up to 200 mA even during the EJP days when the electrical mains power is provided by HQPS. This was not possible previously, due to overly large power requirements.



The best as well as the worst performances were achieved during this period. The best, with the highest figure of availability ever obtained during one run of USM was 98.3%. This was obtained at the end of 1997 (only 0.8% of time being lost due to failures). In addition, a fine record was achieved of 401 hours of USM delivery interrupted only by the MDT, i.e. about three weeks of USM delivery without a single failure (Figure 140). This record is especially impressive since three different modes were delivered (2/3 filling, single bunch and 16 bunch). However, the target value of 95% for the availability was not reached due to a major interruption of 50 hours following the protrusion of a melted RF finger into the path of the beam (see below). The availability over the run concerned with this incident was only 86%. Nevertheless the average beam availability over the period reached 94.5% (Figure 141).

Regarding the Mean Time Between Failures (MTBF), two extremes were also reached during this period (Figure 142): a low value was reached during run 97-04 with 16.5 hours. Such a low figure was to be expected as the two new cavities had to be conditioned with beam. The following run (97-05) still suffered from the new RF system (MTBF of 20 hours) as the cavities also had to be conditioned in single bunch and 16 bunch modes. These two runs made the average of the period drop to a rather moderate 27 hours. However, a very good figure was achieved in run 97-06 with an MTBF close to 65 hours, demonstrating that the new RF system was fully operational.



Failures and preventive maintenance

A major interruption occurred in November 1997 when an RF finger melted into the path of the beam due to the heating in 16 bunch mode.

The statistics of the failures clearly show that 50% of the failures which lasted more than 2 hours occurred during the night or the week-end. Although generally minor, they needed the intervention of an expert to be solved.

The other failures were sometimes due to the weakness of a given piece of equipment (for instance the Linac for which, the switching from 2/3 filling to single bunch lasted more than 3 hours in September 1997) or due to some unpredictable events such as the rupture of a water pipe (twice) leading to a flood in the tunnel.

The preventive maintenance that has been set up has been continuing and is proving to be successful (flushing the water circuitry with nitrogen or acid, regularly checking all the front-end equipment). Furthermore, and following the events of this period, new subjects for preventive maintenance have been determined. First, many radio gammagraphy campaigns have been carried out during the last 6 months to identify the damaged RF fingers. Several were found and replaced during the shutdowns. Following one of the two floods which occurred, it was noted that some flexible pipes installed in the tunnels do not stand up to the radiation. They are going to be changed with another kind of material which has proved to be effective at other institutes.


RF fingers

As noted above, some problems were encountered during the delivery in 16 bunch due to damaged RF fingers. The role of the RF fingers is to shield the bellows in order to provide low impedance for the RF field excited by the beam (image current circulating along the vacuum vessel). The RF power deposited by the beam scales as I2 / Nbunch, where I is the total beam current and Nbunch the number of bunches in the machine. It also depends on the local vacuum vessel impedance and on the effective bunch length. If the contact is not good, the impedance of the bellows assembly is large resulting in high power dissipation.

The RF fingers were damaged either due to incorrect positioning during installation (Cell 4), or to mechanical constraints during the bakeout at 400° C of the long vacuum vessels in the straight sections (Cells 21, 22, 26). Once damaged, the RF power deposited by the beam in 16 bunch mode further degrades the RF fingers. This is accentuated by the fact that the machine is now being operated with a larger RF voltage (and therefore shorter bunches) since the installation of the third RF system.


Following these unfortunate events, it was decided to :

- Operate in 32 bunch mode until the summer shutdown (in place of the scheduled 16 bunch) to have time to replace the defective RF fingers around the ring.

- Operate with a lower RF voltage

- Replace most of the damaged RF fingers and reduce the bakeout temperature

- Improve the bakeout process.



The reasons why the performances of the machine were slightly below the performances of 1996 have been clearly identified: one major incident had some impact on the average availability and the installation of the third RF system led to a moderate MTBF. However, the MTBF of the RF equipment is increasing run by run. In addition, several groups are paying more attention to preventive maintenance on their equipment. These elements give confidence that it will be possible to reach an availability greater than 95% and a MTBF greater than 40 hours in the forthcoming period. The availability during the first half of 1998 is already at 96%.




Commissioning of the third radio frequency acceleration unit for the storage ring

In early 1995 it was decided to construct a third RF transmitter (SRRF3) feeding a third pair of cavities. Besides the reduced power load for the cavity couplers and a gain in reliability and redundancy, this third RF unit opens up other interesting new features.

Based on experience gained from the existing RF units, the third RF unit was constructed using similar hardware for the large components. This included the klystrons, the high voltage power supply (HVPS), the circulator, parts of the low level RF and the 2 five-cell cavities. Some sensitive parts of the RF system, such as the arc detection system, were redesigned in order to reduce the rate of spurious triggering. Also the cavity vacuum equipment was improved with a new fast pressure interlock system, and the NEG pumps of the old design were replaced by titanium sublimation pumps. This proved to be very effective during RF conditioning and commissioning with beam, as it was possible to activate these pumps regularly, so that the vacuum could recuperate quickly after strong outgassing.

As the control system of the existing transmitters was becoming outdated and difficult to maintain or modify, a new control system was developed for SRRF3 and commissioned in parallel with the hardware: This control system, based on an object oriented programming approach, features hardware protection managed independently from the control software by a Programmable Logic Controller (PLC) for the slow interlocks, a hardwired system for fast protection, very powerful diagnostics tools and an easy to use graphical user interface.

The first pieces of equipment were delivered in November 1996 and the commissioning of the system started in August 1997. As noted above it took some time (some hundreds of Ampere.hours) to fully condition the two new cavities for all beam delivery modes. During the three first months of operation, many trips were caused by bursts of reflected power from the cavities, which were typically 50 µs long and as high as 300 kW at the peak. These events were mostly associated with outgassing and were probably due to fast detuning provoked by multipactor in one of the five cells. Continuous operation of the cavities is the only cure for these trips which steadily become less and less frequent.

Some of the early failures were removed gradually during the first three months simply by debugging the system.

The HVPS proved to be very reliable. For example, not a single spurious crowbar firing has occurred since the start-up.



New possibilities with SRRF3


More flexibility and larger safety margin

The fact that the maximum ESRF beam current of 205 mA can be stored with two out of the three RF transmitters represents a large gain in flexibility and security. Klystron acceptance tests, cavity coupler conditioning and extensive R&D work requiring RF power can now be performed without interfering with beam delivery to ESRF users.

It is planned to extend the waveguide network to obtain a level of redundancy that will allow storage ring (SR) operation to be safeguarded even if there is a major failure on any single transmitter.


Increase in cavity voltage

The third pair of cavities has allowed the cavity voltage to be raised from 8 to 12 MV with only a negligible increase in total power and still a reduction of the power per cavity coupler. However, it has turned out that the associated gain in longitudinal acceptance has not shown up on the lifetime which is presently limited by the transverse acceptance. Increasing the transverse acceptance of the SR is therefore one topic of machine physics studies to benefit fully from the enlarged longitudinal acceptance.

Since the lifetime is Touschek limited in both single bunch and 32 bunch operation, the optimum voltage is close to 8 MV, for which the bunches are longest and the energy acceptance is not yet limited by the RF.


Landau damping of multibunch instabilities with cavities 5 & 6

For standard high intensity operation, only 2/3 of the SR is filled to produce strong transient beam loading, where the gap in the bunch train induces the necessary modulation of the accelerating voltage at the revolution frequency f0. The subsequent spread in synchrotron frequencies provides Landau damping of longitudinal multibunch instabilities which are driven by Higher Order Modes (HOM) developing in the cavities; a point of major importance for the ESRF. An additional feature of the third RF system is the possibility to operate it at fRF + f0, such as to modulate the total RF voltage actively even for symmetrical fillings.

This scheme is employed for operation with 32 equally spaced bunches in the SR, where the optimum voltage of about 8 MV requires switching off cavities 5 & 6. As these are no longer temperature regulated, modulation is necessary to operate untroubled by multibunch instabilities at the nominal intensity of 90 mA.



A third RF acceleration unit has been successfully built and put into operation on the ESRF storage ring. It was designed entirely by ESRF staff in order to provide and ideal match to the operation constraints and specificities, and to fit into the ESRF environment, taking into account the experience gained on the existing RF units. Performing such design and construction work has been very beneficial in broadening the knowledge and expertise of the RF group personnel. The project was completed within its budget and on schedule, and was managed so as to minimise inconvenience to ESRF users. This was a challenging issue for a machine which delivers 5600 hours of X-ray beam per year.


Insertion devices

Insertion device vacuum vessels

The first 5 m long 10 mm high (8 mm aperture) vacuum vessel made out of copper plated stainless steel was installed in the machine (on the ID27 radiation safety test beamline) during the 1997 summer shutdown. The conditioning went smoothly, and at the end of run 97-4 no Bremsstrahlung due to scraping of the electron beam was detected. This vessel was then moved to ID28, and two others were installed during the spring 1998 (in cell 23 and 31). One of these two will be moved to the ID9 beamline in spring 1999.

Besides the installation of new segments of conventional undulators, the Insertion Devices (ID) group has been working on some new developments. These include the test of a new scheme of quasi-periodic undulator, the manufacture of a fast switching helical undulator, the manufacture of a 3 Tesla asymmetric wiggler made of permanent magnets and the tests of compound refractive lenses.


Quasi-periodic undulator

In a conventional undulator, the spectrum consists of a series of peaks, the energies of which are precisely an interger multiple of the energy of the fundamental. Because the crystals also diffract the harmonics, the result is that the monochromatic beam produced by the monochromators is always polluted to some extent with higher harmonics. To avoid this difficulty, one may use a mirror to cut off the high energies or use a quasi-periodic undulator as proposed by S. Hashimoto and S. Sasaki. A new magnetic structure of quasi-periodic undulator has been proposed by the ID group which is more compact and gives a higher flux per unit length of undulator (Figure 143).

A prototype of such an undulator has been built and successfully tested on the ID6 beamline showing a suppression of the spectral flux on harmonics 3 (5) of 8.3 (11.3) compared to a conventional undulator. Such an undulator is presently under construction for the MEDEA beamline (ID27).


Fast switching helical undulator

Three linear/helical undulators made of permanent magnets are in operation on ID12 and ID16. The flipping between the left and right polarisations must be carried out by longitudinally translating one magnet array. This cannot be done in less than a few seconds. As a result it is difficult to measure a dichroism signal lower than 10-3. To overcome this difficulty, a novel type of linear/helical undulator has been designed and built by the ID group. It combines an array of permanent magnets which produces a horizontal field and a coil and yoke assembly which produces the vertical field (Figure 144).

The flipping of the circular polarisation is achieved by inverting the current in the coil which is made in a few tens of milliseconds. This undulator is now in user operation and a natural dichroism as low as 10-4 has been recorded.


3 Tesla permanent magnet wiggler

At present all permanent magnet wigglers in operation in the world have a peak field lower than 2 Tesla. To reach a higher field, one must use superconducting technology. Superconducting wigglers are much more expensive and delicate to operate than permanent magnet wigglers. In order to extend the range of peak fields that can be reached with permanent magnet wigglers, a prototype wiggler has been designed and built by the ID group. It is made of two periods of 370 mm and the measured peak field is 3.1 Tesla for a magnetic gap of 11 mm. The profile of the magnetic field is shown in Figure 145. The field is asymmetric in order to produce high energy circularly polarised photons.


Compound refractive lenses

Following the pioneering work of A. Snigirev et al., a detailed theoretical and experimental investigation of compound refractive lenses was carried out on the ID6 beamline. A number of different low Z materials have been tested. The best resolution and transmission were obtained with hole drilled Pyrocarbon and Beryllium materials. In the ID6 beamline, a vertically focusing lens gave a gain of 15 and a transmission of 80% at 24 keV in a 3:1 imaging geometry. In a two plane focusing geometry, a gain of 23 was reached at the same energy and for the same geometry.

In the hard X-ray range, these lenses provide a number of advantages over the conventional X-ray focusing devices (mirrors, bent crystal, Fresnel zone plates). They are inexpensive and highly resistant to the heatload. Their drawback is the limited aperture and the chromatic aberration. For a focal length of 10 m the aperture reaches a maximum of 1.6 mm (fwhm) at around 12 keV. This fits the narrow beam of the undulator beamlines well, but causes a large attenuation for bending magnets or wiggler radiation.

Several undulator beamlines (ID6, ID10, ID13, ID18, ID22) have already been equipped with these lenses placed in the vacuum of the front-end in the white beam of the undulator. They are placed on a remotely controlled X-Z linear stage which allows the selection and alignment of the lens. The user can select one of several lenses which best fits the photon energy and can move the lens to align the image spot to the sample. A few other prototype have been built for ID12 (1-5 keV) , D5 (50 keV) and ID20 (8-20 keV).


Software developments

Significant effort has been devoted to developing the new software RADIA and SRW which are both available from the ESRF web site. RADIA is a full featured 3D magnetostatics code which allows the precise design of insertion devices and other accelerator magnets (dipole, quadrupole). It is based on an integral approach and outperforms the commercial finite element packages in many areas. The software SRW (Synchrotron Radiation Workshop), is presently under development. The currently installed modules allow the accurate computation and propagation of synchrotron radiation in the near field approximation and the fast computation of undulator and wiggler spectra through an aperture including electron beam emittance and energy spread.



By July 1998, there were 37 operational front-ends delivering beam to the users: 24 on beamlines taking beam from insertion devices and 13 taking beam from bending magnets.

With the increasing number of undulator segments on one straight section and the reduction of the minimum gap, the design of the beryllium window assembly, which isolates the storage ring vacuum from the beamline vacuum, has had to be modified in order to sustain the increasing heat load (high power filters and new beryllium window). Nine ID front-ends are now equipped with such a configuration which allows them to close 2 undulator segments (1.6 m long) at a 16 mm gap with a 200 mA current in the storage ring.

A new high power X-ray beam movable absorber has just been designed and will replace the existing one on the most demanding beamlines where three undulators will be used at a 11 mm gap. This absorber will be tested on the ID23 test front-end early in 1999.

Diamond windows have been developed in order to improve the transmission of coherency in the X-ray beam, and also to be compatible with the high power density of the beamlines concerned. A first prototype diamond window was tested very successfully during the run 98-3 on the ID23 test front-end and could sustain the high power of X-rays produced by 3 insertion devices segments at 200 mA.