Machine tuning


Larger brilliance

The remarkable flexibility of the Double Bend Achromat lattice used on the storage ring has been further illustrated by the implementation of low z optics since the last run of 1996. These new optics are based on the reduction of the vertical in the high beta straight sections from 13 m down to 2.5 m whilst keeping the other features of the previous 4 nm optics (distributed dispersion, same beam sizes in the low beta straight sections, ...).

This reduction of vertical betas was a prerequisite for the installation of vacuum vessels of reduced height in the straight sections which will enable the insertion device gaps to be closed, down from the present 16 mm to 11 mm. Thanks to the smaller electron beam sizes in the straight sections, beam scraping on the reduced beam stay-clear insertion device vessels (8 mm inner aperture) is less severe and consequently the beam gas scattering lifetime reduction and bremsstrahlung emission on the beamlines are minimised. Two 2 meter long, 10 mm high vacuum vessels were installed in the ring at the beginning of 1997 without any significant impact on the lifetime. In order to minimise the trend towards the increase of the resistive wall impedance and of the vertical chromaticity providing the necessary damping, the new generation of 10 mm vessels is made of copper-plated stainless steel.

The low z optics also enabled a further gain in brilliance by a factor of 2 in the high straight sections by means of a better matching of the photon beam emittance to the electron beam emittance. This, together with a routinely-achieved coupling figure of 0.7%, leads to a brilliance in the 1020 range, i.e. a gain by a factor of 100 with respect to the Foundation Phase Report goals. The slight drawback to these optics (increase in the vertical divergence of the photon beam) was found to be negligible, and in certain cases beneficial effects were unexpectedly observed on some beamlines.

New upgrades (reduction of the coupling to 0.3%, new optics providing a 3 nm horizontal emittance) are anticipated for the end of 1997. This would push the ESRF brilliance towards the 1021 range which is the maximum achievable figure with the present generation of storage rings (Table 1).

Focusing the X-ray beam downstream along the beamlines by operating the lattice with a non-symmetrical electron beam horizontal envelope in the high straight sections could significantly increase the photon flux on the sample. This could be equivalent to a gain by a factor of 10 in brilliance. The hardware necessary for asymmetrical powering of the quadrupoles of the straight section triplets is being procured and the scheme will be tested in one straight section during the second half of 1997.

Besides probing some of the issues of future diffraction limited rings by running the ESRF at low energy, a reliable operation at 5 or 4 GeV might be of interest for some users. A record vertical emittance of 16 pm (corresponding to 1% coupling) has recently been measured at 4 GeV. Lowering the current limitation induced by the excitation of Higher Order Modes (HOM) in the cavities to a value comparable to that of the 6 GeV operation should be made easier by the impending installation of the third RF acceleration unit and the new possibilities of Landau damping longitudinal instabilities.


Longer lifetime

Like low-energy machines, the ESRF is Touschek lifetime limited in all operating modes because of the small transverse dimensions (the Touschek effect linked to the collisions between electrons is proportional to the electron bunch density). This was illustrated by the significant gain in lifetime made when switching from the 1/3-filling mode to the 2/3-filling mode (with a gap in the bunch train extending over one third of the circumference) in September 1996. The associated increase in lifetime from 40 hours to more than 50 hours at 200 mA is due to the smaller current per bunch which reduces the Touschek contribution. When lengthening the bunch train, the machine becomes more sensitive to HOMs developing in the RF cavities but this has been overcome by careful temperature control of these cavities

In the few bunch mode of operation, a large acceptance in energy is therefore of prime importance for overcoming Touschek limitations and improving the rather moderate lifetime (8 hours at 90 mA in the 16-bunch mode). Investigations are currently being pursued to enlarge the energy acceptance. The transverse-related limitation induced by the injection septum was reduced by moving the septum away from the stored beam (from 14 mm to 19 mm) at the beginning of 1997. This resulted in a significant gain on the lifetime in 16 bunch mode (from 8 hours to 11 hours). A further enlargement of the present ± 3% energy acceptance is aimed at in the course of 1997.


Beam losses

After installing the first 15 mm high vacuum vessels (11 mm aperture) in high beta straight sections, we were confronted with two problems: an unusually high bremsstrahlung level on the outside of the optics hutches and a high activation of the insertion device chamber vessels. A systematic study was undertaken and beam-loss monitors were installed all around the storage ring.

Measurements on ID27 revealed that the bremsstrahlung was created by electrons lost on the chamber wall. Two mechanisms creating these losses were identified and confirmed by measurements with beam-loss detectors:
- vertical elastic scattering of electrons on the residual gas resulting in overly high vertical betatron oscillations
- Touschek scattering between the electrons leading to too great an energy loss of the electrons. These energy reduced electrons were finally lost in the vertical plane as well, probably due to a tune shift onto a vertical resonance.

A major contribution to the activation of the insertion device chambers was found to come from losses during injection. In particular, vertical mis-steering of the injected beam due to closed orbit drifts in the booster could create potential electron losses in the vertical plane on the insertion device vessels.

This was confirmed by measurements with the beam loss detectors and the fact that the highest activation levels were measured in the first straight sections downstream from the injection.

Several measures were taken to reduce the losses on the chambers:
- The switching to the new low bz optics resulted in an increase of the effective vertical aperture, thereby reducing the electron losses on the insertion device chambers during the decay. Additionally the losses in the vertical plane of the electrons with reduced energy disappeared.
- Changing the filling pattern from 1/3-filling to 2/3-filling reduced the Touschek losses by a factor of two.
- Injection and beam decay were performed with the upper and lower jaw of the ID6 scraper closed to a value equivalent to a 30% reduction of the effective vertical aperture. This immediately suppressed the losses on the insertion device chambers due to injection and vertical resonance losses. This was measured with the beamloss detectors and confirmed by the fact that from that time on, the activation on all highly-activated insertion device vessels started to decrease.

It was found by simulation and measurement that incident scattering on a scraper jaw leads to significant secondary losses on downstream vessels. This is why a second vertical scraper was installed in cell 8 (the mechanical assembly of a mini-gap undulator without magnets is used), and a third one in cell 5. Based on these scraper positions, the closure and actions were defined for various situations such as injection, cleaning and beam kill. All these efforts allowed 8 mm aperture vessels to be installed without any major problems so far.



To improve the transverse energy acceptance of the storage ring, the injection septum S3 was moved further away from the stored beam axis. Following this change, the injection had to be re-tuned. The main consequence is that the kicker excitation is now stronger and that the machine is always operated with a closed kicker bump. The same injection efficiency (> 80%) could be achieved. This scheme also allows single-bunch injection without injection saturation up to the stability threshold of the single bunch. Additional alternative solutions were developed in case of a kicker failure. We proved we could accumulate when using only the more downstream of the four kickers. It is also possible to accumulate if only kicker one and three are available.


Longitudinal studies

Longitudinal coupled bunch instabilities (LCBIs) are a major obstacle for the increase of beam current in a storage ring. At the ESRF, threshold limits for multibunch operation have been considerably increased by using fractional fillings, from about 60 mA for a homogeneous filling, to well beyond the nominal intensity of 200 mA for a filling of one third of the circumference. The gap in the bunch train induces a modulation of the cavity voltage and a subsequent spread in synchrotron frequencies. This results in additional Landau damping.

Thesis work was carried out to better understand this phenomenon. A theory of Landau damping of LCBIs, incorporating natural damping, was elaborated and the spread in synchrotron frequencies from beam loading due to fractional fillings was computed. This was verified by experiments and the results permitted the deliberate choice of the appropriate filling ratio (2/3-filling). However, to allow high intensity operation in uniform filling at 6 GeV, a direct modulation produced by the extra pair of radio-frequency cavities to be installed in summer 1997, is envisaged in the near future.




Operation in 1996/1997


From July 1996 to the end of July 1997, 688 shifts were dedicated to beam delivery to the users. This represents a total of 5504 hours (in addition to which 1328 were dedicated to machine studies).

The number of User Service Modes (USM) shifts in 1997 (670 shifts) has not changed significantly compared to 1996 (650 shifts). The next increase will take place in 1998 with 700 shifts (5600 hours). The ultimate goal is 6000 hours per year which should be achieved before the year 2000.


Filling modes

1996 was an innovative year for modes since two new ones were delivered to the users. During run 5 of 1996, the 1/3-filling mode was withdrawn to deliver the new 2/3-filling mode providing a longer lifetime (50 hours at 200 mA). To satisfy a larger number of users and to find a better compromise between the people who need high intensity and the people who need the single bunches for time-resolution experiments, a new hybrid mode was developed: the hybrid 4 mode which was introduced by the end of 1996. It consists of a 1/3-filling bunch train faced with four equally-spaced isolated bunches. The lifetime in this mode reaches 30 hours.

In 1996 the major interest was for the 1/3 and 2/3-filling modes which represent almost 50% of the shifts. Nevertheless, interest in the 16-bunch mode remains high since more than 25% of the shifts were dedicated to this mode. The remaining shifts were distributed between the hybrid modes (1 and 4) and the single-bunch mode. About the same percentages have been maintained for the first half of 1997 although the achieved beam position stability is not as good when the machine is operated in a hybrid mode, in comparison to the 2/3-filling mode. This is mainly during the first hour after the refill due to both the longer refill time and the associated larger variation of the heat-load. This may justify scheduling these modes less often in the future (Figure 143).



The availability over the period considered for this report again reached a high score with an average availability of the X-ray beam of 94.4%: the 'lowest' figure was 93.2% during run 97-02, and the highest was 96.6%, obtained during run 97-03. The remaining 5.6% is shared between the time for the refills (2.2%) and the failures (3.4%) (Figure 144). The fraction of time necessary for the refills can be as low as 1% when the beam is delivered in 2/3-filling, or as high as 4% when delivering the exotic modes which require a cleaning process and/or more frequent refills due to the reduced lifetime (hybrid modes, 16-bunch or single-bunch mode). Towards the middle of 1997, a new selective cleaning method in hybrid mode was successfully implemented for the operation of the hybrid mode. This allows a topping-up in the hybrid 1 mode, thus reducing instabilities due to thermal effects and moreover decreasing the refill time to less than 15 minutes (instead of 25 minutes previously).

The best week ever experienced at the ESRF occurred towards the end of 1996 when the beam was delivered over a full week for 168 hours without a single hitch (availability was above 99%!) (Figure 145).

The Mean Time Between Failures (an essential criterion for the users) also remained at a high level: during the period under consideration, an average of 42 hours was reached. The highest score was 68 hours (run 97-03) but it only reached 29 hours during the run 97-02 which is explained below (Figure 146).

Three years of records indicate that the improvement of the statistics of availability mainly originates from the continuous preventive maintenance on the water circuits, on the front-end equipment, from several improvements brought to the radio-frequency system and from the benefits of the High Quality Power Supply (HQPS). The HQPS has been operating for almost two years now and the conclusions are very positive. During 1996, more than 220 drops recorded on the input electrical mains were smoothed out, thus avoiding beam trips with all their possible direct and indirect consequences. It is clear that the very significant improvement in Mean Time Between Failures described above can be, for a large part, attributed to the untroubled primary power delivered by the HQPS. At the same time, this has enabled intrinsic faults in equipment to be clearly distinguished from actual outside perturbations. Moreover, the absolute low harmonic pollution screened out by the HQPS on site means that all power converters will have increased lifetimes.



The most significant incident occurred in April 1997 during a test on the machine when the electron beam was mis-steered and the X-ray beam drilled a hole in a vacuum vessel (ID27). Two full cells and three front-end modules 1 were put to atmospheric pressure. The vacuum vessel needed to be replaced. The highly efficient reaction of the people involved in this repair limited the time lost in user service mode to 22 hours. The other major beam interruption (5.5 hours) occurred in August 1996 and was due to a burnt transformer in the dipole power supply.


Super Spare Power Supply

Although power supply failures are rare, they can last a long time (replacement of a transformer or another major repair). To avoid loss of time in these situations, a Super Spare Power Supply has recently been built. Its role is to take over from any magnet power supply which fails. Furthermore, a switching board has been installed recently: this allows any magnet family to be connected to the Super Spare Power Supply within less than an hour.



Our objective for the forthcoming period is an availability of 95% whilst keeping a high score for the Mean Time Between Failures (> 40 hours). To do so, we will increase the preventive maintenance work, try to reduce the duration of the refills in hybrid modes and, with the collaboration of the Experiments Division, optimise the mode scheduling in order to minimise mode switching.




Status of the construction of the third radio-frequency acceleration unit for the storage ring

In 1995 it was decided to equip the storage ring with an additional pair of cavities fed by a third transmitter, and the main reasons (essentially an increase in reliability) were reported in the 95/96 Highlights. In the last twelve months, intensive work has been done to enable the installation of the two new cavities during the summer of 1997, and to have an operational transmitter to power them. The new SRRF3 building was officially handed over in September 1996. The klystron garage was delivered and installed in November 1996. The contract with Siemens for the High Voltage Power Supply went very smoothly and the site acceptance tests were successfully carried out in February. Part of the auxiliaries and low-level RF components of the transmitter were made available for HVPS tests in January. The RF low-level minimum configuration for cavity conditioning was ready in June, and the transmitter will be fully commissioned for beam operation by October.

The two RF cavities were delivered with some delay, but without severe repercussions on the time schedule. The RF preconditioning of the first cavity (on the test stand) was completed at the beginning of June and the second cavity by mid-July, i.e. just before installation inside the tunnel, which took place during the 4-week summer shutdown.

RF conditioning with and without beam started on 15 August, and 6 days later we were able to successfully deliver the beam to the users on the scheduled date at a slightly reduced intensity (160 mA) but much above what had been promised (100 mA). Obviously the conditioning of the cavities was not completed in such a short time and a lot of beam trips due to strong outgassing were encountered, but the situation improved steadily (75% of the machine dedicated time was used for this purpose during the last run) and will continue to improve with more and more ampere.hours accumulated. The new cavities will have to be also conditioned in single-bunch and 16-bunch modes. One very positive point was the performance of the RF power plant which was commissioned in parallel to the cavity conditioning and which operated smoothly without a single hitch during the entire run 97-04.

We can say today that this project, which was fully designed and managed by ESRF people, involving staff from the Machine Division but also from the Computing and Technical Services, and others (Safety Group) has been completed very successfully, thanks to the high professionalism and great motivation of all.




Insertion devices

By the middle of 1997, 50 insertion device segments were in operation in the storage ring tunnel serving 26 beamlines. This corresponds to a cumulated length of 75 m. Except for a 4 Tesla superconducting wiggler, all insertion device segments are permanent magnet undulators (75%) or wigglers (25%), each 1.6 m long with magnet blocks located in the air on both sides of the vacuum chamber. Each segment was designed and optimised in collaboration with the beamline responsibles. As a result, we have made a large number of magnetic designs. This is illustrated in Figure 147 which shows the distribution of insertion devices according to their periods. All periods between 23 mm and 52 mm are undulators with peak fields from 0.2 to 0.7 T, while almost all devices with periods ranging from 70 to 230 mm are multipole wigglers with peak fields ranging from 0.8 to 1.9 T. Circularly polarised radiation is produced by three linear/helical undulators (0.5 to 10 keV range) and three asymmetric wigglers (20 to 400 keV).

All undulators are manufactured at the ESRF with multipole and spectrum shimming thereby minimising the closed orbit distortion at all gap values and maximising the brilliance on all harmonics of the spectrum. Figure 148 shows the rms phase error achieved on the last thirteen undulator segments produced at the ESRF. Phase errors typically result in a reduced brilliance on the high harmonics of the spectrum, which are the most sensitive. With such low phase errors (< 2 deg), and assuming the worst case of a filament and mono-energetic electron beam, the brilliance reaches 96% (88%) of that of an ideal undulator on harmonic 9 (15). Taking into account the finite electron emittances and energy spread, close to 100% of the expected brilliance is obtained on all harmonics from 1 to 15.

All undulators built in the last two years are equipped with simple phasing sections [see the 95/96 Highlights]. With such phasing sections, a user can vary the length of the undulator by multiples of 1.65 m without inducing any closed orbit distortion and maintaining proper interference between the undulator segments for any gap.

Figure 149 shows the brilliance routinely achieved in spring 1997. The brilliance on the ID3 beamline (equipped with two undulator segments) reaches nearly 1.1020 photons/sec/0.1%/mm2/mrad2 at around 5 keV. A typical undulator beamline equipped with a single, fully tuneable segment of 42 mm period installed on a high beta straight section produces a brilliance higher than 1.1019 photons/s/mm2/mrad2/0.1%Bwidth for any photon energy between 2 and 20 keV.

The design and construction of a fast-switching variable polarisation undulator, a 3T multipole asymmetric wiggler and an in-vacuum undulator are currently in progress.


Insertion device vacuum vessels

The first 2 m long 10 mm high (8 mm aperture) vacuum vessel prototype (made out of pure stainless steel) was installed on the machine (on the ID27 radiation safety test beamline) during the winter shutdown. The conditioning went smoothly, even if it required more amp.hours than expected. In any case no bremsstrahlung due to scraping of the electron beam was detected.

In March 97, a similar 2 m long 10 mm high (8 mm aperture) insertion device vacuum vessel (made out of pure stainless steel) was installed on the ID11 straight section together with a 34 mm period undulator segment, and a 2 m long 10 mm high (8 mm aperture) vacuum vessel prototype made out of copper-plated stainless steel was placed on ID27. The conditioning on ID11 progressed rapidly and after twoweeks of operation the beamline was authorised to take beam. The situation was even better for the copper-plated vacuum vessel, which thanks to efficient preparation work, always had pressure one decade below that of the pure stainless steel vacuum vessel.

The first 5m long 10mm high (8 mm aperture) vacuum vessel made out of copper plated stainless steel was installed on the machine (on the ID27 radiation safety test beamline) during the last summer shutdown. The conditioning went smoothly, and at the end of run 97-04 no bremsstrahlung due to scraping of the electron beam was detected.

Three other such vessels will be delivered by the end of 1997 and will be installed either during the next winter shutdown or in March 98. Since the conditioning requires time (typically two to three weeks) and in order not to prevent a beamline from taking beam during the corresponding period, all new 5 m long vacuum vessels will be systematically installed on a free straight section of the machine during one full run prior to their installation on a beamline straight section.





In July 1997, there were 33 operational front-ends delivering beam to users: 22 on beamlines taking beam from insertion devices, and 11 taking beam from bending magnets (see Table 2).

With the increasing number of undulator segments on one straight section and the reduction of the minimum possible gap, the design of the beryllium window assembly, which isolates the storage vacuum from the beamline vacuum, had to be modified in order to sustain the increasing heat load. New filters were successfully tested on ID27 during 1996 and there are now five insertion device front-ends in operation equipped with a high power filter configuration, allowing them to close two undulator carriages at 16 mm with 200 mA current in the storage ring. The behaviour of these front-ends is carefully monitored since any malfunction can result in severe damage to different parts of equipment (absorbers, filters, vacuum gauges), because of such very high power densities.

Some developments were also carried out either following the requests from some beamline responsibles or in order to better protect the front-end equipment.

High power X-ray beam absorber: This new movable absorber has just been designed and will replace the existing one for 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 before the end of 1997.

High power windows: We have developed diamond windows that will replace some carbon filter/Be window configurations in order to improve the transmission of coherence in the X-ray beam, and which also will be compatible with the power density of the concerned beamlines. Diamond windows are also being considered for the transmission of very high power densities (three undulator configurations) and for specific beamlines using wigglers which need wider horizontal fans. The first window will be tested on ID23 in August 1997 and others will follow before the end of the year.

Due to the increasing power and power density which have to be transmitted, we have tried to optimise the front-end configuration but the heat load is not the only aspect. We also need to improve the functionality of the front-end in order to better protect all our equipment and also to maintain a high level of reliability.