XBPM FEEDBACKS

The presented excellent stability within the FOFB loop confines the closed orbit to the reference positions of the involved DBPMs to within less than 1 $\mu$m. Unfortunately the reference of these BPMs is not perfectly static. Separate BPMs (e.g. ``Tune DBPM'') and XBPMs are very well suited to independently judge the resulting orbit stability at the photon source points. The analysis of these data revealed a systematic oscillation of the photon beam with slowly changing periodicity ($\approx$45 min) [7]. This effect was originally suspected to be a temperature effect in the 4-channel DBPM electronics but could be finally traced back to the injection clock cycle, which constantly sweeps over the buckets selected for injection during ``top-up'' operation. A corresponding bunch pattern (intensity) dependence in the RF front-end of the DBPM electronics turned out to be the reason for the orbit oscillations. The implementation of a bunch pattern feedback finally eliminated this effect [8].

Figure 2: Slow XBPM feedbacks provide sub-$\mu$m RMS photon beam stability at the first optical elements of presently three beamlines (exemplified by the data taken at a PX beamline over 85 h of FOFB and ``top-up'' operation).

Nevertheless air temperature variations at the location of the DBPM system electronics in the technical gallery together with temperature fluctuations in the SLS tunnel for example due to a beam loss still lead to a change of the FOFB reference on the $\mu$m level. In order to tackle the problem, presently three slow ($\approx$0.5 Hz bandwidth) feedback loops have been implemented involving XBPMs located at a distance of $\approx$8.6 m from the IDs of two protein crystallography (PX) beamlines featuring in-vacuum undulators and one wiggler based material science (MS) beamline. The feedbacks, which are by default only activated for gaps $<$8.5 mm in order to minimize the photon beam profile dependence of the XBPM readings, translate the photon beam position change to a pure angle variation of the orbit at the source point and change the reference of the DBPMs adjacent to the IDs in the FOFB loop accordingly (cascaded feedback scheme). Fig. 2 depicts the variation of the horizontal and vertical reference of the upstream DBPM together with the corresponding stabilized XBPM readings at one of the PX beamlines over $\approx$85 h of continuous FOFB and ``top-up'' operation. The resulting temporal distributions of the photon beam positions exhibit 2nd moments of $\sigma_x$ = 0.37 $\mu$m and $\sigma_y$ = 0.5 $\mu$m for frequencies $<$0.5 Hz. Fig. 3 visualizes a change of the upstream vertical DBPM reference by $\approx -$15 $\mu$m in case of a beam current drop from 350 to 250 mA ($\approx$0.15 $\mu$m/mA) while the reference XBPM position at the MS beamline was kept constant by the XBPM feedback. This example underlines the necessity to maintain a constant heat load on the involved accelerator and beamline components as they exhibit temperature dependencies (see ``thermal'' in Fig. 3) which together with current dependencies of the DBPM electronics give rise to the observed dramatic reference change. Since the beam current of 350 mA was restored after 4 h the measurement allowed to determine the characteristic time constant $\tau$ = 1.7 h for reestablishing thermal equilibrium.

Figure 3: The upstream vertical DBPM reference (red curve) changed by $\approx -$15 $\mu$m in case of a beam current (magenta curve) drop from 350 to 250 mA ($\approx$0.15 $\mu$m/mA) while the corresponding XBPM position at the MS beamline was kept constant by the XBPM feedback.