09/10/2019: Dedicated Discussion about BSW
Present: D. Aguglia, S. Albright, F. Asvesta, H. Bartosik, C. Bracco, J.P. Burnet, G.P. Di Giovanni, J.L. Gomez Costa, B. Mikulec, R. Murillo, E. Renner, T. Todorcevic
Minutes
Some highlight below
-
A review the original specifications
-
Every magnet (BSW1,2,3,4) is a separate circuit
-
Initial control based on RST in a standard approach
-
Looking at the flat top (FT), the difference between the measurement and the reference (slide 4 green vs red curve) is too large to match the specifications
-
This would force EPC to manually tune the converters continuously, which is not possible in operations.
-
Not acceptable > so an Iterative Learning Control (ILC) to modify a few PPM was tested. The ILC works as a reference adaptation method
-
The ILC will cause a deformation of the target control setting to actually reach the reference:
-
It would important to log the current reference, the curve after the learning reference and the measured one. To discuss in the future how to do this.
-
Limitation can be put in the algorithm to limit the ILC action and avoid wild excursions.
-
Warning and error coming from LASER should be propagated in case the converters goes outside the reference by a certain window.
-
BSW can be non-PPM in shape & in amplitude, but it must be PPM in time. KSW and BSW have to decay together. In the past a mis-synchronisation of 150 us between KSW and BSW was investigated and was considered acceptable, but only under the assumption that the BSW is decaying after the KSW and not while the KSW is pulsing at the maximum current.
-
ILC could be used to correct for temperature drift
-
Precision requirement revisited:
-
Achieved precision for BSW1 is in range of +/-150ppm (>97% in +/-125ppm spec.) for the 2nd half of the flat-top; and +/-240ppm over the entire current pulse
-
Achieved precision for BSW4 is in range of +/-200ppm (>93% in +/-125ppm spec.) for the 2nd half of the flat-top; and +/-420ppm over the 2nd half of the current pulseBSW1 worse case on flat-top is +/-125 ppm. The rest is in specs +/- 150 ppm
-
EPC will work on getting better repeatability and use the ILC to correct drifts in temperature.
-
16 power converters to be calibrated, set, tested during HW commissioning.
Question to answer: is there any problem with the spec of 250 ppm, instead of the 100 ppm?
-
Realistic assumption done for the simulation as for a post-LS2 restart
-
Only quads error, fringe fields and the BSW new specifications added as an error.
-
New WP (4.4, 4.45)
-
KSW ratio optimised for the new WP. This WP does not close the orbit bump
-
37 turns injected for HL-LHC beams
-
Worse case scenario. Error of BSW1 in one direction and BSW4 in another direction for all the 37 turns. No error in the decay. It is not needed. The beam shape is defined after the completion of the injection. The initial emittance is determined by the error of the injected beam wrt the circulating beam.
-
Study can be expanded with randomised error and include BSW error on injected beam. EPC will provide the measurements input to ABT for more realistic simulations.
-
NO significance emittance increase due to BSW shot-to-shot reproducibility of +/- 125 ppm.
-
Hesitant to go beyond +/- 125 ppm to keep margin to include other errors and avoid getting closer and closer to the target of 1.7 um.
Conclusions: EPC will keep working on improving the system for reproducibility and correct other source of error. ABT will update the simulation including other errors (e.g. mis-steering) and randomized errors from the BSW. Another meeting will be organised at the end of the year.
Some further points discussed:
Agreement to update the functional specs once the performance are established.
J-P. Burnet asked who measured the fringe fields. The fringe field have been measured in the ABT lab.
B. Mikulec added that the new PSB injection has been simplified and it is planned to inject on a flat-bottom. The option to inject on a ramp should be retained for future optmisations and squeeze the last % of performance. R. Murillo said that this approach should remove the complexity of the Bdl control. At least at the beginning.
C. Bracco said that the BSW amplitude will be kept constant during the injection as part of the requested simplification of the injection process.
G.P. Di Giovanni said that there is the request to have "asynchronous" injection for initial optics studies, e.g. inject a few turns and maintain the BSW pulsing for the maximum amount of time, for instance 150 us.
-> R. Murillo agreed to organise a meeting with OP to discuss the control of the BSW.