User name Seonho Choi
Log entry time 06:16:59 on July25,2001
Entry number 67492
Followups:
keyword=Charge Asymmetry FeedbackDuring this shift, I found several cases where charge asymmetry feedback program has been fooled. Let's look at the example of parity run 1384. On the first plot, we have beam current (top two panels) and charge asymmetry (bottom two panels) with time. I have blown up bottom panels and we don't see overflows, but in this case, whenever there is a beam trip, charge asymmetry becomes very big, on the order of a few thousands ppm. On the second plot, I have histograms of charge asymmetry. On the top two panels, we have the original analysis done by quick_ana.C and the mean value from BCM1 (in this case 240.58ppm) will be used for charge asymmetry feedback. But as we can see, there is high asymmetry tail from beam trips and this tail increases average charge asymmetry artificially. Even if you see the first plot, it is hard to understand why the average charge asymmetry is about 241ppm instead of 100 or 150ppm. So on the two bottom panels of the second plot, I recalculated the average by cutting out this tail. I used asymmetries within +-10000ppm and the average is only 119ppm. Using "incorrect" average of 241ppm, the feedback makes over compensation and the next run, #1385 produces average charge asymmetry -127ppm. (But again this is not completely correct due to high charge asymmetry tail. If I calculate average without the tail, it's about -205ppm.) After several iteration, when equilibrium has been found, the actual charge asymmetry during "real" data may not be zero. When beam trips are very rare, this is not a serious problem. But with current rate of beam trips, there should be an improved algorithm.
Here is a summary of charge asymmetries for the latest few runs.
Run Number Chkfeedback More realistic
1384 241ppm 119ppm
1385 -127ppm -205ppm
1386 -95ppm -220ppm
FIGURE 1
FIGURE 2
