BCM1F calibration studies
Bunch profiles
Fits for CH11, CH14, CH21, CH24
Gaussian fits for bunch 43,
here. Results are summarized below
This gives values of Δt
11,21 = 12.2 ns and Δt
14,24 = 13.6 ns.
Old results
A couple of features are worth noting before the fits are presented. The relation between TDCs and bunch number has changed (n
bunch = (t - 58)/24.9505 + 1). Also, the offset for individual channels may be different than before so that the calibration factors may need to be recalculated. Here I only consider the time range from 0900-1600.
The fits for colliding bunches are
here. Summarized in table,
Channel |
mean 1 (ns) |
σ1 (ns) |
mean 2 (ns) |
σ2 (ns) |
χ2/NDF |
1/1 |
1141.9 |
0.6 |
1143.1 |
2.3 |
19.9 |
1/2 |
1149 |
1.7 |
1154.2 |
2.2 |
6.5 |
1/3 |
1148.2 |
2.4 |
1148.8 |
1.04 |
3.9 |
1/4 |
1145.7 |
1 |
1150.1 |
2.6 |
8.4 |
2/1 |
1140.1 |
1.35 |
1144.5 |
2.1 |
12.14 |
2/2 |
1153.9 |
3.1 |
1154.4 |
1.1 |
4.8 |
2/3 |
1141.9 |
1.5 |
1147.7 |
2.6 |
3.6 |
2/4 |
1144.1 |
1.25 |
1147.75 |
1.6 |
19.10 |
We also want to fit non-colliding bunches, but, alas, it appears statistics are too low (
beam 1 and
beam2). There is indeed some kind of excess around where you would expect (t = 1006 ns and t = 42324 ns, respectively), but these are in channels which show "spurious" bunches. As an example, here is a plot from Fill 1806 which shows this behavior,
This behavior arises in fill 1815 as well. Summarized below is the channel-by-channel behavior.
Channel |
Behavior |
Position |
11 |
Normal |
top |
12 |
Periodic |
near |
13 |
Periodic |
bottom |
14 |
Normal |
far |
21 |
Normal |
top |
22 |
Periodic |
far |
23 |
Periodic |
bottom |
24 |
Normal |
near |
Also, Here is the pre-collision rates by bunch type.
NB: The rate is by block transfer and has a 2 minute binning. The rate is therefore the number of hits per block transfer per 2 minutes.
So, it seems it may prove quite difficult with the current settings to get a fit for non-colliding bunches while separating out the beams. By summing over all colliding bunches in the precolliding range, we should be able to, at the very least, get a beam-inclusive measurement.
Fill 1804
The following analysis was done using data-taken during fill 1804 between midnight and 5am (
details here). The calibration constants used to shift the peaks previously have all been set to 0. This is because there were large inconsistencies between the peak position of the different channels in
BCM1F when I looked at the distributions while applying the correction factors. The bunch is 999, which should correspond to a bucket centered at 31190.6 ns. And of course, the assumption is that the HV is set to 250V during this period.
First we look at fits using a single Gaussian. The fit range is restricted so that the tails are not included.
LINK! Below is a summary of these results.
Channel |
mean (ns) |
σ (ns) |
χ2/NDF |
1/1 |
31193.4 |
5 |
4.7 |
1/2 |
31192.7 |
4.2 |
35.9 |
1/3 |
31192 |
4.6 |
4.5 |
1/4 |
31192.2 |
4.3 |
12.95 |
2/1 |
31193.5 |
4.6 |
12.3 |
2/2 |
31197.8 |
5.4 |
7.5 |
2/3 |
31193.2 |
4.6 |
2.4 |
2/4 |
31192.7 |
3.8 |
16.9 |
If we are sufficiently sensitive, we would like to see an enhancement from beam gas for the incoming and outgoing beam. The outgoing beam is likely to be swamped by post-collision effects (afterglow), but you can see that for channels 12 and, to a lesser extent, 24 there is a bit of an excess of hits before the main peak. This is fitted using a double gaussian,
here. There is definitely an improvement, particularly, for channel 12.
--
NateOdell - 26-May-2011