LHC beam operation of the SLB boards
Collisions
First candidate collisions: here |
SLBs in ECAL barrel registered them at bx 52 |
|
|
Beam splashes
Analysis
The particles produced after a beam splashed traverse the CMS detector resulting in an increase of the production of trigger primitives
and in consequence in the increase in the occupancy of the SLB accumulators.
The excess in the trigger primitive production rate can be searched
using a method based on the computation of a significant deviation from background only counts
or by computing the correlation with respect to a reference distribution.
Description (show) Description (hide)
Search for deviation from background only counts
The instant of the beam splash can be found by summing up all the SLB accumulators and getting the most probable value of the obtained distribution.
The accumulators are then divided into 2 regions:
- A: [0,bx-50[ U [bx+50,1023[ - background region (dominated by cosmics and electronics noise counts)
- B: [bx-50,bx+50[ - signal region (expected region for the splash)
The average number of counts per bunch recorded by the accumulators in these regions is computed as:
.
The significance of the increase of counts due to beam can thus be estimated from:
One can ask if the maximum observed in each accumulator is compatible with the noise counts expected, as measured in the background dominated region A.
The answer is found based on Poisson statistics: the probability that the counts observed at the maximum of an accumulator histogram - - are due to background effects (cosmics+electronics noise) is given by:
From the expression above the following P-value is computed for the total probability that background effects yield or above trigger primitives above the threshold:
Correlation method
Details on the correlation method can be found in Test beam operation of the CMS calorimeter trigger synchronization boards.
The results obtained with these methods are shown below.
2009 results
Friday 20 Nov 2009 |
Comments |
Threshold set to 4 (1 GeV) |
Sum |
Maximum distribution |
Signal intervals defined |
Rate in Background Region (A) |
Rate in Signal Region (B) |
Significance |
|
|
[293; 393[ |
|
|
|
Delays (individual maximum - maximum in the total sum) |
p-value for maximum |
p-value as function of the accumulators summed up |
|
|
|
Delays (from correlation with maximum distribution) |
Correlation |
Significance |
Case studies for some channels |
|
|
|
|
Delays (from correlation with ECAL sum) |
Correlation |
Significance |
Case studies for some channels |
|
|
|
|
List of channels falling behind most probable bunch |
channel list |
List of channels laying ahead most probable bunch |
channel list |
From Sat Nov 7 2009 to Sun Nov 8 2009 |
Comments |
Three sets of histograms have been stored and are shown. Each row in the plots correspond to a different acquisiton time. After each acquisiton the histograms are reseted. - First row aquired in Sat Nov 7 23:37:19 2009 and corresponding run is 120020
- Second row aquired in Sun Nov 8 11:20:21 2009 and corresponding run is 120025
- Third row aquired in Mon Nov 9 01:17:57 2009 and corresponding run is 120042
- Threshold set to 3 (corresponds to approximately 750 MeV)
- Endcap histograms were not logged
|
Sum |
Maximum distribution |
Signal intervals defined |
Rate in Background Region (A) |
Rate in Signal Region (B) |
Significance |
|
|
EB [54; 154[ [54; 154[ [54; 154[ |
|
|
|
Delays (individual maximum - maximum in the total sum) |
p-value for maximum |
p-value as function of the accumulators summed up |
|
|
|
Delays (from correlation with maximum distribution) |
Correlation |
Significance |
Case studies for some channels (2nd dataset only) |
|
|
|
Clear evidence (no delay) Clear evidence (delay=-1) Noisy channel (delay=-6) Noise channel (delay=0) Almost no counts (delay=+1) |
List of channels falling behind most probable bunch |
channel list |
List of channels laying ahead most probable bunch |
channel list |
List of channels not ok (empty histograms mostly) |
channel list |
First LHC beam splash (2008)
(show) (hide)
A consequence of incoming beam is the increase in the occupancy of the SLB accumulators.
First beam was expected around bunch 680 () so an increase of the occupancy in this region was expected.
Dividing the accumulator into 2 regions:
- A: 0 < bx < 500 - cosmics and noise dominated region of the accumulator;
- B: 650 < bx < 750 - expected region for beam;
one can compute the average number of counts per bunch recorded by the accumulators in these regions as
.
The significance of the increase of counts due to beam can be estimated from:
The plots in the table below show the results for ,, and and one can conclude that something was observed in the region 650 < bx < 750 but
with relatively low significance () on a single accumulator basis.
Occupancy increase in 650<bx<750 |
Region A: 0 < bx < 500 |
Region B: 650 < bx < 750 |
Significance for occupancy in region B |
|
|
|
Timing
Looking for the bunch corresponding to maximum number of counts in each accumulator we get the results shown in the table below.
Bunch for maximum in accumulators |
Bunch |
|
Total |
681 |
45 |
|
682 |
2152 |
683 |
3 |
<681 or > 683 |
393 |
One can ask if the maximum observed is compatible with the noise counts expected, as measured in the region A (0<bx<500).
We can give an answer based on Poisson statistics: the probability that the counts observed at the maximum of an accumulator histogram - - are due to background effects (cosmics+electronics noise) is given by:
From the expression above we compute the following P-value: the total probability that background effects yield or above trigger primitives above the threshold is:
Below we show the result obtained for the distribution of the maximum of each SLB accumulator and the corresponding P-value.
Timing results |
Maximum bunch position with respect to bx=680 |
|
|
|
From the distribution for P-value we see that the large part of the accumulators has a low P-value which indicates that the counts observed at the maximum on the histogram cannot be due to background effects only. There is, however a considerable ammount of accumulators with high P-value (counts are dominated by background effects).
Instead of looking to individual timing we can look to the full ECAL barrel integrated timing by summing up all the SLB accumulators. In this case we proceed as follows:
- sum the histograms one by one
- for each new sum re-compute and
- compute the P-value corresponding to
- plot the P-value as function of the number of trigger towers used in the sum
The plots below summarize the results obtained:
Timing results (barrel integrated) |
Accumulator sum |
Zoom for the accumulator sum in region B (650<bx<750) |
evolution with the number of accumulators used |
|
|
|
From the behavior of the P-value distribution we conclude that the first LHC beam was correctly identified by the ECAL SLBs. The delay with respect to the expected bunch (680) was measured to be 2 bunches (50ns).
--
PedroSilva - 12 Sep 2008