MINOS


UCL MINOS Homepage
Back to my MINOS



I have been writing an algorithm to remove crosstalk from data at CalDet.
This is important for tracking and also to correct dE/dX for the relative calibration of the detectors.

1: A cosmic ray muon

The PMT face for 'Left - Horiz Green, Plane 32' and 'Right - Horiz Clear, Plane 34' is shown below. The crosstalk hit in Plane 32 is identified and (optionally) removed. The hit in plane 34 could be crosstalk but the probability of this happening is very low (See Diagram 3) Or it could be noise or a neutron or something.


2: PMT face showing hits in two planes of the above diagram


3: Crosstalk hits from Pixel 1 to other pixels
Crosstalk to non-adjacent pixels is very low - 0.01% in this case. The spot position is also important, if the spot is in the centre, the crosstalk is even lower.


The algorithm works through all the hits in a given two planes and looks at the 8 pixel window around each 'test' hit.
It sums the charge and identifies the largest hit in the window.
The test hit is then flagged as crosstalk if its charge is less than the sum of the pixels around it, multiplied by a user definable factor. This is set to 1 for cosmic/beam muon tracking since hit density on PMT faces is low and low energy deposition from muons combined with photon statistics can lead to crosstalk hits having very comparable energies to real hits.
My crosstalker is a module that runs in Chris' CalDetTracker muon tracking code.
I'd been testing it on cosmic and beam muons previously, Mike generated a crosstalking Monte Carlo which I'm now using. This enables me to run Chris' Tracker with the Crosstalker and compare it to the MC truth information.


4: Number of Crosstalk Hits: Truth vs Tracked

For the most part, the crosstalker works well. It finds about 82% of the crosstalk hits and I've opted to be 'careful' when looking for them, so my purity is above 90%.
You can see a few hits in the wings that I've identified as crosstalk that aren't. There's a corresponding discrepancy in the centre in these planes - I think these are from scattered muons and I've called part of the end of the track crosstalk.


5: De/Dx with crosstalk removed- Truth/Tracked - 4 views

I expect this plot to be flat at 1. There seems to be a dip at the end. This implies that I see more energy deposited after Tracking than the truth tells me.


6: Close up

7: Tracked - Left: Energy Deposited, Right: Number of Hits

To try and explain the discrepancy I plotted energy deposited and number of hits for the Tracked and the Truth. Only Tracked is shown here. Then I divided Tracked by Truth and those plots are shown below.

8: Energy deposited - Tracked/Truth

9: Number of Hits - Tracked/Truth

Both the energy deposited and the number of hits is higher after tracking than the truth. Muons tend to deposit more energy just before they stop and can often produce photons making the end of the track 'messy'. So I believe this effect is due to crosstalk hits, especially interplane ones, being included in the track accidentally if they are very close to the track itself.

Next:

Compare De/Dx after crosstalk folded back. Unfortunately, I don't have information in the Monte Carlo about where the crosstalk actually came from so I'm just going to run the same 8-pixel window algorithm that I use in Chris' Tracker. Hopefully this will make plot 6 even flatter.

I can also try to make the crosstalker into a module that runs properly in the framework if people want to use it...?