triggering with flashcam · triggering with flashcam and a metric for assessing trigger homogeneity...

Triggering with FlashCamand a metric for assessing trigger homogeneity

Aaron Manalaysay17 May 2011

CTA Consortium MeetingToulouse, FranceMC/ELEC/TEL/FPI

In collaboration with the FlashCam group:

2

MC/ELEC/TEL/FPI, CTA Consortium Meeting, Toulouse France

A. Manalaysay, 17 May 2011

Overview

FlashCam

Trigger implementation

Trigger simulation, trigger algorithms

Trigger homogeneity: metric 1

Trigger homogeneity: metric 2

3



FlashCam baseline topology

PMTPreamplifierSlow control

HV

Ring bufferPreprocessed trigger

(data reduction)Slow control

250 MS/s12 bitFADC

Main trigger

Ethernet

Ethernet

Analog transmissionvia CATx

Bus (e.g. CAN)

… many possible implementationsof the technical details in thisblock diagram.

4





HV



250 MS/s12 bitFADC

Main trigger

Ethernet

Ethernet


Bus (e.g. CAN)

FPGA-based trigger,working on digitized data

5





HV



250 MS/s12 bitFADC

Main trigger

Ethernet

Ethernet


Bus (e.g. CAN)

12-bit per pixel, onlywhen triggered

… continuous data flow(don’t want 12-bit perpixel continuous)

So the question becomes, howcan we reduce the data goingto the triggering system?

6



Data reduction for the triggerThe trigger does not necessarilyneed the angular resolution of asingle pixel. Instead, create a patchof pixels that becomes a “triggerpixel”.

Patch of7 pixels(“p7”)

Patch of3 pixels(“p3”)

7



Trigger hardware strategy

CamTrg: External camera triggerClock: Camera clock (e.g. 10 MHz)Sync: Synchronization (e.g. 1 pps)PrepTrig: Preprocessed trigger data

4 x

21 x

Eth

erne

t to

PC fa

rm

Analogue data6 x 20 CAT5

Slow control

Crate # 2

Crate # 3

Crate # 4

Crate # 1FA

DC

boa

rd #

124

pix

els

FAD

C b

oard

# 2

24 p

ixel

s

Inte

rfac

e or

Trig

ger B

oard

FAD

C b

oard

# 2

024

pix

els

8



FlashCam Trigger Simulation MST-size telescope/camera, parabolic mirror (files generated by Konrad)

Gaussian jitter (σ =1ns) applied to all PE emission times

100 MHz NSB

SPE resolution of 0.42

AP proportion of 2E-5, log slope of 3 (factor 3 reduction every 5 PE)

Analog filtering: 6-pass 80 MHz RC filter

Baseline noise: filtered RMS of ~0.1 PE

250 MS/s sampling; 1PE corresponds to 10 ADC bins

… then apply triggering algorithm…

9



FlashCam Trigger Simulation MST-size telescope/camera, parabolic mirror (files generated by Konrad)

Gaussian jitter (σ =1ns) applied to all PE emission times

100 MHz NSB

SPE resolution of 0.42

AP proportion of 2E-5, log slope of 3

Analog filtering: 6-pass 80 MHz RC filter

Baseline noise: filtered RMS of ~0.1 PE

250 MS/s sampling; 1PE corresponds to 10 ADC bins

The following results are preliminary. (should I still showthem? Why not, I have nothing to lose.. in Toulouse).

10



Digital Majority: ≥ 3 pixels above threshold

P7-1

P7-3

P3-4

P3-7

Trigger algorithms

P[N]-[M] = “patches of N pixels, combining M such patches”

,,

(all unique triplets)

(all unique quadruplets)

(all uniqueseptuplets)

11



Afterpulsing, clipping, threshold

All algorithms arecompared in anormalized way. Thismeans, the threshold isadjusted so as to alwaysgive an NSB accidentalrate of 500 Hz.

I am using a clippinglevel of 10 PE in allsubsequent plots.

NSB

12



Threshold variation

Simple to calculate thetriggering efficiency asa function of number ofPE. One performanceparameter is thethreshold for 50% triggerefficiency (shown herefor one energy, onetrigger algorithm).

13



Threshold variationThe threshold for 50% trigger efficiency, for fouralgorithms and three benchmark energies.

14



Effective areas

15



Homogeneity of the triggerThe homogeneity problem: Theefficiency for detection here andhere should be ~equal

16



Homogeneity of the trigger

Quantify the showerposition as the CenterOf Gravity (C.O.G.) ofthe raw PEs.

17




For all cases, the basicrepeated unit is the patch. Soif there is any inhomogeneity,it should repeat in the samepatter with which the patch isrepeated, regardless of howthose patches are grouped.

This symmetry allows me tostack all patches for a giventrigger algorithm, andrepresent the whole cameraas a single patch.

18



Homogeneity of the triggerScatter plot shows the C.O.G.of only those showers thattrigger, taken from the greenband below. As expected,the density of points in thecenter is higher than at theedges.

(p7-1 has no overlappingpatches, so is expectedto have the worsthomogeneity)

19



Scatter plot shows the C.O.G.of only those showers thattrigger, taken from the greenband below. As expected,the density of points in thecenter is higher than at theedges.

… but how to quantify this in away that facilitates a solidcomparison betweenalgorithms?


20



Homogeneity: threshold variationOne idea: break up the patch into threeregions of equal area, compare thetrigger roll-off of the inner region withthat of the outer region. Use the relativevariation at 50% detection efficiency.

21



Homogeneity: threshold variationOne idea: break up the patch into threeregions of equal area, compare thetrigger roll-off of the inner region withthat of the outer region. Use the relativevariation at 50% detection efficiency.

22



Homogeneity: threshold variation

What is good enough? Well, equalization of PMT/ampgains will likely not be better than ~few% level, so thehomogeneity needs to be at least below this level (but notnecessarily an order of magnitude lower…)

23



Homogeneity: threshold variationThis metric could perhaps be improved in two ways:

•Arbitrary choice of trigger efficiency to measure thevariation.

•Only quantifies this variation at a single spot.

24



Homogeneity: roll-off variationAnother idea: again break up the patchinto three regions of equal area, comparethe trigger roll-off of the inner region withthat of the outer region. I call the areabetween the two roll-off curves Δeff.

25



Homogeneity: roll-off variationLess

homogeneious

morehomogeneious

26



Summary and next steps FlashCam trigger is modular, and versatile. Considering two pixel-grouping schemes:

3 pixels or 7 pixels, with many possibletriggering algorithms.

P3-4 shows best effective area P3-7 shows best homogeneity Next: investigate higher NSB, higher AP

rates. What amplitude resolution does the

trigger need? (12-bit? 4-bit?)Thanks for your attention!

triggering with flashcam · triggering with flashcam and a metric for assessing trigger homogeneity...

Documents