IceCube DOM Calibration

Preview:

DESCRIPTION

IceCube DOM Calibration. Jim Braun. Motivation. DOM-Cal – Big Picture. Application runs on DOM Can calibrate all DOMs in parallel Stores calibration data on DOM flash memory Java client Coordinates calibration Reads calibration data from DOM Produces DOM-Cal XML calibration files - PowerPoint PPT Presentation

Citation preview

April 20, 2005 1

IceCube DOM Calibration

Jim Braun

April 20. 2005 2

Motivation

April 20. 2005 3

DOM-Cal – Big Picture

• Application runs on DOM– Can calibrate all DOMs in parallel– Stores calibration data on DOM

flash memory• Java client

– Coordinates calibration– Reads calibration data from DOM– Produces DOM-Cal XML

calibration files• Calibrator class reads XML

file, applies calibration constants

• All code in dom-cal project on glacier

– Authors: John Kelley, Jim Braun, Kael Hanson

DOM hub

DOM-CalClient

XMLFiles

Database

Calibrator.java

Analysis

April 20. 2005 4

Calibration Routine

• ATWD Calibration– Reconstruct PMT voltage waveform from ATWD data– Requires:

• Voltage calibration of ATWD

• Measurement of amplifier gain for ch0, ch1, and ch2

• Frequency calibration of ATWD

• Baseline measurement

• PMT Gain Calibration– Find relationship between PMT gain and applied voltage

• PMT Transit Time Calibration (NEW!)– Find PMT transit time as a function of applied voltage

April 20. 2005 5

ATWD Calibration

• Bootstrap process!

V = 5 x bias_dac / 4096V = 0.0001220 x (0.4 x disc_dac – 0.1 x bias_dac)f = 20MHz

ATWDFrequencyCalibration

ATWDVoltage

Calibration

PulserCalibration

BaselineCalibration

Amplifier ch0Calibration

Amplifier ch1Calibration

Amplifier ch2Calibration

Muons

Mainboard LED

April 20. 2005 6

Pulser Calibration

• Determine relationship between pulser DAC setting and peak voltage– Use known relationship between discriminator DAC and voltage– Set discriminator DAC, adjust pulser DAC until 50% of pulses

cross discriminator threshold

– At this point, the pulser peak voltage corresponds to known discriminator voltage.

– Peak voltage distribution is very narrow

April 20. 2005 7

Pulser Calibration

• Repeat for ~10 discriminator voltages• Relationship is very linear

• Now know pulser peak voltage given DAC setting

• Will use this relationship in amplifier calibration

April 20. 2005 8

ATWD Voltage Calibration

• Determine relationship between ATWD value and signal voltage

• Use known bias DAC voltage relationship– Bias is independent of amplifier gain– Set bias, record average ATWD value

• For each bin (0-127) of each signal channel (0-2) of each ATWD (0-1)

• O(100) samples

– Apply linear fit to ATWD value vs. voltage data

• Pedestal patterns are eliminated

April 20. 2005 9

ATWD Voltage Calibration

• Now know voltage of any ATWD bin given a channel number, bin number, and value

• Requires 768 linear fits!

• ATWD response not entirely linear• Need to calculate and subtract small baseline offset for

each channel during precision measurements

April 20. 2005 10

Baseline Calibration

• New in DOM-Cal 5.0• Measures average baseline offset for each ATWD

channel• Need to measure baseline whenever the internal state of

the DOM changes• Known to affect baseline:

– Analog multiplexer– Mainboard LED power supply– PMT high voltage

• Affects low gain channel the most– Main source of Hagar’s ch1:ch2 charge discrepancy

April 20. 2005 11

Baseline Calibration

• We care most about affect of high voltage and residual baseline from imperfect ATWD calibration

• Take baseline data both with HV off and HV at values spanning DOM operating points

• Store all data points• Use baseline value closest to

operational HV when calibrating TestDAQ data

• For gain calibration, domcal chooses only the voltages where baseline calibration points exist

April 20. 2005 12

Amplifier Calibration• Calibrate high gain channel (0) with pulser

– Pulser too weak to accurately calibrate lower gain channels• Set pulser peak to a known voltage, record peak voltage after amplification

in ch0– Use ATWD ch0 calibration data to find peak voltage

• Maximize ATWD sampling speed to better localize peak

• Record mean and error of O(250) ch0 peaks

• Ratio of mean voltage and known pulser voltage yields amplification factor

April 20. 2005 13

Amplifier Calibration

• Need source of high amplitude pulses to calibrate ch1 and ch2 amplifiers

• Use PMT signals! – new in DOM-Cal 5.0– Muons work well at surface– Mainboard LED needed in deep ice

• LED power supply shifts baseline, need to recalibrate

• For ch1:– Select pulses which have an ATWD peak value of

600-800 counts• Too few ch0 counts -- too much error in ch1 peak voltage• Too many ch0 counts -- ch0 nonlinearity becomes significant

April 20. 2005 14

Amplifier Calibration

• For ch1:– Record ratio of ch1 peak

voltage to ch0 peak voltage for O(250) iterations

– We know ch0 gain, so ch1 gain is given by the product of ch0 gain and voltage ratio.

• For ch2– We now know ch1 gain, use

previous procedure to find ch2 gain

– Slow with muons (<1 Hz)– Slow when discriminator rate is

high– Use LED if necessary

April 20. 2005 15

ATWD Frequency Calibration

• Select mainboard oscillator in ATWD analog multiplexer channel (channel 3)

• At various sampling speed DAC values, count number of bins between positive zero crossings in ATWD waveform

• Average O(100) clock waveforms• Assuming oscillator operates in

spec @ 20MHz, ATWD frequency is given by 20MHz * #bins

• Not quite linear – newer version will sample closer to 850

DAC value

April 20. 2005 16

Gain Calibration

• Capture PMT single photoelectron pulses in ATWD– Glass radioactivity emits enough light

• Apply ATWD calibration to get PMT V(t) waveform

• I(t) given by V(t)/50• I(t) pulse integrated from –4 bins to +8 bins of pulse

maximum (~-14ns - +28ns), yielding SPE charge

• Repeat O(5000) times, histogram charge data, and apply nonlinear fit

April 20. 2005 17

Gain Calibration

Charge (pC)

DiscriminatorEdge

Noise

Single-PhotoelectronPeak

Muons, etc.

Fit Exponential +Gaussian

April 20. 2005 18

Gain Calibration

Repeat from 1200V to 1900V in 100V intervals

1400V

1600V 1700V

1500V

April 20. 2005 19

Gain Calibration

Mean SPE charge vs. voltage is a power law

Gai

n (c

harg

e /

e)

Voltage

Linear log - log fit yields operating point

The number of photoelectrons for any pulse at a given HV is now determined

April 20. 2005 20

Java Client

• In dom-cal project on glacier– Main class: icecube.daq.domcal.DOMCal

• Run with no args for usage instructions

• Reads calibration data from DOM flash– Stores calibration data in local XML files– Stores data in domprodtest database

• Can initiate calibration• Can run calibrations on many DOMs and DOM

hubs in parallel• Most will never need to use the java client

April 20. 2005 21

DOM-Cal XML Files

• Hopefully, users won’t need to know much about the XML files or database structure either!– Access provided through calibration applications

• XML files are reasonably easy to read if needed– Contain:

• DOM hardware ID (No name…..sorry Mark)

• Temperature

• Date

• DAC settings and ADC readings

• Calibration information– Linear fit data

– Baseline data

– Gain histogram data

April 20. 2005 22

DOM-Cal Calibrator

• Icecube.daq.domcal.Calibrator java class provides access to calibration data.– I3DOMCalibration equivalent object in IceTray

• Create a Calibrator for each DOM:– new Calibrator(XML_File);

• Most important routine:– atwdCalibrateToPmtSig()

• Takes an array of raw ATWD data and applies calibration to yield true voltage signal

• Other methods to access raw data, described in javadoc– http://www.amanda.wisc.edu/jbraun/domcaldoc/

April 20. 2005 23

DOM-Cal Results

• Currently running DOM-Cal v4.3 at pole

• Baseline shift apparent in data from String 21 and from last year’s FAT runs

• Analog multiplexer enabled in all string-21 runs – just recently disabled

April 20. 2005 24

DOM-Cal Results

• DOM-Cal 5.0 results are encouraging

• Baseline calibrated to zero for all channels

• No analog multiplexer effects

• Amplifier gains now calculated much more accurately

• Pulse heights now agree between all three ATWD channels

April 20. 2005 25

DOM-Cal Results

• The bad news– Probably won’t see much gain when analyzing String 21 data,

even with DOM-Cal 5.0

• Local coincidence readout rate is a few Hz

• Interval is long enough for baseline to drift

• Hopefully there is a firmware fix!

April 20. 2005 26

DOM-Cal 6.0++

• Other features DOM-Cal may deal with:– PMT transit time – almost finished!

– Ch0 time offset, bandwidth limitation

– Signal droop• Evidence time

constant can be easily measured (Chris W.)

Recommended