Noise model simulation for the DUNE FD DAQ Miquel Nebot, Babak Abi, - PowerPoint PPT Presentation

Noise model simulation for the DUNE FD DAQ Miquel Nebot, Babak Abi, Justo Martin-Albo � 1

Noise modeling: motivation • Started some time ago 
 https://indico.fnal.gov/event/15612/contribution/2/material/slides/0.pdf • Is one of the high priority R&D tasks in the Data selection list : 
 Trigger Development and Testing 
 C. Adding primitive generation in simulation 
 ”Clean” MC waveforms + o ffl ine noise models • Noise model : 
 The physics performance will depend strongly on the noise level 
 Need a tuneable noise model 
 Define configurations: MicroBooNE-like, ProtoDUNE-like, CE interface spec (600e − incoherent, 1500e − incoherent+coherent) 
 Also check as a function of parameters 
 Use ProtoDUNE data when available 
 � 2

Previous work Noise simulation/removal services focused on protoDUNE Jingbo Wang, David Rivera, Robert Sulej, Michael Mooney • Based on MicroBoone parametrisation by IFT • Di ff erent noise generators introduced to Larsoft/dunetpc 
 – Exponential model 
 – MicroBooNE model 
 – Coherent model MicroBoone noise paper, arXiv:1705.07341v1 • Code under debugging, not available yet under repository � 3

Noise modeling: considerations • The approach is hardware based and in time domain. FEMB - 128 ch = 8 x 16FE ASICS-> 8 x 16FE ADC-> 2 COLDATA • In frequency domain we can not Each coldata 2 x 64 measure the phase offset between different components or quick frequency shifts …. • Creating noise components, (harmonic/random…) in time domain regards to Hardware circuits to mimic the frequency specification of (FFT) noise. • We creating a flexible tool having all noise components then use them to mimic a realistic noise model whenever we need. Analog ASICs (FE and ADC) and Cold Digital ASIC (COLDATA) • 128 channels per FEMB = 40 Interface_Agreement_v1 2015 U + 40 V + 48 X Channels 4

Noise modeling: considerations • Noise sources, Model, Characteristics; three types of noise in TPC: 1. Fundamental noise (white/pink noise,…) • Comes from analog readout, proportional to the total capacitive load on the input channel(wire length,..) and channel independent • Simulated already in a basic level in MC (take out and add it latter) 2. Parasitic and EMI (pickup noise, HV/LV PSs ….) • Location (channels) dependency • Coherent noise spectrum characteristics is steady through time. Might need to be monitored daily bases (done offline) • Not simulated in MC 3. Detector noise (Microphonic, ..) • characteristics is not steady!? Should be study • Signal distortion 1. Pedestal drift, death channels • 2 scenarios: MicroBoone noise level and ProtoDUNE noise level 
 tune noise model characteristics like noise geometry, level for collection & induction planes, spectrum,.. To DUNE. The MicroBoone MB scheme is different than DUNE 
 so we focus on protoDUNE-like noise model � 5


 Noise types: 1-Color noise S α ( f ) ∝ 1 • Color noise: 
 f α White, pink, red (Brownian), grey 
 Tuneable: Amplitude • Applied randomly to each channel • Now implemented Pink ∝ =1 
 As learned from microBoone and 35t � 6

Noise types: 2-Coherent noise • Coherent (Common) harmonic noise : periodic (fixed frequency) + random (chirp). 
 Tuneable: Amplitude, Frequency and PhaseO ff set • The Coherent noise model is characterised by a series of wide Gaussian 
 bumps and a number of discrete narrow lines. 
 Coherent noise = exponential background + constant o ff set + Gaussian peaks From 3x1x1 � 7

Noise types: 2-Coherent noise • Coherent harmonic plus non harmonic • LV regulator • 6 (or N) harmonic components with different phase and amplitude • Non harmonic coloured noise (common mode) • Quick frequency drift components (only one ) • HV • 2 (or N) harmonic components with different phase and amplitude • Highest effect on U plane almost zero amplitude on X plane (learned fro microBoone) � 8

Signal distortion: pedestal drift • Pedestal drift (implemented as filtered white noise) subject to be studied in protoDUNE, probably Brownian (red) noise will described better. 
 � 9

Signal distortion: geometry effects Change/tune coherent (common) noise by detector geometry, boards arrangement: 128 channels per FEMB = 40 U + 40 V + 48 X Channels CoherentBundelAmp = [ 1 1 2 1 2 1 2 2 4 5 3 4 4 5 4 3 2 2 1 1 ] ; % amplitude in each FEMB � 10

Noise model 1+2+SD Noise modelled already in Mathlab: • Coloured noise • Coherent noise • LV regulator • HV • Signal distorsion � 11

Noise model 1+2+SD Parameters file DriftSamples= 4494; samplerate = 2e6; APAchannels = 2560; Uchannels = 800 ; Vchannels = 800 ; Xchannels = 960 ; • • Ratio_XtoUV = 0.8 ; % Ratio of noise of Collectio to induction plane • ShapingTimeC = 1 ; % uS • AmpRMSNoise = 3 ; % noise ADC RMS UV • NonC_Noise_model = 1 ; % pink 1 , white 0 , Brownian 2 , Blue -1 , violet -2 • %LVCohNOiseBudle= 32 ; • LVBund_U = 40 ; LVBund_V = 40 ; LVBund_X = 48 ; • HVCohNOiseBudle = 2560 ; • % Coherent nois harmonics • NHarmonCom = 6 ; % number of components • CohPhaseoffset= zeros(1, NHarmonCom) ; % phase different of components would be added randomly • %CohPhaseoffset = [] ; • CohMixAmp = [ 3 2 2 2 2 1 ] ; % Amplitude of components ADC • CN_Frq = [ 5300 11000 23000 49000 55000 99000 ] ; % frequency of components AD • % Coherent nois NON harmonics • Amp_NC_NonHarmo_pink = 1.0 ; • CoherentBundelAmp = [ 1 1 2 1 2 1 2 2 4 5 3 4 4 5 4 3 2 2 1 1 ] ; % amplitude in each FEMB • CN_FEMB_UVX_ratio = [ 1 ,1 , 1] ; • LVRegFreqdirfter= { 'TargetFrequency', 250000, 'InitialFrequency', 450000} • % Add HV Noise • HV_NHarmonCom = 2 ; % number of components • HV_CohMixAmp = [ 2 4 ] ; % Amplitude of components ADC • HV_CN_Frq = [ 33000 110000 ] ; % frequency of components AD • HV_CN_FEMB_UVX_effect = [ 2.5 , 1 , 0.2] ; % Amplitude on different U V X channels 12

Noise addition: raw waveforms Temporary Python/C++ scripts: add noise to extracted waveforms (digits) prodmarley_nue_spectrum_dune10kt_1x2x6_5_20160827T183253_gen_5962ba7e-feda-4c65-a809-586dc6bb8926_g4_detsim_nozs.root 1APA Induction wires 2560Ch 4490 drift samples 2245 µ s Collection wires � 13

Noise addition: raw plus 1+2+SD model Addition of generated noise to 3 channels � 14

Noise addition: raw plus 1+2+SD model � 15

Noise addition: raw plus 1+2+SD model 3 planes wires reco Generated for 
 grouped channels 40 U V 48 X � 16


 Summary and plans • Some noise files available can be used to be added to waveforms for hit finding and trigger studies. 
 12 APA noise file at: /afs/cern.ch/user/b/babi/public/N1 
 readme file for detail of noise parameters • Extract useful information from protoDUNE tests tune the parameters • Request to production: 
 MARLEY events with Radiological background, NO Zero Suppression, NO Noise. • LongTerm: already started the integration in LarSoft � 17

Backup � 18

� 19