Time Calibration Tests Results using a waveform template Cross-talk measurements Stub asymmetry results Kalle Sulanke David Hardtke Bob Stokstad
DOR DOM
Scope at DOR Scope at DOM
One-way time = 1/2 round trip time ONLY IF Symmetry in pulse generation and transmission Signal processing same at both ends Golden Rule of Time Calibration
DOR DOM DOR - DOM testDOR - DOR test DOR-DOR has same clock for transmitting and receiving DOR-DOM Asymmetry Test
V t t= V/( V/ t) Clock period 50 ns Time resolution < 5 ns Rise time ~400 ns -> linear algorithm should be OK
DOR-DOR shifted by one clock tick = 50 ns
V t (ns) t av
Average asymmetry = -1.3 ns RMS asymmetry = 1.1 ns DOR - DOR test DOR
DOR Card Av. Asymm RMS asymm DOM A -1.3 ns 1.1 ns DOM B -1.2 ns 1.1 ns DOR Card Test of Two DOM Configuration DOM A = 20 cm stub DOM B = 40 cm cable DOR DOM A DOM B Runs A_02_04N B_02_04N
Timing Waveform Analysis To get roundtrip time, need to analyze waveform: –Template method -- compare two waveforms and calculate time shift –Fitting method -- Fit each waveform with function Failed fit functions -- Error Function, Gaussians, Polynomials Best fit functions: DOM A_13_188
Waveform Analysis (cont.) Linear fit over limited range, calculate crossing point with pedestal Fit range
Round-Trip times using three methods Waveform comparison rms = 0.9 ns Linear fit rms = 3.9 ns Complicated fit rms = 7.2 ns Note: absolute scale on x-axis not adjusted properly. OK for residuals, however.
Using measured waveform as template appears promising. Simple, fast, accurate. Robust? (likely) Candidate for use in ice. Waveform Analysis Summary
DOR - DOM waveforms
Better symmetry after some component adjustments
DOR - DOM typical results: Round trip time rms ~ 1 ns for 3.4 km cable in lab using template Boards otherwise "quiet" No transmission of data in other twisted pair.
Cross-talk studies DOR - DOM 3.4 km Ericsson quad on spool Measure round trip rms deviation with and without data xmission at (1 Mbit/s) in other twisted pair
Cross-talk measurement results round trip residual (rms ns) DOM A DOM B Data transmission off to Ericsson quad Data transmission on in Ericsson quad Runs X01_A,B NOX01_A,B Does Cross-talk occur in quad or on DOR card?
Cross-talk measurement results, cont. round trip residual (rms ns) DOM A DOM B Data transmission on to another quad Data transmission on in Ericsson quad Runs X01_A,B X03_A,B => Most Cross-talk occurs in quad
Cable cross-talk rms =4.4 nsNo cable cross-talk rms =1.5 ns Round Trip Time Residuals
Cross-talk, cont. Look at pedestal region: Communications OFF Communications ON
Cross-talk Calculate rms of signal in baseline region: Data transmission in another cable => Most cross talk is in cable
Cross-talk, cont. Try to reduce cross talk by using trapezoidal communications pulse Result rms DOM Arms DOM B 4.4 ns 6.0 ns => No Help Runs X05_A,B_trap
Timing error budget for clock calibration is 5 ns total -including frequency, offset, asymmetry DOM clock calibration randomsystematic quad sum frequency(sqrt 2 * intrinsic)2.82 clock drift1 offset(1/2 up-down asymmetry, component variations) 4 DOM clock calibration subtotal Conclude need to shut down data transmission during RAPCAL in order to meet timing requirement, given cross- talk levels in Ericsson quad. Synchronized shutdown of communications for calibration is now the planned operating mode.
Cable Stub Tests No Stub Configurations DOR DOMaDOMb 15 cm terminated unterminated 3.4 km cable end DOR DOMbDOMa 15 cm1750 cm3.4 km 12 DOR, DOMa, and DOMb are actually one DOR card
Cable Stub Tests Stub Configuration DOR DOMb DOMa 130 cm 1750 cm3.4 km 3
Stub test results (preliminary) DOMaDOMb asym rms (NB 1 clock tick = 50 ns. Above absolute asymmetry is due to systematic logic error and is < 1ns.) DOR card has single clock for all 3 channels => Can measure up - down asymmetry Use centroid of positive portion of pulse These results suggest that 1.3 m cable stub does not introduce an asymmetry with measurable effect on time calibration