CTF3 May 2013 Data Analysis

FONT Meeting 2nd August 2013 CTF3 May 2013 Data Analysis Phase Profile Plots and ADC Droop Correction with IIR Filter Jack Roberts

Overview • Phase profile with unfiltered Mixer/Diode. • Calibration with IIR filter, Mixer/sqrt(Diode). • Phase profile with filtered, Mixer/sqrt(Diode). • Unfiltered vs. Filtered phase reconstruction. • Mean pulse to pulse phase stability. Jack Roberts

Recap: Unfiltered Mixer/Diode Calibration CALIBRATION LO SCAN • Cancel ADC droop by using Mixer/Diode instead of Mixer/sqrt(Diode). • Absorb beam current A in to calibration constant a. • Fitted values for a: • M1: 1.67 • M2: 3.11 • M3: 1.92 • Best resolution near zero crossings of LO scan. • Plots for the points circled here are outlined in red. • All the plots are mean subtracted. 9 4 3 8 Jack Roberts

Phase Profile – No filtering, Mixer/Diode Calibration. Jack Roberts

Phase Profile – No filtering, Mixer/Diode Calibration Jack Roberts

Phase Profile – No filtering, Mixer/Diode Calibration • Good agreement between M1 and M3 near zero crossings of ADC output in LO scan. • Saturation effects near min/max of ADC output in LO scan due to sin(phi)>1 values. Jack Roberts

ADC Droop Exponential Fit • Fit an exponential to the diode data then use the fitted decay constants in an IIR filter: • Fitted values for τ (in 357 MHz samples), Diode ADCs: • M1: 518 ± 1 • M2: 440 ± 1 • M3: 457± 2 • Also have droop in mixer channel but can’t fit exponential to it. • Use the mean of the diode ADCs: τ = 472. • Can get these values from lab tests (on going). • Fitted values for τ can vary up to ~20% dependent on sampling window used, LO phase etc. • Resolution seems fairly insensitive to τ. Jack Roberts

ADC Droop Exponential Fit • Left: Points are mean diode channel data for each LO phase setting. Black lines are the exponential fit using the fitted decay constants. • Seeing cross-talk from Mixer? Jack Roberts

IIR Filter • Implement an IIR filter to remove the exponential droop using the fitted decay constants: • Can now perform a calibration using Mixer/sqrt(Diode) as planned: • Fitted values for a: • M1: 52.5 • M2: 84.1 • M3: 56.9 Jack Roberts

Phase Profile – Filtered, Mixer/sqrt(Diode) Calibration Jack Roberts

Phase Profile – Filtered, Mixer/sqrt(Diode) Calibration • At a first glance, very similar to the unfiltered calibration plots. Jack Roberts

Resolution Comparison – Filtered vs. Unfiltered • Resolution appears to be a bit better for the filtered calibration. Jack Roberts

Comparison – Unfiltered vs. Filtered M1 Jack Roberts

Comparison – Unfiltered vs. Filtered M3 • There are differences in the relative scale of the reconstructed phase from the two calibrations. Jack Roberts

Filtered vs. Unfiltered Calibration Constant Mean Mean • Filtered/unfiltered calibration constants divided by mean to make them comparable. • Only mean value used in phase reconstruction. • Differences vs. sample no. will cause phase shifts. Mean Jack Roberts

Mean Pulse Phase Stability Jack Roberts

Mean Pulse Phase Stability • Only short runs, but around 0.5o – 1.0o of jitter in the mean pulse phase. Jack Roberts

Summary • The IIR filter appears to do a good job but lab tests needed to accurately determine decay constants. • Difference in relative scale of reconstructed phase for unfiltered Mixer/Diode and filtered Mixer/sqrt(Diode) calibrations. • Can look in to cross-talk between the mixer and diode. • Side notes: • Installation of kickers, rearrangement of magnets etc. in TL2 for the feedforward system in progress. • M2 can’t be repaired due to radiation. Planned to test using horizontal pickups instead of vertical. 4th monitor will be built. • May be delays in installing M3 in its final location. • Thinking about how to integrate FONT DAQ with CTF3 data. Jack Roberts

CTF3 May 2013 Data Analysis