l1s modulator performance measurement - stanford … · improve l1s modulator stability ... • two...
TRANSCRIPT
L1S modulator performancemeasurement
Tao Tang
Outline
PurposeMeasurement setupMeasured signalTrigger sourceResultsFuture work
Purpose
Better understand L1S jitter problemImprove L1S modulator stabilityCross validation with other measurementsCollect data for thyratron tuning procedureCollect data for automated thyratron tuningCollect data for thyratron performance
Measurement setup
Oscilloscope:• Tektronix DPO 4054B, 2.5GSps, 500MHz BW, vertical
resolution (Hi Res) 11 bits (~980ppm)• Typical measurement setup: 100k point, full BW, ~3
measurement per secondVoltage offset:
• Preample Instruments 1855 Differential Amplifier• 100MHz BW, 50 Ohm, 7ppm resolution
Trigger generation:• SRS DG535 delay generator
Signal measured
PFN signal from Ross divider• Connected to thyratron anode• Buffered by new HV regulation board outside thyratron
cabinentBV signal from MKSU
• Measured at klystron tank with capacitive divider• Buffered by MKSU
Trigger signal from RF system and modulator• Increase pulse width to 5ms for better visualization by DG535
Signal measured - PFN
Source
Signal conditioning
Channel 1: Buffered PFN voltageà Scope (1MOhm)Channel 4: Buffered PFN voltageà Diff-ampà Scope (50Ohm)Channel 3: Buffered BVà Scope (1 MOhm)Channel 2: Buffered BVà Diff-ampà Scope (50 Ohm)Ferrite on each line to reject common mode signalRaw data without scaling (scaling factor needed)
Trigger
Clock base: AC line zero crossing (120Hz)Modulator trigger:
• Always 120 Hz• Acceleration trigger, stand-by trigger (6us before beam)
RF trigger/accelerator trigger• Variable rate• Fixed delay relative to clock base
Charging waveform
Measurement conditions• Infinite persistence• Long time scale (10min)• Low sample rate• 1 hour difference in screen time and real time
Charging waveform
Two-state charging due to AC phase• 120Hz trigger vs 60Hz AC line• Two charging states with opposite resonant peak polarity• Switching between two states due to operation condition
(frequency change)Possibly additional switching noise due to SCR voltageregulation in AIP stations
• SCR voltage regulation board working at 60° spacing(2.78ms)
• 6 SCR formed 6 poles
Large scale view (PFN charging)
CH1: PFN, CH2: Reconditioned acceleration trigger, CH3: BV, CH4: PFN w. offset
SCRswitching?
PFN charging
CH1: PFN, CH2: BV w. offset, CH3: BV, CH4: PFN w. offset
Very close to60° spacing(2.78ms)
Switching between two charging states
Charge state #1 Charge state #2
CH1: PFN, CH2: Reconditioned acceleration trigger, CH3: BV, CH4: PFN w. offset
Charge state change due to beam rate?
CH1: PFN, CH2: Reconditioned acceleration trigger, CH3: BV, CH4: PFN w. offset
Noaccelerationtrigger
PFN waveform shape change
CH1: PFN, CH2: Reconditioned acceleration trigger, CH3: BV, CH4: PFN w. offset
Thyratron instability
Thyratron instability• Self breakdown
- High gas pressure / reservoir voltage- Impact next a few pulses, much lower BV (beam drop out)- Can not start modulator
• Random recovery- Low gas pressure / reservoir voltage- Impact almost every pulse- High PFN instability
Random recovery
CH1: PFN, CH2: BV w. offset, CH3: BV, CH4: PFN w. offset
Worse PFNStability
Self breakdown
CH1: PFN, CH2: Thyratron grid, CH3: BV, CH4: PFN w. offset
Thyratron selfbreakdownNo grid trigger
Data acquisition system
Using Tektronix scope on sun-ray network• Can not measure PFN and BV at the same time• Jumping between two different configurations every 5 minute• Statistics for about 900 data points with a 200-point window
Save screen capture at regulate interval• Infinite persistence
Take measurements at regulate interval• Mean, minimum, maximum, standard deviation, sample count
Matlab script running on lcls-srv01• Daily cleaning routing problem• Lcls-srv02 for long term running• Matlab 2012a pause memory leak/inaccurate timing
Modulator stability
Data with less than 800 samples are ignored
Waveform with different
12/3 4:47 34ppm 12/3 5:18 170ppm
L1S phase
L1S low spike count
PFN voltage
Data with less than 800 samples are ignored
Beam voltage
Data with less than 800 samples are ignored
Sample count
Future work
Understand PFN stability measurementTest station?