How to extract digitizer peaks with the SCS Toolbox¶

[1]:

import toolbox_scs as tb
import matplotlib.pyplot as plt
%matplotlib notebook
plt.rcParams['figure.constrained_layout.use'] = True

First, to explore the data contained in a run, we open a DataCollection using open_run() which is equivalent to the extra_data.open_run() function:

[2]:

proposal, runNB = 2956, 13
run = tb.open_run(proposal, runNB)

Each channel of each digitizer stores a string that is used to describe which device is connected to that channel. A dictionnary of all channels can be accessed as follows:

[3]:

tb.digitizer_signal_description(run)

[3]:

{'FastADC_Ch0': 'CHEM DIAG X-ray PD',
 'FastADC_Ch1': 'PPL I0 at in-coupling window',
 'FastADC_Ch2': 'CHEM APD',
 'FastADC_Ch3': 'PPL Reflectometer PD',
 'FastADC_Ch4': '',
 'FastADC_Ch5': 'PPL 800 nm I0 ILH PD',
 'FastADC_Ch6': '',
 'FastADC_Ch7': 'No Description',
 'FastADC_Ch8': 'No Description',
 'FastADC_Ch9': 'CHEM DIAG Laser PD',
 'FastADC2_Ch0': 'No Description',
 'FastADC2_Ch1': 'No Description',
 'FastADC2_Ch2': 'No Description',
 'FastADC2_Ch3': 'No Description',
 'FastADC2_Ch4': 'No Description',
 'FastADC2_Ch5': 'diode 2 - diam = 5mm',
 'FastADC2_Ch6': 'I0_400 nm',
 'FastADC2_Ch7': 'diode 5  - diam = 3mm + Filter (Ti 400 nm / Polyimide 200 nm)',
 'FastADC2_Ch8': 'diode 6 - 10x10 mm + filter longpass > 550 nm',
 'FastADC2_Ch9': 'diode 8 - APD diam = 3 mm + Filter (Ti 400 nm / Polyimide 200 nm)'}

As is customary, we can always list the available mnemonics for the run as follows:

[4]:

tb.mnemonics_for_run(run).keys()

[4]:

dict_keys(['sase3', 'sase2', 'sase1', 'maindump', 'bunchpattern', 'bunchPatternTable', 'npulses_sase3', 'npulses_sase1', 'BAM414', 'BAM1932M', 'BAM1932S', 'nrj', 'nrj_target', 'M2BEND', 'tpi', 'VSLIT', 'ESLIT', 'HSLIT', 'transmission', 'transmission_col2', 'GATT_pressure', 'UND', 'UND2', 'UND3', 'XTD10_photonFlux', 'XTD10_photonFlux_sigma', 'XTD10_XGM', 'XTD10_XGM_sigma', 'XTD10_SA3', 'XTD10_SA3_sigma', 'XTD10_SA1', 'XTD10_SA1_sigma', 'XTD10_slowTrain', 'XTD10_slowTrain_SA1', 'XTD10_slowTrain_SA3', 'SCS_photonFlux', 'SCS_photonFlux_sigma', 'SCS_XGM', 'SCS_XGM_sigma', 'SCS_SA1', 'SCS_SA1_sigma', 'SCS_SA3', 'SCS_SA3_sigma', 'SCS_slowTrain', 'SCS_slowTrain_SA1', 'SCS_slowTrain_SA3', 'HFM_BENDING', 'VFM_BENDING', 'AFS_DelayLine', 'AFS_FocusLens', 'PP800_PhaseShifter', 'PP800_SynchDelayLine', 'PP800_DelayLine', 'PP800_HalfWP', 'PP800_HWP_POWER', 'PP800_FocusLens', 'FFT_FocusLens', 'hRIXS_det', 'hRIXS_delay', 'hRIXS_index', 'hRIXS_norm', 'hRIXS_ABB', 'hRIXS_ABL', 'hRIXS_ABR', 'hRIXS_ABT', 'hRIXS_DRX', 'hRIXS_DTY1', 'hRIXS_DTZ', 'hRIXS_GMX', 'hRIXS_GRX', 'hRIXS_GTLY', 'hRIXS_GTRY', 'hRIXS_GTX', 'hRIXS_GTZ', 'XRD_DRY', 'XRD_SRX', 'XRD_SRY', 'XRD_SRZ', 'XRD_STX', 'XRD_STY', 'XRD_STZ', 'XRD_SXT1Y', 'XRD_SXT2Y', 'XRD_SXTX', 'XRD_SXTZ', 'FastADC0peaks', 'FastADC0raw', 'FastADC1peaks', 'FastADC1raw', 'FastADC2peaks', 'FastADC2raw', 'FastADC3peaks', 'FastADC3raw', 'FastADC4peaks', 'FastADC4raw', 'FastADC5peaks', 'FastADC5raw', 'FastADC6peaks', 'FastADC6raw', 'FastADC7peaks', 'FastADC7raw', 'FastADC8peaks', 'FastADC8raw', 'FastADC9peaks', 'FastADC9raw', 'FastADC2_0peaks', 'FastADC2_0raw', 'FastADC2_1peaks', 'FastADC2_1raw', 'FastADC2_2peaks', 'FastADC2_2raw', 'FastADC2_3peaks', 'FastADC2_3raw', 'FastADC2_4peaks', 'FastADC2_4raw', 'FastADC2_5peaks', 'FastADC2_5raw', 'FastADC2_6peaks', 'FastADC2_6raw', 'FastADC2_7peaks', 'FastADC2_7raw', 'FastADC2_8peaks', 'FastADC2_8raw', 'FastADC2_9peaks', 'FastADC2_9raw'])

Let’s assume that we are interested in the APD signal on diode 8, which corresponds to Ch9 of Fast ADC 2. We will load the raw traces of this channel (mnemonic FastADC2_9raw) and extract the peaks.

This is, in principle, all done automatically in the load() function. The only piece of information that is required is which bunch pattern to use for the extraction ('sase3' if the device is looking at the FEL or 'scs_ppl' if the device is looking at the PP laser), so that all data loaded simultaneously (XGM, BAM, digitizer, …) is pulse-aligned. By default, data from FastADC2 uses the ‘sase3’ bunch pattern (parameter fadc2_bp), so we can in this case omit this parameter in the load function.

A peak-finding algorithm analyses the averaged trace and finds the location of the peaks and performs trapezoidal integration on the interval centered on the peak.

[5]:

proposal = 2956
fields = ['SCS_SA3', 'nrj', 'FastADC2_9raw']
run, ds = tb.load(proposal, 13, fields,)
ds

We see here that ds has a variable FastADC2_9peaks which has dimensions ['trainId', 'sa3_pId'], which is exactly what we wanted: the raw traces were automatically transformed into vectors of length sa3_pId.

If, for instance, the recorded signal was an optical laser(OL) pulse and that the OL pattern was different from the sase 3 bunch pattern, we would need to specify fadc2_bp='scs_ppl' in the load function.

We can always inspect how the peak extraction was performed by using the following function:

[6]:

good_params = tb.check_peak_params(run, 'FastADC2_9raw')
good_params

bunch pattern sase3: 400 pulses, 96 samples between two pulses
Auto-find peak params: 400 pulses, 96 samples between two pulses

[6]:

{'pulseStart': 6759,
 'pulseStop': 6776,
 'baseStart': 6743,
 'baseStop': 6755,
 'period': 96,
 'npulses': 400}

This shows a plot with the first and last pulses identified by the peak-finding algorithm and displays the region of integration and the region for baseline subtraction. The function returns a dictionnary good_params that has all parameters necessary to perform the trapezoidal integration over the digitizer trace.

Summary with another example¶

Check digitizer signal descriptions and load data¶

[12]:

proposal, runNB = 2953, 1
print(tb.digitizer_signal_description(tb.open_run(proposal, runNB)))

fields = ['FastADC0raw', 'chem_Y']
run, ds = tb.load(proposal, runNB, fields, fadc_bp='sase3')

{'FastADC_Ch0': 'CHEM DIAG X-ray PD', 'FastADC_Ch1': 'PPL I0 at in-coupling window', 'FastADC_Ch2': 'CHEM APD', 'FastADC_Ch3': 'PPL Reflectometer PD', 'FastADC_Ch4': '', 'FastADC_Ch5': 'PPL 800 nm I0 ILH PD', 'FastADC_Ch6': '', 'FastADC_Ch7': 'No Description', 'FastADC_Ch8': 'No Description', 'FastADC_Ch9': 'CHEM DIAG Laser PD'}

The period from the bunch pattern is different than that found by the peak-finding algorithm. Either the algorithm failed or the bunch pattern source (sase3) is not correct.

Check integration parameters¶

[13]:

params = tb.check_peak_params(run, 'FastADC0raw', bunchPattern='sase3')
plt.figure()
tb.get_dig_avg_trace(run, 'FastADC0raw').plot()

bunch pattern sase3: 400 pulses, 96 samples between two pulses
Auto-find peak params: 521 pulses, 93 samples between two pulses

[13]:

[<matplotlib.lines.Line2D at 0x2b57b8d0d588>]

Define and check new integration parameters¶

[14]:

params['period'] = 96
params['npulses'] = 400
params['pulseStart'] = 22491
params['pulseStop'] = 22525
params['baseStart'] = 22480
params['baseStop'] = 22485

params = tb.check_peak_params(run, 'FastADC0raw',
                              bunchPattern='sase3', params=params, )
params

bunch pattern sase3: 400 pulses, 96 samples between two pulses
Auto-find peak params: 400 pulses, 96 samples between two pulses

[14]:

{'pulseStart': 22491,
 'pulseStop': 22525,
 'baseStart': 22480,
 'baseStop': 22485,
 'period': 96,
 'npulses': 400}

How to extract digitizer peaks with the SCS Toolbox¶

If the peak-finding algorithm fails¶

Summary with another example¶

Check digitizer signal descriptions and load data¶

Check integration parameters¶

Define and check new integration parameters¶

Reload data with raw trace and extract peaks using the new integration parameters¶