Time resolved experiments (without timing tool)

Time resolved experiments (without timing tool)

Introduction

Pump probe experiments prove time-dependent structural changes within a crystal. Chemical reactions are initiated by a flash from the excitation laser and proved by an XFEL pulse after a precisely controlled delay.

At SACLA, we use an interleaved data collection scheme. The ratio depends on the experimental setting. For example:

1:1 (laser 15 Hz, XFEL 30 Hz) Light-Dark-Light-Dark-Light-Dark-....
1:2 (laser 10 Hz, XFEL 30 Hz) Light-Dark-Dark-Light-Dark-Dark-....

The light from the excitation laser is monitored by a photo-diode and the readout signal is stored in the SyncDAQ DB with the XFEL tag number. By looking at this value, we can tell whether an XFEL pulse was accompanied by the excitation laser.

Running Cheetah with image classification

To enable image classification, start cheetah-dispatcher with the --pd1_name option. The signal name for the photo-diode is available from the laser operator. If you have more than one sensor, you can specify in the --pd2_name and --pd3_name arguments.

cheetah-dispatcher --pd1_name=xfel_bl_3_st_4_pd_laser_fitting_peak/voltage

Now you have a new text box in the GUI. Here, you have to specify the threshold to tell excited (light, laser-on) images from non-excited (dark, laser-off) images. Consult your laser team for the value. If you specify "0" for all photo-diodes, image classification is disabled.

In time-resolved mode, Cheetah sorts images into the light dataset and dark1 dataset. Images that meet the criteria go into the light dataset. Images that do not meet the criteria AND just after a light image go into the dark1 dataset.

Below is an example; Suppose the threshold is 0.5.

RESULT     TAG     SIGNAL
light       1       0.9
dark1       2       0.0
light       3       1.0
dark1       4       0.0
light       5       0.9
dark1       6       0.1
...
light      101      0.9
dark1      102      0.1
REJECTED   103      0.1 # Excitation laser did not fire for some reason
REJECTED   104      0.1
light      105      1.0
...
light      201      0.9
light      202      0.9 # Excitation laser fired for some reason
light      203      0.9
dark1      204      0.1
light      205      1.0

Images 103 and 104 are rejected because it is not right after a light image. Note that images 202, 203 and 204 are not rejected.

--submit_dark_to

To work with 10 Hz excitation, add --submit_dark_to=2. Now images are classified into light, dark1 and dark2.

RESULT     TAG     SIGNAL
light       1       0.9
dark1       2       0.0
dark2       3       0.0
light       4       0.9
dark1       5       0.0
dark2       6       0.1
...

You can go up to --submit_dark_to=9.

--submit_dark_any

If you do not want to separate dark1, dark2, ... but want to keep them together, use --submit_dark_any=1. Now, no rejection occurs at all.

RESULT     TAG     SIGNAL
light       1       0.9
dark        2       0.0
light       3       1.0
dark        4       0.0
light       5       0.9
dark        6       0.1
...
light      101      0.9
dark       102      0.1
dark       103      0.1 # Excitation laser did not fire for some reason
dark       104      0.1
light      105      1.0
...
light      201      0.9
light      202      0.9 # Excitation laser fired for some reason
light      203      0.9
dark       204      0.1
light      205      1.0

Analysing the result

You can easily make CrystFEL list files as follows and process them separately. By dividing images as early as possible, you can reduce the risk of mixing them up.

ls ../*-light/run*.h5 > light.lst
ls ../*-dark1/run*.h5 > dark1.lst