Sensor and Raft testing.


5 Νοε 2013 (πριν από 3 χρόνια)

60 εμφανίσεις

Laboratory Information
Management System for
Sensor and Raft testing.


Ivan Kotov

for BNL group

March 22, 2012


Single sensors and Raft Tower Modules have to be tested
and characterized before the focal plane assembly

large volume of testing

high throughput is required

can not be achieved without automation

automate tests from the very beginning

CCD test facility was set up at Brookhaven National
Laboratory Instrumentation Division

Study contract sensors have been tested

sensor format from 2k x 0.5k to 4k x 2k

>50,000 image files (FITS format) were produced, analyzed and

Introduction (cont.)

in testing mode

is as important as

an image is useless without knowledge of
measurement conditions: temperature, bias voltages,
clock levels and timing etc

Automation approach is based on

Rts2 (robotic telescope system 2)

bash scripts

MySQL database

Express analysis, C/C++ code in ROOT framework

details are in

Data volume, EO tests

Image data

single sensor: 32 MB /2s

RTM: 288 MB /2s


single sensor:

currently over 200 values

picoammeter tables

RTM: ASPIC & CABAC registers

temperature, biases etc

where it goes: header, DB,

Analysis code

variety of development platforms is in use

ROOT framework, C/C++ code

Python code



storage of analysis results

data base entries with analysis summary

version control

Production needs

cyber security of operations

data storage


version control


workflow monitor

inventory tracking


instrument calibration and
maintenance record & schedule


Laboratory Information Management
System, LIMS

Requirements & Proposed Solutions

cyber security of operations

computers in LSST clean rooms, buildings 510
and 535B are a mixture of machines running
Linux and Windows operating systems.

machines will be on a private local area network
(LAN) behind a dual homed gateway machine

allow outgoing traffic from the machines

not allow incoming traffic to go directly to the internal

machines can be accessed by logging into the
gateway machine

Requirements & Proposed Solutions

data storage

Use existing RHIC and ATLAS Computing
Facility (RACF) at BNL which is managed and
maintained by the RACF staff. Data files are
stored on a dedicated file server residing in
the facility.

files are backed up nightly using RACF
backup system and are archived using the
IBM High Performance Storage System
(HPSS) also within the RACF.

These backups provide both disaster
recovery and long
term availability of the

Requirements & Proposed Solutions


MySQL database will be used as the LIMS
back end

This database will be installed on an RACF

Write access to the database will be limited
to machines in the clean room and an
authenticated web interface

a read
only replica of the database can be
installed at remote locations

Requirements & Proposed Solutions

version control

A Subversion (SVN) server will track all code
and scripts used in test procedures in the
clean rooms as well as changes made to the
code and scripts.

Versions of all code and scripts used for a
particular test will be recorded in the database
allowing subsequent searches on this

End of presentation

Measurement metadata logging & browsing

Information about environment conditions and instrument settings for measurements of
interest is stored in FITS headers but

search and data retrieve is slow (files need to be open, read etc)

there is no mechanism to narrow down the search (what files to search


something more flexible and convenient is needed

Fast, flexible and convenient access to metadata can be achieved using a database
(DB). The ease of access allows one

keep track of performed measurements

search and compare

plot essential measurements parameters


open source product

multi platform capability

performance and reliability

user familiarity and wide knowledge base

The metadata logging is performed by running the executable from a measurement
script when a set of measurements is completed

the executable is built from C/C++ code

table altering is supported

tables are defined in the code header file

change the table layout easily

keep track of the changes

visualization GUI was developed in Python using PyGTK open source package

DB GUI example

horizontal axis

time of observation

Express analysis of
Fe data (1)

standard input signal ( K

1620 e

provide absolute calibration of the entire
electronics chain in a very straightforward

what CCD parameters can be measured:

system gain

system noise



Fe data

Fe data are obtained by multiple CCD exposures to a 10 μCi source

source swings over CCD surface on a motorized arm mounted inside the

swing time is minimized to 6 sec to reduce the pile up of X
ray clusters

arm motion is synchronized with the CCD exposure

the simplest form of the rts2 command is

d C0
s ’for 100 { E 6 FEARM.!CURPOS+=8000 }’

measurements and analysis are automated

series of bias exposures (6s darks)

Fe data analysis outline

base line subtraction (
IEEE Transactions on Nuclear Science, Vol.57,
p. 2200
2204, August 2010, DOI:

bias exposures

evaluated from the
Fe image itself

out noise noise in a.d.u.

cluster finding algorithm to find X
ray hits

cluster seeds = pixels with amplitude above 5


seeds are ordered automatically in amplitude decrease order


n pixels zone around a seed is analyzed ( usually n = 3)


n zone is cleared

statistical analysis of X
ray clusters

gain determination

charge transfer efficiency


Fe express analysis: gain

The conversion gain is determined from fitting K

lines in cluster total amplitude

fit function is the sum of two Gaussians

line width = read
out noise + natural line width

initial parameter estimates and appropriate range are crucial for fit robustness

initial estimate of the K

peak position is done by finding the bin with the

maximum content in the X
ray amplitude distribution

device 106
07, T=

initial distribution

after correction

X direction, serial transfer CTE=0.999911

Y direction, parallel transfer CTE=0.999996

Fe express analysis: CTE

device 106
07, T=

Fe express analysis: CTE

X direction, serial transfer CTE=0.999911

CTE is calculated using least square regression method assuming
exponential amplitude dependence on coordinate.