About
the DCL - Test Procedures
July
12, 1999 CCD Test Procedure Summary - Draft 1
(view PDF
file)
Introduction
This
is a brief outline of the testing of CCD detectors at the
Detector Characterization Laboratory. (A detailed, 'working'
Test Procedure is also available.) This procedure describes
the approach for a typical set of tests: Read Noise, Dark
Current, Charge Transfer Efficiency, Gain, Linearity, Full
Well, Output Node Sensitivity, Quantum Efficiency, Point
Spread Function, and Uniformity. Note that this document
does not cover all testing activities at the DCL. Other
tests may be run based on results of any of the above tests,
or with special requirements.
1.
Test Setup Characterization
The
Test Electronics are characterized for a given readout frequency
to account for possible design deviations. The test signal
is injected at the CCD to electronics interface. Characterization
includes measurement of gain (in uV/LSB), noise bandwidth,
linearity and pulse response.
2.
CCD Read Noise
CCD
Read Noise is measured by shifting the CCD charge away from
a given amplifier. This allows for the exclusion of other
sources (e.g., dark current, charge injection, etc.) from
the data. Three full frames are acquired for a given readout
frequency; each frame is corrected for baseline drift. An
FFT analysis is performed to look for interfering noise
sources. Noise is calculated as a mean standard deviation
for all three frames.
Noise
is also measured in the classical way, by reading the CCD
image using a short exposure time. Two frames are acquired
and subtracted from each other to remove image irregularities.
The difference between those two results is indicative of
the noise component due to clocking, dark, cosmics and other
sources.
Noise
is measured within 10 kHz to 100 kHz frequency range at
-80C and -90 C.
3.
Dark Current
Dark
Current is measured at -80C and -90C. At this temperature
range, dark rate may be as low as 2 electrons per hour.
A very long exposure is required for an accurate measurement
of dark current. Binning in the serial register is used
(where possible), to reduce exposure time. At least three
different exposure lengths are used for a given temperature.
Three frames are taken for each exposure time. The frames
are corrected for baseline drift and cosmic events. The
dark current is calculated as an average of all frames.
In
addition to average dark current, dark nonuniformity, histograms
and a map of 'hot pixels' are computed.
4.
CTE Gain and Sensitivity using an Fe55 source
An
Fe55 X-ray source is used to determine the absolute Charge
Transfer Efficiency (CTE) and system Gain. System Gain and
electronics Gain are used to calculate CCD Output Node Sensitivity
(Sv). Measurements are performed at -80C and -90C. Three
frames are taken for each temperature. Each frame is baseline
corrected if necessary. Standard methodology is used to
extract serial and vertical CTE and Sv from each frame.
An average of values from the three frames is used.
5.
CTE, Gain, Sensitivity, Linearity and Full Well using flat
field illumination
CTE,
Gain, Linearity and Full Well are determined from series
of flat field frames. The primary objective here is to characterize
CTE versus signal intensity. The same data set is used to
calculate linearity, full well, and system gain.
Measurement
are made at -80C and -90C. At a given temperature, a set
of exposure times is designed to span the expected dynamic
range of the CCD. Data acquisition is set to overscan in
both vertical and horizontal directions. For each exposure
time at least two frames are acquired. Illumination is monitored,
its value is averaged and recorded for each frame and used
for data correction.
CTE
is calculated from the amount of charge trailing the last
pixel (EPER method). Gain is calculated using the Photon
Transfer Curve method. Linearity is determined by fitting
a straight line to the mean frame value and computing the
relative deviation. Full Well is determined from the Photon
Transfer Curve as a point where the shot noise curve departs
from the theoretical slope. Sensitivity is determined from
the system Gain and electronics Gain measured at section
1.
6.
Quantum Efficiency and Uniformity
Quantum
Efficiency (QE) is measured by comparing the CCD readings
with that of a calibrated, NIST traceable photodiode. The
detector focal plane is first measured for flatness and
intensity by scanning with a reference photodiode. After
that, the CCD is moved into the focal plane position and
three frames are taken for each temperature: -80C and -90C.
Each frame is normalized to the reference diode readings.
QE is calculated as an average for all three frames. Those
normalized frames are used to compute the uniformity of
QE.
QE
measurements are performed with monochromatic illumination,
with an approximate bandwidth of 10 nm. The spectral range
is 200 nm to 1100 nm. The range of 200 nm to 400 nm is sampled
at 25nm intervals; the range 400 nm to 1100 nm is sampled
at 50 nm intervals.
7.
Quantum Efficiency stability
This
test verifies how much charge trapping takes place on the
CCD back surface (due to backside thinning process). The
CCD is maintained at dark, and flushed before being exposed
to series of rapid, flat field exposures; no flushing takes
place between exposures. Relative responses are plotted
as a function of time. The results are computed by comparing
the values in the first frame with those in the plateau,
which is established in the last few frames.
8.
Point Spread Function
The
Point Spread Function (PSF) is measured by scanning a pinhole
in the object plane of the optical relay (Offner system),
projecting its image in the focal plane and collecting frames
for each position of the pinhole. The pinhole image is diffraction
limited, and approximately 6 um in diameter.
The
pinhole position is first adjusted so the image is centered
on a selected pixel. Then, the pinhole is scanned in the
x-y direction, +,- one step in each direction, in 5 um increments.
Data is acquired for each step. PSF is calculated by fitting
a Gaussian peak to the data. This test is performed with
a monochromatic light (approx. 10 nm bandwidth) for UV (200
nm to 400nm), VIS and IR.
8a.
Pixel Uniformity Test
This
test is to show how an object image (star) will respond
to 'spatial dithering', due to possible nonuniformity of
response within the CCD pixel area. For this test, the optical
system is modified to have an f# similar to the application
optics. A pinhole image is projected on the CCD focal plane
and the pinhole is scanned in 5 um (?) increments over a
50 x 50 um square (?). This test is performed for a few
selected locations over the detector area using monochromatic
light.
