Upload
brinly
View
54
Download
0
Embed Size (px)
DESCRIPTION
Optimisation of an EMCCD. Reduction of parallel clock induced charge (CIC). Investigation of dark current and the effect of “Dither”. Measurement and reduction of serial register CIC. Some astronomical results. (Application of Dither to a large format CCD). SDW Munich 2009. - PowerPoint PPT Presentation
Citation preview
Optimisation of an EMCCD
SDW Munich 2009 www.qucam.com
- Reduction of parallel clock induced charge (CIC).
- Investigation of dark current and the effect of “Dither”.
- Measurement and reduction of serial register CIC.
- Some astronomical results.
-(Application of Dither to a large format CCD).
Cuts through EMCCD bias frames
SDW Munich 2009 www.qucam.com
Clock induced charge the dominant noise source.
Opt
imis
atio
n pr
oces
s
EMCCD primer
SDW Munich 2009 www.qucam.com
H
1
H
2D
C
H
2H
V
H
1
H
3
H
2
H
3
H
1
H
2
H
3
H
2D
C
H
2H
V
H
1
H
3
Conventional part of register EM part of register
t1
t2
t3
EMCCD primer
SDW Munich 2009 www.qucam.com
EM register Link section Serial register Conventional Amp.EM Amp.
Image Area.
Store Area.
604 elements 468 elements
1056 columns
16 elements 16 elements
E2V CCD201
CIC produced in all sections of the CCD.
Multiplication noise
SDW Munich 2009 www.qucam.com
Output of EM register in response to inputs between 1 and 5 e-.
Note that an output signal of 400 e-
could result from an input of either4 or 5 e- with almost equal probability
-> “Multiplication noise”
www.qucam.com
Multiplication noise: Monte Carlo model
SDW Munich 2009
Overall EM gain=1000
Average
Passage of 10 seperate photo-electronsare followed through the EM register.
Inverted Mode
SDW Munich 2009 www.qucam.com
Inverted mode operation reduces dark current
SDW Munich 2009 www.qucam.com
E2V CCD201, T=293K
Holes are attracted from the channel stops.These then populate the surface of theCCD mopping up surface dark current.
Inverted mode operation increases CIC
SDW Munich 2009 www.qucam.com
P+ P+n
p
+4V
Electron producedby impact ionisation
P+ P+SiO2
Electrode
n
p Pixel charge packet
-8V
Channelstop
During integrationsurface populatedwith holes that suppress surfacedark current.
During charge transfer whenthe pixel comesout of inversionthe holes produceclock induced charge.
Measuring parallel CIC in an FT CCD
SDW Munich 2009 www.qucam.com
103
7 ro
ws
103
2 ro
ws
ImageStore
Last row of imagewill contain 2069rows of CIC.
First row of paralleloverscan will contain only 1038 rows of CIC.
Integration Transfer Readout
So the parallel CIC should show a step in vertical cuts through bias frames that include a parallel overscan area.
Next frameintegrating
CCD201
Non-inverted mode reduces parallel CIC
SDW Munich 2009 www.qucam.com
Inverted operation Non-inverted operation
Cuts through bias images that contain a parallel overscan.
CIC from 1032row transfers
Summary
SDW Munich 2009 www.qucam.com
Inverted Mode Non Inverted Mode
Low Dark current
Huge CIC
High Dark current
Low CIC
But…..dark current non-linear with time!
SDW Munich 2009 www.qucam.com
CCD201 data
Non-inverted dark current suddenly drops by a factor of almost 100!
CCD201 data
Non-inverted dark current versus exposure time
SDW Munich 2009 www.qucam.com
CCD201 data
Non-Inverted mode conclusions
SDW Munich 2009 www.qucam.com
Non inverted mode required for low parallel CIC.
For short exposures the corresponding increase in dark current is not seen.
Non-inverted mode operation preferred for EMCCDs
P+ P+SiO2
Electrode
n
p Pixel charge packet
-8V
Channelstop
The suppression of dark current could be explained by `Dither` (Jorden et al. `Secrets of E2V Technologies CCDs` SDW 2004). However, this explanationrequires the presence of holes at the surface. The low CIC seems to indicate the very opposite (??).
Measurement of serial-clock generated CIC
SDW Munich 2009 www.qucam.com
Removed by DG operation
CIC left behind by previous line readout
CIC from current line readout
Sum of the two
The CCD201 contains a dump gate (DG)structure to assist in rapid clearing. It canalso be used to measure serial generated CIC.
Reduction of serial clock generated CIC
SDW Munich 2009 www.qucam.com
New image dimensionsfor purposes of test.
Complete readout of pipeline for every row of image.
Measurement of serial-clock generated CIC
SDW Munich 2009 www.qucam.com
EM register Link section Serial registerEM Amp.
Image Area.
Store Area.
DG
Reduction of serial-register CIC
SDW Munich 2009 www.qucam.com
Reducing the serial high clock voltage from 10 to 8.5V reduced serial CIC .Lower voltages gave poor CTE.
Final CIC levels
SDW Munich 2009 www.qucam.com
Model used to find relative proportions of pre-EM-register and in-EM-register generated CIC events.
Pre-reg = 0.02e- per pixel , In-reg = 0.011e- per pixel. Total 0.013e- per pixel.
Dual-EMCCD spectroscopy systemon William Herschel Telescope La Palma
SDW Munich 2009 www.qucam.com
Red arm of ISIS spectrograph: CCD201 Blue arm : additional CCD201
SDW Munich 2009 www.qucam.com
CCD201 cryogenic EMCCD camera
SDW Munich 2009 www.qucam.com
Cataclysmic Variable: white dwarf and less massive donor orbiting around their common centre of gravity. Orbitalperiods from 5 minutes to > 12 hours. Most of the light is emitted from an accretion disc surrounding thewhite dwarf.
Spectrographic observations show the double emission lines produced by the highvelocity material orbiting within the accretion disc.
EMCCD spectroscopy: astronomical results
Artists impression, Mark GarlickSDSSJ1433
Appearance of an EMCCD spectrum
SDW Munich 2009 www.qucam.com
Targetg´=18.5
Reference
0.22A per pixel dispersion.Mean intensity of continuum=0.08e-/s per wavelength stepExposure time=30s
SDSSJ1433
EMCCD spectroscopy: astronomical results
SDW Munich 2009 www.qucam.com
With an EMCCD we can use short exposures to obtain time resolved spectra of the accretion disc. It is then possible tomeasure the to-and-fro motion of the white dwarf and constrain the mass of the secondary star.
Radial velocity of the white dwarf.
Ser
ies
of t
ime
reso
lved
spe
ctra
Tulloch, Rodriguez-Gil, Dhillon, MNRAS 397, L82-86, 2009
EMCCD spectroscopy: astronomical results
SDW Munich 2009 www.qucam.com
Actual EMCCD spectrumModel spectrum: 3e- read noise
This type of time-resolved high-dispersion spectroscopy would have been impossible with a conventional detector.
SDW Munich 2009 www.qucam.com
2k x 6k pixelframe-transfer CCD42C0.Conventional invertedmode operation
Intended for Eddington.Now destined for Mercatortelescope in La Palma
High-speed photometry,short exposure time. Use of Peltier cooler.
Aside: Dither clocking in a large format CCD
´Dither´ induced cosmetic defects
SDW Munich 2009 www.qucam.com
V1 V2 V3 “Wobble” sequence repeated at intervals ranging from 1ms to 4sat temperatures from213 to 233K during exposure.
Flat field
+3V
-9V
Use of ´Dither´ with a CCD42CO
SDW Munich 2009 www.qucam.com
Profile through 6 defects after 10,000 dither clock cycles.Charge is conserved, defect amplitude < 1e- / cycle.
T=220K
Use of ´Dither´ with a CCD42CO
SDW Munich 2009 www.qucam.com
Use of ´Dither´ with a CCD42CO
SDW Munich 2009 www.qucam.com
Approximately equal to an extra 10 degrees of cooling.
Optimisation of an EMCCD
SDW Munich 2009 www.qucam.com
End of presentation
www.qucam.com