Talk Roadmap

Paola Amico et al. - The 2007 ESO Instrument Calibration Workshop - January 12, 2007 Slide 1/35

The detector monitoring projectP. Amico

P. Ballester, W. Hummel, G. Lo Curto, L. Lundin, A. Modigliani, L. Vanzi (all @ the European Southern Observatory)


Talk Roadmap

• General Introduction to detector characterization and testing

• ESO’s detector roadmap: from vendors to science.

• Detector monitoring in La Silla• Detector monitoring in Paranal• The monitoring plan


Once upon a time there was the perfect scientific

detector:• Detect 100% of photons

• Large number of pixels

• Time tag for each photon

• Measure photon wavelength

• Measure photon polarization

Up to 99% QE

over 355 million pixels

No - framing detectors

No – defined by filter

No – defined by filter

PLUS READOUT NOISE (~2e-)!!

Detectors are now nearly perfectBut the devil is in the details ………


Just to name a few:Signal

• QE (wavelength, T, t)

• Fringing• Linearity, well

depth (T,bias, flux!)

• Gain precision and stability

• Persistence• Crosstalk• Charge Transfer

Efficiency• Stray light

Noise

• Bias Level• Dark current• Mux glow• Cosmic rays• Read noise• Microphonic noise• Pickup noise• Odd-even column

effect

Many wonderful ways to be defective

Many behaviours to measure

985 nm 990 nm 995 nm

1000 nm 1005 nm 1010 nm

Fringing of Silicon sensor


“Persistence” in Hawaii-2RG

• switch from LPE to MBE does not eliminate persistence

• latent image can be seen for many hours

• persistence on all arrays tested

Courtesy G. Finger

QuickTime™ and a decompressor

are needed to see this picture.


“Persistence” in Hawaii 2RG

• depends on fluence not on flux

• N < Nsaturation = 105eno persistence (?)

• switch from LPE to MBE does not eliminate persistence

• latent image can be seen for many hours

• Causes: traps, temperature effects?

Courtesy G. Finger


QC

Once upon a time there was a detector…

Vendors

ODT IRI

Characterization

HQ

1.

Monitoring

Ins+SciOps

LPO

Ins+SciOps

Paranal

La Silla

2.


1. In-house characterization• CCD pedigree (type, specs, output, pixel size)

• RON and Conversion factor (inverse gain, CONAD)

• Quantum Efficiency (QE) • Cosmetic defects• Linearity (peak-to-peak, rms, range, sat. level)

• Dark Current• Cosmic X-ray events• Charge Transfer efficiency• Measurements conditions (T, pre-scan/overscan, bias voltages, date) and sample images.

• Additional tests: IR - sensitivity profile of pixel.


In-house characterization (cont’d)

Courtesy G. Finger


Plasma cleaning devicesPlasma cleaning devicesSuccessful plasma cleaning inside the assembled OmegaCam cryostat

After some minutes the large volume (approx. 200 l) of the cryostat was filled by a violet glowing plasma. A normal cleaning process takes not longer than 10 minutes!

Courtesy S. Deiries


Comparison of cosmetic quality 40K / 80 K

Cut levels -250 e /200 e , DIT 900 sec

T=40 K T=80 K

Courtesy G. Finger


2. La Silla: 18 CCD detectors

INSTRUMENT N type pixels

Speedkps

RON (e-)

Gain-1

(e-/adu)

FEROS 1 EEV 2kx4k

225 4.8 3.0

WFI 8 EEV 2kx4k

200 4.5 2.0

EMMI B 1 TK1024 1kx1k

52 6.5 2.9

EMMI R 2 MIT/LL 2kx4k

225 5.3 1.3

SUSI2 2 EEV 2kx4k

200 4.7 2.3

CES 1 EEV 2kx4k

100 3.4 1.1

EFOSC2 1 LLes/UV flooded

2kx2k

58 8.4 1.2

HARPS 2 EEV 2kx4k

416 4.6 1.3


La Silla CCDs (optical)• Weekly linearity tests with beta light (only FEROS and HARPS with LEDs) performed by SciOps.

• Semi automatic tool with GUI to acquire, analyze, e-mail, publish data on the web (one click per action). Feedback is displayed on the screen

• Test results e-mailed to instrumentation engineer and instrument scientist.


La Silla CCD detectors

• Values measured: bias level, average count rate, gain, ron, linearity, shutter delay. Database spanning several years.

• We use the traditional Photon Transfer method to derive gain from photon statistics (5%)

• Method documented (among others) in Downing, 2006SPIE.6276E...8D:

• Linearity relays on beta light (or LED) stability to compute average count rates at different exposure times -> shutter stability required (with the exception of HARPS for which the LEDs are downstream of the shutter).


LED vs. Beta Light in FEROS

The HARPS CCD is not accessible; beta light tests are therefore not feasible.Similarly, for FEROS, insertion of the beta light is inconvenient, as it has an impact on the instrument stability.Instead LEDs, placed just above the CCDs are used.This technique is proven to be fully satisfactory using the FEROS LEDs (high stability power source, no feedback).

P. FrancoisP. FrancoisP. Francois


La Silla CCD detectors

• Uniform scheme across instruments for detector testing.

• The look & feel of the interface is the same.

• The end product is the same• There is a database spanning several years to be used for trending.


Detectors in ParanalOPTICAL DETECTORS

Instrument Detector Size Pixel [ m]

Format Wavelength range [nm]

FORS1 Tek 2048EB4-1 2048x2046 24 S 330-1100 FORS2 MIT/LL CCID-20 2x2048x4096 15 M of 2 330-1100 FLAMES (Giraffe) E2V 44-82 2048x4102 15 S 370-950 UVES Blue E2V 44-82 2048x4102 15 S 300-500 UVES Red E2V 44-82

MIT/LL CCID-20 2048x4102 15 M of 2 420-1100

VIMOS E2V 44-82 2048x2440 15 M of 4 360-1100 NAOS OPT E2V CCD50 128x128 24 S 450-1100

Total of 25 detectors. 44 more expected

INFRARED DETECTORS Instrument Detector Size Pixel

[ m] Format Wavelength

range [ m] ISAAC Rockwell Hawaii HgCdTe 1024x1024 18 S 0.9-2.5 SBRC Aladdin 1024x1024 27 S 1-5 VISIR DRS BIB Si:As 256x256 50 S 5-25 CONICA SBRC Aladdin 3 InSb 1024x1024 27 S 1-5 NAOS Rockwell Hawaii HgCdTe 1024x1024 18 S 0.9-2.5 SPIFFI Rockwell Hawaii HgCdTe 2040x2048 18 S 0.9-2.5 MIDI Raytheon IBC SI:As 320x240 50 S 5.0-25 AMBER Rockwell Hawaii HgCdTe 1024x1024 18 S 0.9-2.5 CRIRES science

Raytheon Aladdin III 4x1024x1024

27 M of 4 0.95-5.2

CRIRES SV Raytheon Aladdin III 1024x1024 27 S 1-2.2 (5)


Quality checks in Paranal: CCDs


Optical detector monitoring:RON for FORS2


Optical detector monitoring:CONAD and linearity for GIRAFFE


Quality checks in Paranal: IR arrays


NIR detector monitoring:RON and dark level (ISAAC

Aladdin)


MIR detector monitoring: RON of VISIR


What can be improved?• Basic parameters (BIAS, RON) are measured and

monitored daily.• Fundamental parameters (gain) still not

implemented for all instrument.• Other measurements (e.g. linearity, fringing,

contamination, etc) not yet monitored.• Measurements of the same quantity are made

differently for different instruments (different pipelines).– “measured in a raw file (high gain mode) as

100x100 pixels sigma, corrected for fixed-pattern contribution”

– “measured on single raw frames, with no corrections”

• La Silla and Paranal independent from each other (which is partly compatible with the different operational scheme).


The detector monitoring plan• A true Chile-Germany interdepartmental collaboration:

Sciops & Instrumentation (Paranal, La Silla) + QC + Data Flow System Dept. (DFS), +…

• Started as an IR detector monitoring project for Paranal.• Its scope was widened last year, to include optical

detectors.• Once the group of “volunteers” was formed, the first step

had been to collect all the documentation available and condense it in a document. This document summarizes the goals and objectives of the plan, presents an extensive description of all (well, almost all) the tests and measurements ever conceived for detectors and attempts at describing the algorithms we believe are best to measure a certain quantity.

• It is not meant as a strict to-do-list to be completed, but as a driver for an ambitious project.

• It is open to a wide audience (instrument scientists, instrumentation engineers, astronomers, etc) and it welcomes the contributions and expertise of many.


Goals of the project• Standardize the test procedures whenever applicable (e.g. define a unique way to measure linearity).

• Unify test procedures for IR and Optical detectors whenever applicable and describe the differences in all other cases.

• Unify the measurements procedures and the use of data reduction recipes and algorithms.

• Utilize available resources, such as data taking templates, pipeline recipes, existing reporting tools (QC web pages, Autrep), previous experience (e.g. LaSilla test procedures, ODT, IRI, etc).


Specific GoalsPARAMETERS OPT NIR MIR ACCURACY [UNITS] M F P Amplifier Glow X X X Signal (ADU) Statistics on

Bias/Dark N 14

Bias Level X <1 (1 digit [ADUs])

Bias Frames Statistics

D 1

Bad pixels, Cosmetic Quality

X X X N/A Bias Frames (OPT) Dark Frames (IR) Flat frames (ALL) analysis

Y 2

Contamination X UV Flats analysis Y 16 Conversion Factor X X X 2 digits [e-/ADU] Transfer Curve

TDI images BA M

3

Cosmic Ray Sensitivity X Long darks statistics

N 17

Crosstalk (multiple ports) X X X 1/300,000 Bright star sequence

Y 8

CTE (V/H) X 10-6 EPER (overscan); Signal variance;

O 6

Dark Current X X X 2 digits [e-/px/s]–[e-/px/hr] Dark Frames Statistics

D 1

Dark Signal Nonuniformity variable and fixed (DSNU +FPN)

X <10% BA 5

Full Well Capacity X 1% Linearity curve BA 4 Linearity X X X <1% Transfer Function BA 3 Persistance (Remanence) X X X Dark current

measurements after illumination

Y 13

Readout Noise (RON) X X X 2 digits [e- RMS] Bias Frames Statistics (OPT) Dark frames Statistics (IR)

D 1

50Hz Pick up Noise X X 2 digits (ADU) FFT analysis Y/N

7

Microphonic Noise X BA 8 N-pixel correlated Noise X M 9 Odd-Even column effect X D 2 Shutter Pattern/Error X Linearity curve

Shutter procedure Y/N

11

Spatial Uniformity, Photo Response non-Uniformity (PRNU) - Fringing

X X X <2% Analysis of flats M/N

10

Stability X X N/A Visual inspection, bias frames

D 15

Stray light X Flat frames(filters) W 18 Temperature stabilization X Y 12


More Specific Goals

Parameter Importance Operating conditions Amplifier Glow Low – it is becoming rarer as array

design improves for Optical detectors. It can be large for IR (e.g. CRIRES).

Independent of temperature Proportional to integration time Non-uniform (apparent only for those pixels close to the output amplifier).

Conversion Factor (Gain, amplifier response)

High – affects absolute photometric measurements

Amplifier gain and offsets can be affected by temperature and applied voltages.

Cosmic ray effects Low- important for long exposure, therefore affects observing strategies

Location (altitude) and surrounding materials. Both cannot be changed. Usually stable.

Crosstalk

Medium – affects spatial resolution Important at design level but usually stable. Independent of pixel location, usually scales with signal.

Charge Transfer Efficiency in CCDs Medium – Photometric accuracy can be affected of CTE is poor

Important at design level, but usually stable. Modern CCDs exceed “5 9s”.

Dark current and DSNU High – affects photometric accuracy

Varies strongly with temperature Can change in time Changes with exposure time (IR)

Non-linearity High – affects photometric accuracy Usually stable, but needs to be characterized accurately.

Pick-up noise Low Depends on DITs, time, readout speed

Pixel Response Non-Uniformity, Fringing

Medium – affects photometry and can affect wavelength calibration accuracy in spectrometers

Varies with wavelength (fringing) and can depend on illumination conditions.

Quantum efficiency High – affects absolute photometric accuracy

Depends on temperature (IR/OPT) Affected by contamination

Read out noise High – affects all aspects of astronomical observations. Good indicator of instrument’s electronics problems

Can change with temperature, may be influenced by external electromagnetic interference.


Responsibilities & status

Definition of the plan

Calibration group+ La Silla+Instrumentation (LPO & HQ)+…

Preparation, implementation and maintenance

ISs+Software group (Paranal)

Execution SciOps (Astronomers + TIOs)

Development and maintenance of pipeline recipes

Software Development Division (Gar) +DHA (LPO)

Analysis QC Gar + SciOps

Monitoring of results ISs (LPO)


CPL Detector Monitoring Functions

- Study of operational pipelines-RON, BIAS/DARK supported for most instruments-First development required is for a common detector linearity recipe.

- CPL Implementation-Algorithms distributed as CPL data reduction functions-Pipeline recipes invoke a single function

- Pipelines Implementation-Instrument templates may have to be retrofitted


ReferencesNIR/MIR Spectro-Imagers:ISAAC/CONICA/SINFONI/VISIR

Today @15:20

E. Mason

Integral Field Spectroscopy: VIMOS/FLAMES/SINFONI

Thu @09:00

P. Amico

Calibrations of VLTI Instruments: AMBER/MIDI

Fri @09:00

S. Morel

Six years of WFI Operations

Fri @13:30

F. Selman

Questions?

Talk Roadmap

Documents

Transcript of Talk Roadmap