First results of the tests campaignFirst results of the tests campaign in VISIBLE in VISIBLE

for the demonstratorfor the demonstrator

12 October 2007SNAP Collaboration MeetingParis

Marie-Hélène Aumeunier

C.Cerna, A.Ealet, E.Prieto

Demonstrator ObjectivesDemonstrator Objectives

GoalGoal: Validate the performances slicer concept and test the calibration procedure

Straylight Measurement controlled at 10-3

Wavelength Calibration at the nanometer level

Flux Calibration better than 1 %

The demonstrator optical bench:The demonstrator optical bench:

o Source: Halogen lamp + monochromator + optical fibero Steering mirror: to scan the FoV by step < 1/100 of a pixelo Demonstrator: the same characteristics as the SNAP spectrograph in board (low spectral resolution + under sampled IR)o Detectors: Camera Apogee in visible

Rockwell HgCdTe in IR

Steering Mirror

Slicer Imager

DetectorOffner

PrismExit of optical

Fiber

PLANPLAN

1- Demonstrator PSFs vs simulation

2- Wavelength calibration of emission lines

3- Flux calibration of emission lines and QTH spectrum

PSFs MeasurementPSFs Measurement

x,y λ x,y λ

simulation Measures

x,yλλ

λ=500nm

x,yλ

Double verification:

Measure of PSF shape

Measure the optical losses

λ=900nm

PSF within the central slice when the point source hits the slicer

center

At the input simulation, we need:- λ (provided by monochromator)- (x,y) into the slice (given by steering mirror)- position of the initial pixel on the detector plane (fitted manually)

Objectives

The demonstrator: an opportunity to test the simulation with real data

Check the alignment of the complete instrument at ≠ (x,y,λ)

Check the diffraction effect by the slice edge

Method

PSFs MeasurementPSFs Measurement PSF shape WRT λ

Wavelength (nm)

Sp

ati

al FW

HM

(p

ixels

)

Sp

ectr

al FW

HM

(p

ixels

)

grating 1

grating 2

Lines simulated as Dirac

In the spatial direction: maximum ∆FWHM sim / exp ~ ½ pixels at the best focus

In the spectral direction: Emission lines have a broad width Gap for spectral FWHM at 600 nm due to the change of

monochromator’s grating

Ideal focus+ Perfect detector

Best focus adjusted

Wavelength (nm)

PSFs MeasurementPSFs Measurement Optical Losses

y-coord within the slice (slice unit)

Effi

cie

ncy

y-coord within the slice (slice unit)

Effi

cie

ncy

900 nm

500 nmFlux losses at the slice edges

Wavelength (nm)

Op

tical lo

sses (

%)

At λ>600 nm : optical losses predicted with a precision better than 2 %

At λ < 600 nm: the highest flux losses the PSF is finer, the diffraction by the slice edges are more important

Current Work: implement in simulation « dead area » between 2 slices responsible of higher optical losses

slice 3 slice 4slice 2

slice 3 slice 4slice 2

PSFs MeasurementPSFs Measurement

Good agreement of data with the simulation:

No severe default detected from one slice to another:

PSF shape homogenous along the slicer width

Optical losses at the slice edges checked better than 2

% for λ > 600nm

Conclusion

Wavelength CalibrationWavelength Calibration Adjustment of dispersion curves(1)

Sources:

- emission lines

- spatial extended source

y1

y4

y2

y3

y5

λ1

λ2

λ3

λ4

λ5

pixelx yf

y-coord (pixels)

λ1

λ2

λ3

λ4

λ5

y1 y4y2 y3 y5

Spectrum on detector

Adjustment of the dispersion curve fx at x-coord given

Extract the lines center

(barycenter)

Halogen lamp +

monochromator

∑ images done by scanning the slicer width with the steering

mirror

Wavelength CalibrationWavelength Calibration Adjustment of dispersion curves(2)

Dispersion curves at the slice center VS SIMULATION

Dispersion curves in agreement with simulation at 95 %

Wavelength CalibrationWavelength Calibration Calibration Procedure of punctual emission lines

1. Compute the wavelength using the dispersion curves of each slice

2. Correct slit effect for punctual sources

3. Calibration error = λtrue - λfitted

pixelxfitted yf with ypixel = center of lines on the detector

Slit Effect (definition)

When the point source scans the slicer width (i.e spectrograph slit), the lines center moves also on the detector

Causes calibration error between the real and fitted λ

Visible pixel

λ Dispersion

Correction of slit effect

Mean of 5 slices: average the spectrum of each slice weighted by the flux into the slice

Spatial dithering: average the images done when the point source position is shifted of a random value (Normal distribution of RMS 1/5 slices)

Procedure steps

Wavelength CalibrationWavelength Calibration Calibration of emission linesdem

onst

rato

rsi

mula

tion

613 nm

612 nm

835 nm

823 nm

405 nm

485 nm

Data in agreement with the result predicted by simulation Slit effect corrected with mean of 5 slices No need of spatial dithering Offset (1 nm) for the high wavelength current work

IN SPEC

14.4Si nm 1.44Si µm

Estimation of Silicium line Center

Error calibration of Si line < 1 nm using the mean of 5 slices

Stars Calibration (simulation) Stars Calibration (simulation)

Error on Redshift from galaxy emission lines

Hα

0.87H

syst nm

1.5H H fittedz z

0.002 5 /1000z

Galaxy spectrum simulated in IR arm at z=1.5

Wavelength CalibrationWavelength Calibration Conclusion

Wavelength calibration possible with classical procedure (despite of low spectral resolution + undersampled)

Calibration error < 1 nm using the information within the 5 slices (mean of 5 slices)

Perspectives: Fly calibration

find adequate calibration lamp: it is difficult to use blended lines

Flux CalibrationFlux Calibration Objectives

Goal:

Test calibration procedure able to find the initial flux source better than 1 %

%12222 otherncalibratioSTATflux

< 0.33 %Which means ?

Calibrate all effects that damages the source flux better than 1/3 %:

Optical losses: diffraction, coating,… Detector: quantum efficiency, pixel response (fringing for

CCD, intra-pixel sensitivity variation, CTE for IR detector, etc)

Difficult ! AND the PSF is under-sampled in the IR range and so sensitive to intra-pixel variation

Flux CalibrationFlux Calibration Method

Telescope Focal Plane (slicer)

1/10 of a pixel

slice 1

slice 0

slice -1

A library of reference images to characterize the spectrograph response for any (x,y,λ)

Done from reference source (point source) with the flux known better than 1 %

Reference images

Reference source Φref

Source Φobj at unknown position

ref

objk

χ2 minimization per pixel

Linear Interpolation of flux per slice m

A method to calibrate the object flux at unknown position from the library

Find the « nearest » reference images

Interpolate the object image from the selected images to deduce the ratio k

Flux Calibration of emission linesFlux Calibration of emission lines

Calibration of 300 images (monoλ) at 700 nm

No dithering Spatial dithering

k-theoretical = Φobjet/Φref = 1

k adjusted from 2 references images

Mean error= 0.22 %Std of mean = 0.39 %

Mean error= 0.17 %Std of mean = 0.14%

Results at 450, 500, 700 and 900 nm

At 700 nm, flux error << 1% (95% CL)

Dispersion error improved with spatial dithering (4 images)

Precision of k-theoritical to improve for best measurement

(95% CL)

Result

Flux Calibration of spectrumFlux Calibration of spectrum Method

First step: Select the reference images

λ

x,y

slice 2 slice 3 slice 4

Mask @ 700 nm

Extract a region of image associated to a narrow band of spectrum

λ

slice 2 slice 3 slice 4

x,y

Flux Calibration of spectrumFlux Calibration of spectrum Method

First step: Select the reference images

Extract a region of image associated to a narrow band of spectrum

λ

x,y

slice 2 slice 3 slice 4

Minimization method applied to

“monochromatic” image

Flux Calibration of SpectrumFlux Calibration of Spectrum

j

iij

First step: Select the reference images

1 –Minimization per pixels

2 –Minimization on ratio of slice flux

Sensitive to the position within the slice

Object Point

Reference Points

Selected Points

1/20 of a slice = 20 arc sec

1/20 of a slice

one slice

the same minimization as the one developed for the mission line

Method

Flux calibration of spectrumFlux calibration of spectrum

Second step: Find the ratio k(λ, ∆λ) between the reference spectra and the spectrum to calibrate

Method

Linear interpolation of the flux captured in each slice by narrow band

Flux Calibration of QTH spectrumFlux Calibration of QTH spectrum

Wavelength (nm)

Flu

x E

rror

(%)

SN

R

k-theoretical = Φobjet/Φref = 1

k adjusted from 2 references images

<S/N>=45μ=0.44 %σ=2.72%

<S/N>=204μ=0.15 %σ=0.36 %

<S/N>=204μ=0.06 %σ=0.50 %

Tests conditions zone 1[450-620 nm]

zone 2[621-830 nm]

zone 2[831-950 nm]

Result

Flux Calibration of QTH spectrumFlux Calibration of QTH spectrum

Wavelength (nm)

Flu

x E

rror

(%)

SN

R

With spatial dithering <S/N>=45μ=0.18 %σ=1.41%

<S/N>=204μ=0.02 %σ=0.15 %

<S/N>=204μ=0.05 %σ=0.24 %

Result

Flux Calibration of QTH spectrumFlux Calibration of QTH spectrum

95 % CL

Calibration error WRT SNR

No degradation when the point

source moves within the slice 4 images dithered enough to

improve calibration Error Flux < 1% when SNR > 100

Calibration error WRT y

Calibration error WRT spatial dithering

Results

Flux CalibrationFlux Calibration

Slicer technology helps the flux calibration

Move the point source of a 1/ 10 of a slice

Conclusion

Fly calibration

Even moving by small steps in the FoV (1/10 pixel), the detector image changes completely :

The PSF is cut differently by the slicer The sampling of each PSF’s part is also different

• Adapt the method to calibrate secondary stars from fundamental starsReference library made with artificial lamps at ground

made with fundemantal stars in fly

Intensity spatial variation do not the same

Take into account the sky background and the variation of detector gain map

ConclusionConclusion

INFRARED campaign in coming Adapt calibration procedure in fly : find adquate calibration source,

Good agreement between PSFs measurements and simulation (shape and optical losses) Straylight in specification (<10-3) at 500 and 700 nm

Visible campaign Report:

Good results for wavelength calibration of emission lines: correction of slit effect works (mean of 5 slices) Accurate flux calibration (1 %) for emission lines + spectrum @ k=1

The optical performances of slicer are checked in visible

The calibration procedures have been tested and validated

Perspectives

SPARESSPARES

PSFs MeasurementPSFs Measurement PSF shape WRT y-coord

y-coord within the slice (slice unit)

Sp

ati

al FW

HM

(p

ixels

)

y-coord within the slice (slice unit)

Sp

ectr

al FW

HM

(p

ixels

)

Complete PSF

Uniform along the slice width

No diffraction effect

PSF cut by the slice

Sensitive to the position within the slice

Diffraction by the slice edge: PSF widening

PSF projected along the spatial direction :

PSF projected along the spectral direction :

Steering Mirror CalibrationSteering Mirror Calibration

y=3.00078

Flux total = ∑ 5 slices

slice 3slice 2 slice 5slice 1 slice 4

y1 3 4 52

x1 2 3 4 5

Method:- scan the slicer width with monochromatic

point source by step of 1/10 of a pixel- guess the slice edges around the curves

intersection of flux per slice

Goal: calibrate the relative position along the

slice width given by the steering mirror

Steering Calibration at 700 nm

Result:

Precision < 1/10 of slice width

Camera CalibrationCamera Calibration Fringing (1)

Definition: sensitivity variation of pixel as a function of λ due to interference among the incident and reflected beam within the CCD layers

Consequence

Input spectrum: Halogen lamp Spectrum on detector

Wavelength (nm)

(semi-log scale)

Wavelength (nm)

(semi-log scale)

Camera CalibrationCamera Calibration Fringing (2)

Fringing characterization:

- Fit the QTH spectrum on detector

- Compute the response per pixel i and per λ

∆λ/pixel < 5 nm 5 < ∆λ/pixel < 10 nm

Result:

Fringing responsible of flux

variations: up to 10 % for λ<700 nm up to 5 % for λ> 700 nm

Low spectral resolution smoothes the

flux variations at high wavelength

Pupil Mirror

Slit Mirror

slice 0slice -1

slice 1

SLIT 0SLIT -1 SLIT 1

CROSS-

TALK

Only the slice 0 well aligned with the optical axis is lighted on the pseudo-slit 0

In case of straylight, the light coming from the slices may be also seen by the pseudo-slit 0

Ghost Images (CROSS-TALK )

slicer

Straylight

Straylight MeasurementStraylight Measurement Principle

Straylight MeasurementStraylight Measurement Experimental method

slice 1

slice 2

slice 3

slice 4

slice 5

PSF [slicer] PSF [spectrograph]

PSF at the spectrograph entrance saved into the slice

3

PSF into the slicer plane

Method: Sum N images of punctual monochromatic source to extract possible light coming from the slice 3 onto the slits 1 and 5

1

3

5slice 5

slice 1

slice 5

slice 1

3/3

1/31/3 F

FK

3/3

5/35/3 F

FK

slice 1

slice 43

5

2 3 4 5

1

5

1

5

1

3Slit 1Slit 2

Slit 3Slit 4

Slit 5PSF on the focal plane (log scale)

Flux ratio gives the cross-talk

Straylight MeasurementStraylight Measurement Results @ 500 nm

4

K3/1=7x10-4

(95 % CL)

5321

3

3

3

3

slit 1

slit 2

slit 3

slit 4

slit 5 As predicted by simulation, the PSF parts at the edges of the slit 2 and 4 spreads out of the slice image because of the spectrograph PSF convolution

K5/1=7.7x10-4

(95 % CL)

Straylight MeasurementStraylight Measurement Conclusion

Straylight controlled better than 10-3

Sraylight measurement limited by the detector noise

WavelengthK3 /1

(95 % CL)K3 /5

(95 % CL)

500 nm 8x10-4 9.5x10-4

700 nm 4.8x10-4 2.5x10-4