characterization of vanilla liverpool physics 30/1/2007 introduction

16
Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction Vanilla image pixels average, standard deviation (STD) and common mode (CM) noise depend on frame number and settings (use Digital Mode). Change Internal ADC Dark Ref. Change Row number (regional interest).

Upload: reuben

Post on 24-Jan-2016

28 views

Category:

Documents


1 download

DESCRIPTION

Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction Vanilla image pixels average, standard deviation (STD) and common mode (CM) noise depend on frame number and settings (use Digital Mode). Change Internal ADC Dark Ref. Change Row number (regional interest). - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Characterization of Vanilla

Liverpool Physics30/1/2007

Introduction

Vanilla image pixels average, standard deviation(STD) and common mode (CM) noise depend on frame number and settings (use Digital Mode).

• Change Internal ADC Dark Ref.• Change Row number (regional interest).• Change ADC Clock Period.• Under light illumination.

Page 2: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Output change with ADC dark referenceDigital model, device in dark and use default settings.Change ADC dark reference.Mean value increase with ADC ref. and frame number.STD value increase with frame No., but decrease with ADC ref.

Fig. 1 Out average and STD change with frame number.

0 10 20 30

130

140

150 Change Internal ADC dark reference2800--3200

VtestF1.org

Frame number

STD value of each frame

2800 2900 3000 3100 3200

Ou

t p

ut

( A

DU

)

0 10 20 300

200

400

600

Change Internal ADC dark reference2800--3200

VtestF1.org

Frame number

Mean value of each frame

2800 2900 3000 3100 3200

Ou

t p

ut

( A

DU

)

Page 3: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Output change with row numberDigital model, device in dark and use default parameters.Change row number from 520 to 100.Mean value and the slope decrease with row No.STD value and the slope decrease with row No.

Fig. 2 Out average and STD change with row number.

0 10 20 30370

380

390

400

410

linear increase

initial drop

Change row number520--100

VtestF4.org

Frame number

Mean value of each frame

3000 400R 300R 200R 100R

Out

put

( A

DU

)

0 10 20 30132

134

136

138

140

142

Change row number520--100

VtestF4.org

Frame number

STD value of each frame

520 400 300 200 100

Out

put

( A

DU

)

Page 4: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Fig. 3 Average output change with frame number.

Average output shift effectFull size image (520 rows) almost linearly increased with increasing

frame number (ADC Dark Ref.=3000 mV).Small size image (10 rows) shows non-linear transient depend on

frame number due to heating effect.Initial drop could be due to reset effect.

0 5 10 15 20375

380

385

390

395

400

405

Frame number

Out

put (

DN

)

testH27

520 rowsMean output402.896

0 500 1000 1500 2000360

370

380

390

400

410

420

430

440

Frame number

Out

put (

DN

)

testH32

10 rowsMean output436.185

Page 5: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Average output shift effectADC Dark Ref. selected for 2800~3200 mV. Recommend for 3000 mV. MEAN<0, when <2750 mV.For full size image (520 rows), the mean value shifts 1.25 DN/frame. This leads to extra CM mode noise with increasing frame number.

Fig. 4 Average output and slope change with ADC Dark Ref.

0 100 200 300 400 500

0.4

0.6

0.8

1.0

1.2

Slope change(with default settings,17 ADC clock)

Change row number520--100

VtestF4.org

Row number

Slope value shift

C

Slo

pe (

DN

/fram

e )

2800 3000 32000

200

400

600

ADC Dark Ref ( mV )

Dark baseline shift(with default settings,17 ADC clock)

Change Internal ADC Dark Ref.

VtestF4.orgMean value shift

B

Mea

n ou

tput

( D

N )

Page 6: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Average common mode noise shift effectIn the dark condition.The noise not very sensitive to ADC Dark Ref.The noise increases with increasing row number (integration time).

Fig. 5 Average common noise change with ADC Dark Ref.

2800 3000 32002.0

2.1

2.2

2.3

2.4

ADC Dark Ref ( mV )

Dark baseline shift(with default settings,17 ADC clock)

Change Internal ADC Dark Ref.

VtestF4.orgNoise value shift

Noise

Noi

se (

DN

)

0 100 200 300 400 500

1.4

1.6

1.8

2.0

2.2

Mean noise change(with default settings,17 ADC clock)

Change row number520--100

VtestF4.org

Row number

Noise value shift

Noise

Noi

se (

DN

)

Page 7: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Change Internal ADC clock periodIn dark condition.Average output has an initial drop, then almost linearly increase

with frame number.Both output and STD increase with increasing ADC clock period

(integration time).

Fig. 6 Output and STD change with frame number.

0 5 10 15 20

380

400

420

440

linear increase

initial drop

Change Internal ADC clock period12--45

VtestH3.org

Frame number

Mean value of each frame

12 15 20 25 30 35 40 45O

ut p

ut (

DN

)

0 5 10 15 20

135

140

145Change Internal ADC clock period12--45

VtestH3.org

Frame number

STD value of each frame

12 15 20 25 30 35 40 45

STD

( D

N

2 )

Page 8: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Change Internal ADC clock periodIn dark condition.Pixel count cut off at ~300 DN. FWHM ~180 DN. FWHM increases

with Internal ADC clock period.The slope of out put to frame number linearly increases with

increasing Internal ADC clock period (integration time).

Fig. 7 Mean and slope depend on Internal ADC clock periods.

200 400 600 800

0

10000

20000

FWHM~180 DN

Cut off~300 DN

Change Internal ADC clock period12--45

VtestH3.org

Output ( DN )

Mean value of each frame

12 15 20 25 30 35 40 45

Pix

el c

ount

0 10 20 30 400

1

2

3

slope=0.125+0.0698*clock

0.125

0.0698 DN/frame/period

Change Internal ADC clock period12--45

VtestH3.org

ADC clock period

Mean value shift

P3--20

Slo

pe

( D

N/f

ram

e )

Page 9: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Noise feature in the darkIn dark condition.Some (quite a lot of ) flake spots were seen as well as some weak

straps. The noise to output plot shows the distribution relate to PTC

feature.

Fig. 8 Common mode noise feature in the dark.

Common mode noise image

Col

umn

0100200300400500

0

50

100

150

200

250

300

350

400

450

5000

5

10

15

20

25

30

35

testH1

200 300 400 500 600 700 8000

2

4

6

8

10

12

14

16

18

20

Output (DN)

Noi

se (

DN

)

0

100

200

300

400

500

600

700

testH1

Page 10: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Noise changes with ADC clock periodIn dark condition.Most pixel out put in 300—500 DN region show low noise.Some pixel out put in 500—600 DN region show large noise.The average noise increases with increasing ADC clock period, and

the slope is close to that of the average out put.System noise ~2DN (at 12 ADC clock period).

Fig. 9 Average and slope depend on Internal ADC clock periods.

0 10 20 30 40

2.0

2.5

3.0

3.5

4.0

Noise slope=1.178+0.0684*clock

Mean slope=0.125+0.0698*clock

Change Internal ADC clock period12--45

VtestH3.org

ADC clock period

Noise value change

350--430

Ave

rag

e n

ois

e (

DN

)

200 400 600 800

0

5

10

center~400 DN

Change Internal ADC clock period12--45

VtestH3.org

Output ( DN )

Noise distribution

12 15 20 25 30 35 40 45N

ois

e (

DN

)

Page 11: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Count distributions under illuminationDevice under illumination.The out put was cut off at ~600 DN (~300 DN in dark).The count distributions shift to low value with decreasing ADC

clock period (small integration time).The maximum output (cut off) also reduced with decreasing ADC

clock period (small integration time).

Fig. 10 Count distributions at various ADC clock periods.

0 1000 2000 3000 4000

0

1000

2000

3000

600 cut

12

All cut off ~600 DN,small clock period has small outputdistribution.

Change Internal ADC clock period12--45

VtestH9.org

Output ( DN )

Count distribution

12 15 20 25 30 35 40 45

Cou

nt

Average output image

Col

umn

0100200300400500

0

50

100

150

200

250

300

350

400

450

5000

500

1000

1500

2000

2500

3000

3500

4000

testH9

Page 12: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Output and STD change with frame number Device under illumination.Average output increases rapidly in the first a few frames, due to

reset effect.The STD (non-uniformity) reaches maximum at frame 2 might due

to combination of reset and other (heating) effect.

Fig. 11 Frame average output and STD.

0 5 10 15 20400

600

800

1000

Change Internal ADC clock period12--45

VtestH9.org

Frame number

STD

12 15 20 25 30 35 40 45

ST

D (

DN

2 )

0 5 10 15 20800

1000

1200

1400

1600

Change Internal ADC clock period12--45

VtestH9.org

Frame number

Average output

12 15 20 25 30 35 40 45

Ou

t p

ut

( D

N )

Page 13: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Linear fitting of average outputDevice under illumination.Linear fitting for frames 5—20, at various ADC clock periods.The intersect relate to the average illumination.The slope seems linearly with increasing clock period, but the value

is much smaller than that in the dark.

Fig. 12 Linear fitting of average output.

10 20 30 401200

1300

1400

1500

1600

1700Change Internal ADC clock period12--45

VtestH9.org

ADC clock period

Mean value shift

P5--20

Inte

rsec

t ( D

N )

10 20 30 400.04

0.06

0.08

0.10

0.12

0.14

Change Internal ADC clock period12--45

VtestH9.org

ADC clock period

Mean value shift

P5--20

Slo

pe (

DN

/fram

e )

Page 14: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Noise and distribution under illuminationUnder illumination.The noise image corresponds to signal image due to photon noise

(PTC) effect.The noise to output plot shows the count distribution and PTC

feature.

Fig. 13 Noise image and noise to out put plot.

Common mode noise image

Col

umn

0100200300400500

0

50

100

150

200

250

300

350

400

450

5000

5

10

15

20

testH9

600 800 1000 1200 1400 1600 1800 2000 2200 24000

1

2

3

4

5

6

7

8

9

10

Output (DN)

Noi

se (

DN

)

0

50

100

150

200

250

300

350

400

450

500

testH9

Page 15: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

PTC and gain under illuminationUnder illumination.PTC almost linear in 1500—2500 DN region. The larger slope in

600—1300 ND region might relate to weak light?The PTC not sensitive to ADC clock period.The corresponding gains are ~38 electron/DN (1500—2500 DN)

and ~25 electron/DN (700—1100 DN).

Fig. 14 Linear fitting of average output.

10 20 30 4020

25

30

35

40

45

Change Internal ADC clock period12--45

VtestH9.org

ADC clock period

Gain changes with ADC clock period

700-1100 1500-2500

Gai

n (

ele

ctro

n/D

N )

1000 2000 3000 40000

20

40

60

80

100

linear PTC

larger slopecut off?

PTC less sensitive to clock period

Change Internal ADC clock period15--45

VtestH9.org

Output ( DN )

Noise depend on clock period

15 20 25 30 35 40 45

Noi

se 2 (

DN

2 )

Page 16: Characterization of Vanilla Liverpool Physics 30/1/2007 Introduction

Conclusion 1. Average output increased with increasing internal

ADC dark ref. (2800--3200 mV), row number (integration time), ADC clock period and frame number (heating effect). The output shift per frame in the dark was bigger than that under illumination.

2. The CM noise increased with increasing row number (integration time) and ADC clock period, but less significant to internal ADC dark ref.

3. PTC shows good linearity in 1500—2500 DN region and larger slope in 600—1200 DN region. The gain (slope) fitted from PTC was ~38 electron/DN. ADC readout noise will be dominate at small integration time.