2005-05-251 HAWAII-2RG NOISE SNAP Collaboration Meeting M.Bonati, D.Guzman, R.Smith Caltech.

2005-05-25 1

HAWAII-2RG NOISE

SNAP Collaboration Meeting

M.Bonati, D.Guzman, R.Smith

Caltech

2005-05-25 2

The questions

• Is detector material limiting noise performance?

• Is this a property of the 1.7 m cutoff material/process, or the higher operating temperature?

• Is the problem with the noise floor or the starting CDS noise?

Requirement / Goal = 8e- / 5e- @ fowler ~16

...yes

…both

…CDS too high

2005-05-25 3

Confusing effects

• Is detector noise floor masked by dark current, mux glow, or particle hits?

• Is mux degrading detector noise?

• Is controller degrading mux noise?

…slightly. Upgradeable, but not urgent.

…not now. RTS may intrude later.

…particle hits in substrate are a concern.

2005-05-25 4

How do we know detector material limits noise performance?

• Image pixel noise >> bare mux, ref pixels

• Dependence on T, not seen in ref pixels.

• Excess noise goes away when reset switch is closed.

2005-05-25 5

Temporal noise map shows

reference pixels are much quieter than image

• Need >400 samples to see intrinsic width of histogram

2005-05-25 6

Temporal Noise Histograms vs. T

2005-05-25 7

Fowler Noise at 140K

Excess noise goes away when reset switch is closed.

2005-05-25 8

Is it just the temperature?

2005-05-25 9

• At 77K CDS noise is still much higher than for 2.5m HgCdTe.

• True fowler limit may be due to mux glow, dark current, drift or Cosmic Rays.

2005-05-25 10

Dark current vs. Temperature

• Modal dark current is negligible.

• ~0.2e-/read mux glow has small effect at Fowler 64.

Dewar showed no light leak previously, but this has not been tested recently.

2005-05-25 11

Sample Up the Ramp

• Average of all channels is quite linear but individual channels sho considerable scatter.

• ….Drifts?

• ….Cosmic ray hits?

2005-05-25 12

115K

Hot pixels frozen out

Dark images vs TM

ovie

140K

lots of hot pixels

Many diffuse cosmic ray hits

2005-05-25 13

Cosmic Rays ?

• Difference of successive 300s dark images

• 100K

• Fowler 32

Some are very diffuse

Mov

ie

2005-05-25 14

Long darks show flat field patternM

ovie

Dark Flat

2005-05-25 15

e-/ADU vs T

4% rise from 85 to140K

Inverse Gain

2005-05-25 16

Bare mux performance

….Not currently a limitation.

2005-05-25 17

Bare mux 006, ch 1, optimum biases

CDS noise (ADU)

Reset mode Image pixels spatial temporal

Reference pixels spatial temporal

Line, Pulsed 16.8 7.2

Global 10.1 11.8 6.3 8.8

Global + dummy before reset

10.2 6.8

Global + dummy after reset

9.7 6.2

Line, Continuous 5.8 6.8 5.8 6.8

gain = 1.1 to 1.4 e-/ADU

PREVIOUS RESULTS

2005-05-25 18

2005-05-25 19

To reduce bias sensitivity…

Vload

Vo

Column bus

CellDrain

Vreset

RGRow Sel

Col Sel

BiasGate

BiasPower

Ci

Rload

OutputDrain

VddABG tracks BP so bias current ~constant. [Can

use 32 ch readout]

VddA = BP eliminates body effect, but…

…. VddA still acts like phantom gate connection here so VddA=BP

increases net effect of VddA drift.

Vishay J5082.4mA, 700K,

0.05%/C

375

2005-05-25 20

Bare mux029 -- CDS noise

Configuration Spatial noise min.exp 300s

Temporal (mode)

min.exp 300s

BG = BPVddA = BP

5.2 5.7 6.0

BG = BPIndependent VddA

5.25 5.6

ANU bias boardBG=BP, VddA=BP

5.8 6.0 5.4

ANU bias boardSeparate BP, BG, VddA

6.4 7.1 5.9 6.1

Mux030,SDSU biases, BG=BP,

4.9 5.0 4.6

Noise in ADU, gain = 1.1 to 1.4 e-/ADUContinuous line reset, all pixels on ch 1, ~1 sec/read. BP=3.3V, BG=2.4V, VR=0.6,

2005-05-25 21

DRIFTS in CDS frames

Configuration Drift (ADU)RMS of means for 200 frames

Min.exp 300s

SDSU biasesBG=BP, VddA=BP

4.3 36

ANU bias boardBG=BP, VddA=BP

1.4 16

ANU bias boardSeparate BP, BG, VddA

0.9 1.6

Mux030,SDSU biases, BG=BP,

1.5 4.9

RMS variation of mean in ADU. gain = 1.1 to 1.4 e-/ADUContinuous line reset, all pixels on ch 1, ~1 sec/read. BP=3.3V, BG=2.4V, VR=0.6

Drift reduced by shield installation? …less convection around JFET currrent source?

2005-05-25 22

ANU bias loopback tests

• ANU drift and noise are dominated by the Leach 8 channel video board drift and noise, so they are indeed very good.

• We plan to improve the drift of the Leach board by a factor of two by a component change, and will then remeasure.