Preliminary Design Review Read Noise Experiment Preliminary Design Review October 5 2001 Don Figer SPACE TELESCOPE SCIENCE INSTITUTE

Goals of Review Demonstrate that we know how to measure read noise, gain, linearity Choose preferred experiment setups Choose items to purchase Generate actions 13 November 2018

Definition of Read Noise Read noise is the uncertainty in the measurement of charge in a pixel. Note that system read noise is greater than the intrinsic read noise of the detector. Contributions to read noise include: shot noise in FETs Johnson noise (resistive elements) drifts in reference (ground) voltages dB/dt in a ground loop temperature variations pickup of ambient dE/dt, i.e. local transmitters settling effects A/D noise 60 Hz noise 13 November 2018

Maximizing S/N S/N can be increased by reading the detector many times (note that this is not reducing read noise, strictly speaking) Read noise can be reduced by minimizing noise sources, i.e. by shielding, stabilizing temperatures, bandwidth-limiting, etc. Figure 5. Non-destructive reading reduces effective read noise. Data (red) for ALADDIN array, as measured by Co-I Fowler, compared to 1/n1/2 (blue). 13 November 2018

Definition of Gain Gain is the conversion factor between ADU and e-, and applies to the whole A/D chain 13 November 2018

Definition of Linearity Linearity is a measure of the change in the rate that charge is accumulated under constant source flux as a function of the amount of charge already stored. Ideally, the response would be perfectly linear, i.e. that charge would be accumulated at the same rate while a pixel is empty up to the point where it is near full capacity. Practically, pixels deviate from linearity and become less responsive as they fill up. 13 November 2018

Source of non-Linearity See Solomon (1998; PhD Thesis) and Benson et al. (2000; SPIE 4131, 171) 13 November 2018

NGST Requirements NDC0200 (from NGST Doc. #641) 13 November 2018

RGL Experiment Requirements (from NDC0200) Parameter Requirement Read Noise1 1.2 e- RMS Gain NR Linearity2 1using any sampling mode consistent with NGST read modes. Attributes 10% of NGST system noise goal to uncertainty in measuring read noise. 2although NGST photometric accuracy must be <1% 13 November 2018

Photon Transfer Method for Measuring RG Method described in Mortara & Fowler (1981SPIE..290...28) Gain is slope of line Read noise is square root of (y-intercept/gain2) 13 November 2018

Photon Transfer Method for Measuring RGL Method described in Mortara & Fowler (1981SPIE..290...28) g=0.5 ADU/e- Ndot=100 e-/s sR=10 e- Npixels=100 13 November 2018

Proposed RGL Experiment Procedure Drain depletion regions by blanking detector and allowing enough time for trapped charge to randomly bleed out of traps Stabilize detector bias and temperature Blank off detector Obtain bias/dark ramp Remove blank to allow flux of a few hundred e-/s/pixel Obtain another ramp Repeat sequence for range of variations, i.e. read mode 13 November 2018

Alternate Read Noise Experiment Procedure Blank off detector Read detector twice Repeat sequence for range of variations, i.e. read modes 13 November 2018

Proposed Experiment Variations Read modes: dwell multiple sampling Fowler Fowler with multiple sampling Temperature: 3 levels (a through c, b optimal) covering NGST range Bias levels: 2 levels covering NGST requirement (a) and goal (b) for well capacity Combinations: 1a2b3a, 1b2b3a, 1c2b3a, 1d2b3a, 1c2a3a, 1c2c3a, 1c2b3b 13 November 2018

Proposed Experiment Duration Dominant step in terms of schedule is step 1 of primary proposed experiment. Alternate experiment will take less than a day to perform for all variations. Time estimate: 7 days Extended scope: variations in wavelength variations in bandwidth-limiting techniques, i.e. digital vs. analog filtering 13 November 2018

Proposed Experiment Designs Standard TFST hardware (dewar, Leach controller, etc.) Light source (approximately spatially flat) 13 November 2018

Data Reduction/Analysis Procedure Subtract dark frames from illuminated frames Locate sample area that is free of defects and large pixel-to-pixel variations Compute variance over sample area Compute average illumination over sample area Plot variance versus signal Fit a curve to extract read noise and gain Plot signal versus time Fit a curve to extract non-linearity 13 November 2018

Data Reduction/Analysis Procedure for Alternate Read Noise Experiment Subtract bias/dark frames from each other Locate sample area that is free of defects and large pixel-to-pixel variations Measure standard deviation in difference frame over sample area Single-frame read noise is standard deviation divided by root 2 13 November 2018

Expected Performance The accuracy in the read noise measurement is a big question mark We will be able to measure gain with errors less than a percent, using a few thousand pixels in the sample region 13 November 2018

Schedule Because read noise and dark current noise will dominate all measurements that we make, we should perform these experiments first Estimate that the RGL test set would take about 10 days 13 November 2018

Costs (Shopping List) nothing 13 November 2018

Risks Our system will be too noisy to perform the measurements. This is almost always true for a new system and results in a prolonged period of debugging. 13 November 2018

Recommendations While waiting for detector, perform sample tests of read noise with controller and shorted cables 13 November 2018

Actions Test Leach noise with cables 13 November 2018

References http://www.mso.anu.edu.au/observing/detlab/ccdlab/ccd/ccdchar/linearity/linearity.htm 13 November 2018