Characterization of the CSST Survey Camera CCDs: I. basic electro-optical performance

arXiv:2609.19105 · astro-ph.IM · Submitted 2026-09-16 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics. The day in astrophysics: observations, models and what the data forces.

Vera: We're looking at the paper: "Characterization of the CSST Survey Camera CCDs".

Jocelyn: Here's the gist of the paper. The Chinese Space-station Survey Telescope (CSST) is a flagship mission designed to carry out a large-area high-resolution multiband imaging and slitless spectroscopic survey over a decade This paper is the first in a series that present ground-based characterization of the 31 flight-model charge-coupled devices (CCDs) of the CSST Survey Camera Basic Electro-optical Performance Baseline This research establishes a performance baseline for all 31 flight-model CCDs from their bias frames, dark frames, and flat fields Key results include a full-well capacity of ≥ 80 ke− (limited by the test electronics), non-linearity < 1.4% in the range of 7-70 ke−, median readout noise ∼ 3.6 e−, dark current ≤ 0.02 e− pix−1 s−1 at the operating temperature of −85◦C, high quantum efficiency across the survey’s spectral range (255–1000 nm), typical pixel response non-uniformity 99.9994% at the flat-field level of 10 ke− All 31 CCDs meet or exceed the Survey Camera’s specifications.

Vera: Detector Architecture and Testing Setup The CSST-SC mosaic focal plane consists of 31 9k × 9k charge-coupled devices (CCDs) with a pixel size of 10µm × 10µm Two types of CCDs are installed: GM1201A from China Electronics Technology Group Corporation No.

Jocelyn: 44 Research Institute and CCD290-99 from Teledyne e2v GM1201A uses standard silicon with different coatings for near-ultraviolet (NUV) and visible (VIS) subtypes, which we refer to as 44NUV and 44VIS, respectively CCD290-99 has more variants: standard silicon with NUV coating (e2vNUV), standard silicon with astro-multi 2 coating (e2vVIS), and deep-depletion silicon with astro-multi 2 coating (e2vDD), which is significantly more sensitive at wavelengths above 650 nm than standard silicon CCDs.

Vera: The test bench comprises a flat-field illumination system and various imaging test apparatuses All measurements are carried out at the working temperature of −85◦C except for dark current, which is also measured at −75◦C and −95◦C.

Title and authors: Vera: Thinking about the title of this paper, "Characterization of the CSST Survey Camera CCDs: I. basic electro-optical performance," it really frames what we're looking at—it’s not a deep dive into the science yet, but just making sure the camera hardware is actually capable of doing its job.

Jocelyn: And you see that this work is coming from a team working on this flagship mission, which means they’re focused on making sure the detectors are reliable for ten years of continuous operation in space.

The paper's summary: Vera: The paper summarizes their testing by detailing how they derived several key performance indicators from the PTC flat-field images, which is basically using those calibration images to calculate things like quantum yield and charge transfer efficiency.

Jocelyn: They explain how the PTC slope relates to the quantum yield alpha divided by conversion gain G in units of electrons per ADU, and that conversion gain needs to be calibrated individually for every single channel of each CCD they tested.

Subrahmanyan: The paper also discusses a "brighter-fatter effect" which bends down the PTC more as flat-field intensity increases, and they say that you need the correlation between three neighboring pixels to recover linearity without adding too much random noise.

Vera: They also detail how they measured readout noise by looking at the standard deviation of pixel values in electrons from bias frames taken at negative eighty-five degrees Celsius, and then taking the median of those noises for each channel.

Jocelyn: And they quantified charge transfer efficiency, or CTE, which is given by one minus Ntrap over Nflatntrans—they found that the CTE values were greater than nine point nine nine nine four six for the e2vVIS CCD at ten kilo-electrons parallel.

Subrahmanyan: That high CTE value is critical because if charge transfer isn't efficient, you lose signal when stacking images, which impacts your ability to reconstruct faint sources or measure subtle structures in the sky.

Vera: They also measured the PRNU or pixel response non-uniformity, typically finding it less than one percent across the whole system.

Jocelyn: So in short, they’ve characterized how well each detector handles light before we even see what those detectors are seeing from deep space observations.

The paper's improvements: Vera: One thing the paper points out is that several detector effects require further study, specifically noting that the e2vNUV CCD shows a marked gradient in its flat field at three hundred nanometers but not at four hundred seventy nanometers.

Jocelyn: That means there are subtle differences between the silicon coatings, and they suggest they need to investigate those specific spectral dependencies for each detector type more closely.

Subrahmanyan: From a modeling perspective, this gradient suggests that our current models might oversimplify the response of certain detector subtypes across different wavelengths, so we need to refine those models.

Vera: Then there’s the suggestion to use semi-physical models to effectively reconstruct the brick wall pattern in NUV because that’s a dominant non-uniformity for some of these CCDs.

Jocelyn: That sounds like a practical step; if you can model that known pattern, you can correct for it and get a cleaner image, especially since they're looking at the two hundred fifty nanometer to one thousand nanometer range.

Subrahmanyan: And in terms of noise modeling, incorporating those median readout noise estimates from the test results—which they found to be about three point six electrons root mean square—will help AI systems better estimate the true signal-to-noise ratio when processing flight model data.

Vera: That's a practical application; knowing that the median readout noise is that specific number helps us build more accurate noise models for our actual science pipelines.

Conclusion: Jocelyn: So to wrap up, the paper confirms that the basic electro-optical performance of the CSST-SC flight model CCDs meets all the specified requirements, with full-well capacity exceeding eighty kilo-electrons.

Vera: And they've verified that their response non-linearity stays below one point four percent over a specific range and less than one percent over another, which is exactly what we needed to see for the survey’s science drivers.

Subrahmanyan: The dark current being no more than zero point zero two electrons per pixel per second at negative eighty-five degrees Celsius confirms that the hardware is stable enough for long-term operation in space environments.

Jocelyn: They also found the median readout noise is about three point six electrons root mean square with those test electronics, and the quantum efficiency covers two hundred fifty nanometers to one thousand nanometers.

Vera: And they verified that the PRNU is typically less than one percent, and charge transfer efficiency was greater than nine point nine nine nine four six for the e2vVIS CCD at ten kilo-electrons parallel.

Subrahmanyan: Overall, this characterization of the CSST Survey Camera CCDs: I. basic electro-optical performance provides a solid foundation that validates the hardware before we start pouring massive amounts of data into the survey over that decade.

Jocelyn: It’s a necessary first step to ensure the science we get out of this telescope is as clean and reliable as possible.

Vera: It's just good to see these numbers solid, and then we can move on to the next set of papers that will show us what actual data looks like coming from this instrument.

Key Laboratory of Space Astronomy and Technology, National Astronomical Observatories, Chinese Academy of Sciences · University of Chinese Academy of Sciences · Kavli Institute for Astronomy and Astrophysics, Peking University

astro-ph.IM

Submitted: 2026-09-16

Updated: 2026-10-29

Comments: 22 pages, 17 figures, accepted by RAA, comments are welcome

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 85/100

The gist: The Chinese Space-station Survey Telescope (CSST) is a flagship mission designed to carry out a large-area high-resolution multiband imaging and slitless spectroscopic survey over a decade This paper

Key concepts

Full-Well Capacity
This is the maximum amount of charge (electrons) that a CCD pixel can hold before it overflows and loses information. The test results showed this capacity is at least 80 ke−, though it was ultimately limited by the readout electronics rather than the silicon itself.
Readout Noise
This represents the inherent electronic noise present when reading out the signal from a pixel. The median readout noise measured was about 3.6 e− rms with test electronics, which is a critical factor in determining how faint signals can be reliably detected during image processing.
Charge Transfer Efficiency (CTE)
CTE measures how effectively charge is moved from one pixel to the next during readout. The measured values were very high, exceeding 99.994% for serial transfer and 10 ke− parallel, indicating minimal charge loss during this critical step.
Pixel Response Non-Uniformity (PRNU)
This refers to small, random variations in the sensitivity of individual pixels across the detector. The typical PRNU was found to be less than 1%, which is important for accurately modeling and correcting systematic errors in scientific data.

Terminology

Summary

The Chinese Space-station Survey Telescope (CSST) is a flagship mission designed to carry out a large-area high-resolution multiband imaging and slitless spectroscopic survey over a decade This paper is the first in a series that present ground-based characterization of the 31 flight-model charge-coupled devices (CCDs) of the CSST Survey Camera

Basic Electro-optical Performance Baseline

This research establishes a performance baseline for all 31 flight-model CCDs from their bias frames, dark frames, and flat fields Key results include a full-well capacity of ≥ 80 ke− (limited by the test electronics), non-linearity < 1.4% in the range of 7-70 ke−, median readout noise ∼ 3.6 e−, dark current ≤ 0.02 e− pix−1 s−1 at the operating temperature of −85◦C, high quantum efficiency across the survey’s spectral range (255–1000 nm), typical pixel response non-uniformity 99.9994% at the flat-field level of 10 ke− All 31 CCDs meet or exceed the Survey Camera’s specifications.

Detector Architecture and Testing Setup

The CSST-SC mosaic focal plane consists of 31 9k × 9k charge-coupled devices (CCDs) with a pixel size of 10µm × 10µm Two types of CCDs are installed: GM1201A from China Electronics Technology Group Corporation No. 44 Research Institute and CCD290-99 from Teledyne e2v GM1201A uses standard silicon with different coatings for near-ultraviolet (NUV) and visible (VIS) subtypes, which we refer to as 44NUV and 44VIS, respectively CCD290-99 has more variants: standard silicon with NUV coating (e2vNUV), standard silicon with astro-multi 2 coating (e2vVIS), and deep-depletion silicon with astro-multi 2 coating (e2vDD), which is significantly more sensitive at wavelengths above 650 nm than standard silicon CCDs. The test bench comprises a flat-field illumination system and various imaging test apparatuses All measurements are carried out at the working temperature of −85◦C except for dark current, which is also measured at −75◦C and −95◦C.

Performance Metrics Derivations

The paper details the derivation of several performance indicators from the PTC flat-field images

**((3) The PTC slope is equal to the quantum yield α divided by the conversion gain G in units of e−ADU−1 (Janesick 2001, 2007) The quantum yield is given by the product of the PTC slope and the conversion gain The conversion gain is needed in other tests and must be calibrated individually for each channel of each CCD The quantum yield is a property of the silicon material itself and is affected by the electric field strength in the silicon (Janesick 2007) The brighter-fatter effect bends down the PTC more and more as the flat-field intensity increases (Downing et al. 2006; Ma et al. 2014; Rasmussen et al. 2016) The correlation between three neighbouring pixels is sufficient to recover the PTC linearity without introducing excessive random noise The brighter-fatter effect is more pronounced with e2vDD CCDs, so that their correlation matrix has to include five neighbouring pixels. 3.4 Readout noise is determined as follows: for each pixel, standard deviation of its values in units of electrons is calculated from the bias frames taken previously during the dark current measurement at −85◦C The resulting standard deviation may be considered as the readout noise of the individual pixel, even though all pixels in the same channel are read through the same signal path, which should produce the same readout noise in an idealized system The readout noise of each channel is then taken to be the median of all the pixels’ readout noises in the channel 3.7 CTE quantifies how well charges are transferred between pixels The CTE is given by (Janesick 2001; Ryon & Grogin 2019) CT E = 1 − Ntrap/Nflatntrans, where ntrans = 4616 for the parallel CTE The CTE values are greater than 99.99846 for the e2vVIS CCD at 10-ke− parallel. 4 summary and discussion We have characterized the CCDs for the CSST-SC flight model and verified that their basic electro-optical performance meets the specifications The full-well capacity exceeds 80 ke− and is actually limited by the readout electronics The median readout noise is about 3.6 e− rms with the test readout electronics The QE is measured from 250 nm to 1100 nm The PRNU is typically less than 1% The CTE is greater than 99.994 at 1 ke− serial. Several detector effects require further studies. The e2vNUV CCD shows a marked gradient in its flat field at 300 nm but not at 470 nm. The PRNU is quantified by the relative fluctuations of pixel values in the flat-field images after subtracting contributions from the photon noise, dark-current noise and readout noise (collectively, the random noise). 4 summary and discussion We have characterized the CCDs for the CSST-SC flight model and verified that their basic electro-optical performance meets the specifications The response non-linearity is below 1% over the 14–56 ke− range and below 1.4% over the 7-70 ke− range. The full-well capacity exceeds 80 ke− and is actually limited by the readout electronics The dark current spans more than an order of magnitude from one CCD to another. All of them meet the requirement of no more than 0.02 e− pix−1 s−1 at the operating temperature of −85◦C. The median readout noise is about 3.6 e− rms with the test readout electronics The QE is measured from 250 nm to 1100 nm. The PRNU is typically less than 1%. The CTE is greater than 99.

Improvements for AI systems

  1. Improve image simulation pipelines by incorporating measured pixel response non-uniformity of less than 1% across all flight model CCDs to create more accurate data models for CSST image simulations. This allows for better prediction of systematic errors in derived scientific results such as precision cosmology and stellar astrophysics.

  2. Enhance data processing algorithms to accurately correct the brighter-fatter effect by applying the correlation matrix summation method, as this technique is used to remove the non-linearity caused by the brighter-fatter effect. This correction will improve flux independent Point Spread Function estimation for deep surveys like Euclid or LSST.

  3. Develop more robust noise modeling in data analysis software by incorporating median readout noise estimates from test results, which are found to be about 3.6 e minus rms with the test readout electronics. This allows AI systems to better estimate the true signal-to-noise ratio when processing flight model data.

  4. Create advanced dark current subtraction routines that account for temperature variations across different CCD subtypes, as the paper shows that the choice of -85 degree C as the working temperature for the focal plane is a reasonable balance between performance and engineering cost. This will ensure accurate background subtraction regardless of minor thermal drifts in space.

  5. Implement charge transfer efficiency correction modules using the measured Charge Transfer Efficiency (CTE) values, which are greater than 99.9994% in both serial and parallel directions with 10 ke minus flat fields. This directly improves the accuracy of source reconstruction by mitigating charge trapping effects during image stacking.

  6. Integrate wavelength-dependent PRNU correction models into image processing pipelines, specifically using semi-physical models to effectively reconstruct the brick wall pattern in NUV, as this is a dominant non-uniformity for certain CCDs. This will enhance photometric accuracy across the entire 255 nm to 1000 nm survey range.

Abstract

The Chinese Space-station Survey Telescope (CSST) is a flagship mission designed to carry out a large-area high-resolution multiband imaging and slitless spectroscopic survey over a decade. The survey's science drivers, ranging from precision cosmology to stellar astrophysics, impose stringent requirements on the performance of its focal plane detectors. This paper is the first in a series that present ground-based characterization of the 31 flight-model charge-coupled devices (CCDs) of the CSST Survey Camera. Measurements of basic electro-optical performance are reported here. Key results include: full-well capacity at least 80 ke- (limited by the test electronics), non-linearity <1.4% in the range of 7-70 ke-, median readout noise 3.6 e-, dark current at most 0.02 e-pix-1 s-1 at the operating temperature of-85, C, high quantum efficiency across the survey's spectral range (255-1000 nm), typical pixel response non-uniformity <1%, and charge transfer efficiency >99.9994% at the flat-field level of 10 ke-. All 31 CCDs meet or exceed the Survey Camera's specifications. These measurements establish a performance baseline and provide inputs for CSST image simulation and data processing pipelines

Sources

Related papers