🔭 astrophysics

High-Precision Photometry with a scientific CMOS Camera: II On-Sky Testing of the Marana camera at the NGTS facility

On-sky testing at the NGTS facility demonstrates that the Andor Marana scientific CMOS camera achieves photometric precision comparable to traditional CCDs while detecting 20% more photons per unit time due to its faster readout, confirming its suitability for high-precision astronomical time-series photometry.

Original authors: Ioannis Apergis, Daniel Bayliss, Paul Chote, James McCormac, Peter J. Wheatley, Morgan A. Mitchell, Jorge Fernández Fernández, Sam Gill, Edward M. Bryant, Toby Rodel, Leonidas Asimakoulas, David R. An

Published 2026-03-18

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ioannis Apergis, Daniel Bayliss, Paul Chote, James McCormac, Peter J. Wheatley, Morgan A. Mitchell, Jorge Fernández Fernández, Sam Gill, Edward M. Bryant, Toby Rodel, Leonidas Asimakoulas, David R. Anderson, James A. Blake, Sara L. Casewell, Fintan Eeles-Nolle, Faith Hawthorn, James S. Jenkins, Monika Lendl, Isobel S. Lockley, Maximiliano Moyano, Sean M. O'Brien, Suman Saha, Alexis M. S. Smith, Philip G. Steen, Jose I. Vines, Richard G. West, Tafadzwa Zivave

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to take a perfect photograph of a tiny, faint star as a planet passes in front of it. In the world of astronomy, this is like trying to spot a gnat flying in front of a distant streetlamp. To do this, you need a camera that is incredibly sensitive, fast, and quiet.

For decades, astronomers have relied on CCD cameras (the same technology found in older digital cameras) for this job. They are the "reliable old horses" of the astronomy world: steady, accurate, but a bit slow and heavy.

Recently, a new contender has entered the ring: the CMOS camera. You might know these from modern smartphones. They are the "race cars" of the imaging world: incredibly fast, lightweight, and getting better every year. But are they fast enough and precise enough to replace the trusted CCDs for serious science?

This paper is the report card from a head-to-head race between the two, held at the NGTS observatory in the high, dry mountains of Chile.

The Race Setup

The scientists took a brand new, high-tech scientific CMOS camera (called the Marana) and mounted it on one of the NGTS telescopes. Right next to it, on a neighboring telescope, they kept the old-school iKon-L CCD camera.

They pointed both telescopes at the exact same patch of sky, at the exact same time, using the exact same filters. They watched eight different exoplanets (planets orbiting other stars) as they transited (crossed in front of) their host stars. It was a controlled, side-by-side test to see who could measure the star's brightness more accurately.

The Contenders

The Veteran (CCD):
Think of the CCD as a meticulous librarian. It reads the data slowly and carefully. It has a very high "sensitivity" to red light (which is great for seeing cooler stars), but it takes a long time to "read" the data from the sensor. After taking a picture, it needs a few seconds to reset before it can take the next one. This creates "dead time" where it misses photons (particles of light).

The Sprinter (CMOS):
The Marana CMOS is like a high-speed data courier. It reads the data 70 times faster than the CCD. It can snap a picture and be ready for the next one in a blink of an eye. It has very low "noise" (static), meaning the image is cleaner. However, its sensitivity to red light is slightly lower than the CCD's, and it has a unique "dual-gain" system that handles bright and dim light differently.

The Results: Who Won?

The race wasn't a simple "one wins, one loses" story; it depended on the conditions.

1. The "Bright Star" Sprint (The 10-second exposure)
When looking at bright stars, the CMOS camera actually collected more light overall.

The Analogy: Imagine two people trying to catch rain in buckets. The CCD has a slightly bigger bucket (better sensitivity), but it takes 3 seconds to empty it and get ready for the next drop. The CMOS has a slightly smaller bucket, but it can empty and refill it almost instantly.
The Outcome: Because the CMOS is so fast, it catches 20% more rain (photons) over the same amount of time. This extra light makes the data slightly clearer for bright stars, even though the CCD is technically more sensitive to red light.

2. The "Faint Star" Hurdle
When the stars were dimmer, the CCD had a slight edge.

The Analogy: The CCD is like a deep-depleted sponge that soaks up red light very well. The CMOS is a bit less efficient at soaking up that specific red light. Since the NGTS telescope uses a filter that blocks blue light and lets red light through, the CCD's natural strength gave it a tiny advantage for very faint targets.

3. The "Noise" Factor
Both cameras produced incredibly clean data. The "static" or "graininess" in the images was so low that both cameras reached the theoretical limit of what is possible from Earth's surface (a limit caused by the atmosphere twinkling, known as scintillation).

The Verdict: The CMOS camera proved it is just as quiet and precise as the veteran CCD. It didn't introduce any new "bad habits" or errors.

The "Glitch" and the Fix

The paper also looked at a specific quirk of the CMOS camera. Because it is so advanced, it uses two different "gears" (High Gain and Low Gain) to handle light. If a pixel is right on the edge of switching gears, it can get a little confused.

The Finding: The scientists found that for the stars they were watching, this "gear-switching" confusion happened so rarely and affected so few pixels that it didn't mess up the science at all. It's like a car shifting gears so smoothly you don't even feel it.

The Bottom Line

The paper concludes that CMOS cameras are ready for prime time.

They are not just "good enough" to replace CCDs; in many ways, they are better. Their speed allows them to catch more light, they are lighter (making telescopes easier to balance), and they are just as precise.

The Future:
Just as smartphones replaced point-and-shoot cameras for everyday photography, this study suggests that CMOS sensors will soon replace CCDs for professional astronomy. They offer the speed and precision needed to find more exoplanets, track satellites, and monitor the sky with a level of detail we haven't seen before.

In short: The old guard (CCD) served us well, but the new generation (CMOS) is faster, smarter, and ready to take the wheel.

1. Problem Statement

While Charge Coupled Device (CCD) cameras have been the standard for astronomical time-series photometry (e.g., in Kepler, TESS, and NGTS), Complementary Metal-Oxide Semiconductor (CMOS) technology has advanced rapidly, offering lower read noise and faster readout speeds. However, there was a lack of comprehensive on-sky validation comparing modern scientific CMOS sensors directly against established deep-depleted CCDs in a high-precision exoplanet survey context. The authors aimed to determine if CMOS sensors could match or exceed the photometric precision of CCDs for bright star monitoring, specifically addressing concerns regarding read noise, dark current, dynamic range, and systematic noise (red noise).

2. Methodology

The study utilized the Next Generation Transit Survey (NGTS) facility at the ESO Paranal Observatory in Chile.

Instrumentation:
- Test Subject: An Andor Marana 4.2B-11 scientific CMOS camera (2048 × 2048 pixels, 11 µm pixel size, back-illuminated, HDR mode).
- Control Subject: An Andor iKon-L 936 deep-depleted CCD camera (2048 × 2048 pixels, 13.5 µm pixel size), the standard instrument for NGTS.
- Setup: The Marana CMOS was mounted on NGTS Telescope ID 06, while the iKon-L CCD was mounted on the neighboring Telescope ID 01. Both telescopes used identical NGTS custom filters (520–890 nm) and were pointed at the same fields.
Observation Strategy:
- Simultaneous Observations: The team conducted simultaneous observations of 8 known bright exoplanet host stars (including WASP-4, WASP-95, KELT-10) and TESS candidates.
- Exposure Parameters: Both cameras used a 10-second exposure time. The CMOS readout time was 42 ms, while the CCD required ~3 seconds (including readout and dead time), resulting in a significantly higher duty cycle for the CMOS.
- Conditions: Observations covered both dark nights (no moon) and bright nights (full moon) to test performance under varying sky background conditions.
Data Reduction:
- A unified pipeline was used for both datasets, including astrometric solving (WCS), source extraction (SEP), and relative photometry.
- Detrending: Systematic errors (atmospheric extinction, color terms) were corrected using the SysRem algorithm and Gaussian Processes (Matern 3/2 kernel) via the allesfitter package.
- Noise Modeling: Theoretical noise models were constructed including read noise, dark current, sky background, photon shot noise, and scintillation noise (using Young's formula).

3. Key Contributions

Direct On-Sky Comparison: This is one of the first studies to directly compare a modern scientific CMOS (Marana) against a high-end deep-depleted CCD (iKon-L) under identical observing conditions and reduction pipelines.
HDR Mode Validation: The study validates the use of the Marana's High Dynamic Range (HDR) mode, which combines High Gain and Low Gain channels, for astronomical photometry, confirming that the transition between gain channels does not significantly degrade photometric precision for the targets observed.
Duty Cycle Analysis: The authors quantified the advantage of the CMOS camera's fast readout, demonstrating that despite slightly lower Quantum Efficiency (QE) in the red, the CMOS collects more photons per unit time due to reduced dead time.
Sky Background Sensitivity: The paper highlights that the CMOS sensor is less sensitive to the red-shifted night-sky airglow (OH emission lines) compared to the deep-depleted CCD, resulting in lower sky background noise for the CMOS in specific conditions.

4. Key Results

Photometric Precision:
- The Marana CMOS achieved a precision of 500 ppm at a 30-minute timescale for a $T=10$ mag star, comparable to the iKon-L CCD.
- For bright stars ( $10 < T < 12$ ), the CCD showed slightly lower noise (better precision) because the NGTS filter is optimized for the CCD's red-sensitive QE, and the target flux dominates the noise budget.
- For fainter stars ( $T > 12$ ), the CMOS outperformed the CCD. This is attributed to the CCD's higher read noise (12.9 $e^-$ vs. 1.57 $e^-$ for CMOS) becoming a limiting factor as signal decreases.
Photon Collection Efficiency:
- Despite the CCD having higher QE in the red (NGTS band), the CMOS camera collected 20% more photons per unit time for solar-type stars. This is due to the CMOS's 70x faster effective readout speed (42 ms vs. 3 s dead time), allowing for a much higher duty cycle.
Noise Characteristics:
- Read Noise: The CMOS read noise was negligible even for faint stars, whereas the CCD read noise became significant for stars fainter than $T=13$ mag.
- Dark Current: While the CMOS has higher dark current than the cooled CCD, it was never a dominant noise source for the 10s exposures used.
- Red Noise: Both cameras exhibited minimal correlated (red) noise, indicating that the CMOS introduces no significant instrumental systematics for bright star photometry.
- Saturation: The CMOS showed higher RMS for very bright stars ( $T < 9.5$ ) due to pixel saturation and charge conservation issues in the anti-blooming architecture, whereas the CCD exhibited "bleeding" but maintained better linearity in extreme saturation.
Transit Recovery: Light curves for 8 transiting exoplanets (e.g., WASP-4) derived from the CMOS were consistent with those from the CCD, with fitted parameters (radius ratio, orbital period) agreeing within uncertainties.

5. Significance and Conclusion

The study concludes that modern scientific CMOS cameras, specifically the Andor Marana, are highly suitable for high-precision time-series photometry and can effectively replace CCDs in exoplanet surveys.

Advantages: The CMOS offers superior performance in the photon-limited regime (fainter stars) due to ultra-low read noise and significantly higher duty cycles. Its anti-blooming architecture also reduces contamination from saturated bright stars compared to CCD bleeding.
Future Applications: The authors suggest that CMOS sensors are ideal for applications requiring rapid data capture, such as tracking satellites, monitoring asteroids, and detecting transits of rapidly varying targets.
Impact: As CMOS technology continues to advance, it is poised to become the standard detector for ground-based time-domain astronomy, offering a combination of speed, sensitivity, and precision that traditional CCDs cannot match.