A search for photometric variability towards the globular cluster M3 with the TWenty Inch Survey Telescope

TL;DR

This study validates the TWIST facility, a 50 cm sCMOS telescope, for high-cadence time-domain photometry through a 67-night, white-light campaign on globular cluster M3, achieving a 10 s cadence over a 199-day baseline. It reports the discovery of four SX Phoenicis variables (V301–V304) and confirms cluster membership, with V302 identified as a fundamental-mode pulsator enabling a distance modulus of $ (m-M)_V = 15.02 \pm 0.10$ (true modulus $ (m-M)_0 = 14.98 \pm 0.10$) consistent with literature; it also delivers updated periods and amplitudes for 203 RR Lyrae stars and detects Blazhko-like modulation in 53 stars. A key methodological advance is a new flat-field correction technique that removes scattered-light–driven vignetting while preserving dust features, improving long-baseline photometric stability. The analysis shows that microlensing events in M3 are negligibly rare for small-aperture surveys, and three foreground M-dwarf flares were observed as short-lived transients, establishing TWIST’s sensitivity limits for non-periodic variability in dense stellar fields.

Abstract

We present the commissioning results and first scientific observations from the newly installed TWIST observatory - a 50 cm telescope equipped with an sCMOS camera providing a $36.1\times24.1$ arcmin$^2$ field of view - housed in the former SuperWASP-North enclosure. We conducted a 67-night, 199-day baseline white-light monitoring campaign centred on the globular cluster M3 aimed at characterizing stellar variability within the cluster while also assessing the photometric performance of the newly commissioned system. We report the discovery of four new SX Phoenicis variables (V301-304), confirm their cluster membership, and identify fundamental-mode pulsation in one, allowing an independent period-luminosity-based distance estimate to M3. We revisited 231 previously known RR Lyrae stars, providing updated period measurements for 203 and white-light amplitudes for 198. We detected Blazhko-like modulation in 53 stars and characterized the modulation parameters for 28. Notably, we measure periods and amplitudes for the unclassified variables V286 and V287 for the first time. We also identify three foreground flaring M dwarfs, and assess the feasibility of detecting microlensing events in M3, concluding that expected rates are negligible. Alongside the scientific results, we introduce a new correction technique for flat-field images affected by scattered light and present a full characterization of the observatory's photometric capabilities. These results demonstrate the scientific utility of TWIST for high-cadence time-domain surveys using modest-aperture instrumentation.

Figures (19)

• Figure 1: The TWenty-Inch Survey Telescope in its enclosure on La Palma, Canary Islands. This image was taken from the southern entrance of the building, facing North.
• Figure 2: The observing limits of TWIST. Left: The minimum observable altitude (black line) measured to the nearest degree versus the azimuthal angle, sampled every 10$^\circ$. The white-shaded region shows the observable altitude/azimuth combinations. The concentric circles are lines of constant altitude and the polar angle represents the azimuthal angle. The dashed grey lines are loci of points of a fixed declination, illustrating the sidereal motion of an object of that declination. Right: The observable hour angles (shaded white) visible for a given declination and the total unobstructed time per night (black dashed line) at that declination. This time does not include sun or moon-induced visibility restrictions.
• Figure 3: Overview of the M3 survey observations. The red bars represent the total number of 10-second exposure, on-target images captured each night, and the blue points indicate the median half-flux diameters of all images taken on that night. Nights with a lower-than-expected number of images typically reflect either delayed starts to observations or early terminations due to environmental triggers. The two large gaps in the data are the result of deliberate non-observing periods, not adverse weather or other external factors. The month labels reflect the beginning of the month in the year 2024.
• Figure 4: The $36.1\times24.1$ arcmin$^2$ survey field centred on M3. The locations of known variables from clement_variable_2001 that have a corresponding Gaia source are shown. RRL, SXP, semi-regular, eclipsing binary, W Virginis and uncategorized stars are shown with open circles, squares, diamonds, triangles, stars, and pentagons, respectively. The newly discovered SXP-type variables V301--304 and the unclassified variables V286-287 are labelled. North is up and East to the left. The image was created by aligning with spalipylyman_spalipy_2021 and stacking 360 calibrated frames (totalling one hour of exposure) with measured half-flux diameters less than 2 arcsec taken during dark sky environments. The 5$\sigma$ limiting magnitude of the stacked image for a 5-pixel aperture radius is around G = 22.1.
• Figure 5: Overview of the flat-field correcting process. a) The untouched master flat from 2023 December 13 -- the first night of observations. b) The flat image detrended by a bivariate spline fit. c) Cross-correlation map of the spline-detrended flat with an example dust doughnut. d) The spline-detrended flat overlayed with the locations of the detected doughnuts. e) The fit mask; masked areas are shown in white. f) The RBF-detrended master flat. g) The master vignetting pattern used for the whole survey. h) The corrected master flat image.