Can a Planet Have a Moon? Teaching Exomoon Detection with a Disco Ball Motor
Abstract
Classroom treatments of exoplanet detection usually focus on the transit method. What gets omitted is that moons of exoplanets could also host life — and that with a small motor and a slight modification to the standard transit experiment, you can show students what an exomoon signature looks like in a light curve. Published in MNU Journal in 2023.
Contents
Literature and calculations checked through 2026-10-24; classroom-hardware cutoff: 11 July 2026.
This post describes the paper “Ein Analogieexperiment zur Suche nach Exomonden” (An Analogy Experiment for the Search for Exomoons), published in MNU Journal in 2023 together with Alexander Küpper [[1]].
The Gap in the Curriculum
Most physics and astronomy teaching units that address the search for extraterrestrial life focus on exoplanets. The transit method gets visualised, a light curve gets plotted, and the lesson ends with: some exoplanets are in the habitable zone. The end.
What tends to get omitted: moons of exoplanets — exomoons — are plausible targets in habitability discussions. A moon orbiting a planet in a conventional habitable-zone orbit receives a related stellar flux, but habitability also depends on its atmosphere, mass, tidal heating, magnetic environment, and many other factors. The classroom value is the detection problem, not a claim that a habitable-zone planet automatically supplies a habitable moon.
The pedagogical gap is this: the transit method is routinely demonstrated in analogy experiments, but the extension to exomoon detection is almost never treated experimentally, even though it is a natural continuation of the same experiment with only minor modifications. Our paper is an attempt to close that gap.
What an Exomoon Signal Looks Like
When only a planet transits a star, the resulting light curve shows a dip as the planet moves in front of the star. In the uniform-brightness, non-grazing limit, the normalised flux during the fully superimposed phase is:
$$I(t) = \frac{A_s - A_p}{A_s} = 1 - \frac{A_p}{A_s} \tag{1}$$where the dip depth $\delta = A_p / A_s$ is determined by the ratio of the planet’s cross-sectional area to the star’s. The areas are projected areas as seen from the detector; for a real star and planet, both effectively at the same distance, $\delta = (R_p/R_s)^2$.
When the planet has a moon, the situation is more complex. While planet and moon both lie inside the stellar disc, the light curve is governed by:
$$I(t) = \frac{A_s - (A_p + A_m - A_{pm}(t))}{A_s} \tag{2}$$where $A_m$ is the moon’s cross-sectional area and $A_{pm}(t)$ is the time-dependent overlap between the planet’s and moon’s projected discs (the moon is orbiting the planet, so this overlap changes during the transit). This is the mutual-transit light curve of Sato and Asada [[2]]. For disc radii $r_p \ge r_m$ whose centres are a distance $d$ apart, the overlap is a lens made of two circular segments:
$$A_{pm} = r_p^2\,(\theta_p - \sin\theta_p \cos\theta_p) + r_m^2\,(\theta_m - \sin\theta_m \cos\theta_m) \tag{3}$$where $\cos\theta_p = (d^2 + r_p^2 - r_m^2)/(2 d r_p)$ and $\cos\theta_m = (d^2 + r_m^2 - r_p^2)/(2 d r_m)$ fix the half-angles that the common chord subtends at the two centres. Eq. (3) holds for $r_p - r_m < d < r_p + r_m$. Once the discs separate ($d \ge r_p + r_m$), $A_{pm} = 0$; while the moon is entirely covered ($d \le r_p - r_m$), $A_{pm} = A_m = \pi r_m^2$. During ingress and egress, $A_p$, $A_m$ and $A_{pm}$ are replaced by the parts of each area that lie on the stellar disc.
In this projected-disc model, additional dips and asymmetries can appear in the light curve [[3]], [[4]]. The moon can transit slightly before the planet (causing a small flux dip before the main transit begins), or pass in front of or behind the planet during the transit (temporarily reducing the combined occulting area, causing a brief flux recovery in the middle of the dip), or trail the planet across the star (causing a small dip after the main transit ends). In front and behind give the same overlap, so the light curve alone cannot tell them apart; Sato and Asada describe the recovery as an apparent increase in brightness whenever one body occults the other [[2]]. The signature depends on the relative sizes of planet and moon, their orbital motion, stellar limb darkening, cadence, noise, and the geometry of the particular transit.
These signatures are small and can be confused with starspot crossings [[3]], [[4]], stellar variability, instrumental systematics, and planet-only models. A moon the size of ours, transiting a Sun-like star, would block about 6 ppm of the light (the calculation is in a later post).
The two most discussed candidates are much larger. Kepler-1625 b-i, proposed as a moon roughly the size of Neptune, rests largely on one Hubble transit with a moon-like dip of about 500 ppm [[5]]; Kepler-1708 b-i, about 2.6 Earth radii, came from a survey of 70 cool giant planets found by Kepler [[6]]. A reanalysis concluded that neither planet is likely to host a large moon [[7]]; the original authors replied, in the preprint of their response, that both candidates remain viable but need further observations [[8]].
Microlensing has produced candidates too; the one on NASA’s exomoon page “could be either a moon and a planet – or a planet and a star” [[9]], and a microlensing event cannot be observed again [[4]]. No exomoon had been unambiguously confirmed by September 2026 [[10]]. In the classroom analogue, the features are deliberately enlarged so that their geometry is visible: a moon sphere 0.12 lamp radii across, the illustrative proportion of Figs. 1 and 2, hides $0.12^2 \approx 1.4\%$ of the lamp, about 30 times the Kepler-1625 b-i dip.
The Experimental Setup
The starting point is a standard transit analogy experiment [[11]]: a sphere (the planet) on a rod, moved slowly around a lamp (the star) by a slowly rotating motor. A light sensor — an Android smartphone running phyphox, or an Arduino with a suitable sensor — records the illuminance over time. The resulting light curve shows the characteristic symmetric transit dip.
The modification is straightforward: attach a small battery-powered motor to the planet sphere, with a smaller sphere (the moon) on the motor’s arm. The motor we used is a disco ball motor — inexpensive, widely available, and with a rotation speed that works well relative to the transit timescale if you choose the geometry appropriately.
The result is a physical system with two independent circular motions. It is an optical analogy, not a dynamically scaled model of a star–planet–moon system:
- The planet orbiting the star (driven by the main slow-rotation motor)
- The moon orbiting the planet (driven by the disco ball motor)
When this system transits the “star” (the lamp), the light sensor records a compound light curve with the exomoon signatures described above.
One technical note on sensors: High sample rate matters here. The exomoon signatures are brief features on top of the transit dip, and a sensor that samples too slowly will average them out. We found that the TI SensorTag CC2650, despite being a reasonable choice for the basic transit experiment, has a light sensor sample rate of only 1.25 Hz — too slow in the published apparatus. The Android phone, Arduino, and Pasco sensor configurations used there provided higher rates; that result should not be generalised to every phone, board, application, firmware version, or sensor.
How brief the features are follows from the geometry. Take a moon at distance $a > r_p + r_m$ from the planet’s centre, on a circular orbit of period $P_m$ seen edge-on. A mutual event, the whole stretch during which the two discs overlap, then lasts
$$T_{pm} = \frac{P_m}{\pi} \arcsin\frac{r_p + r_m}{a}$$per passage, which for $a \gg r_p + r_m$ becomes Sato and Asada’s $2(r_p + r_m)/(a\omega)$ with $\omega = 2\pi/P_m$ [[2]]. At 1.25 Hz a reading arrives every 0.8 s, so a mutual event gets about $T_{pm}/(0.8\,\mathrm{s})$ points, and tracing its shape needs several. The rate matches the 800 ms conversion time of the SensorTag’s OPT3001 light sensor; the chip also has a 100 ms mode [[12]], [[13]], and whether a given firmware and app use it is a separate question.
Reading the Light Curves
The paper presents two distinct light curve types that emerge from the experiment, each with a different exomoon orbital configuration [[1]]. Figs. 1 and 2 compute both from Eqs. (2) and (3) with illustrative radii; they are not the published measurements.
Type 1: The moon’s orbital period is short relative to the transit duration. Multiple exomoon signatures appear within a single transit. These include:
- A small dip before the main transit begins (moon transiting alone)
- Asymmetric ingress/egress (moon leading or trailing the planet)
- A brief flux recovery during the transit (moon passing behind or in front of the planet, reducing the total occulted area)
- A small post-transit dip (moon still in front of the star after the planet has exited)
The recoveries repeat. The discs overlap whenever the moon passes behind the planet or in front of it, twice per moon orbit, so a fast moon brings the flux back to the planet-only level every half moon period [[2]] (Fig. 1).
Type 2: Specific orbital phase alignment where the moon moves directly behind the planet around mid-transit, while the planet is fully in front of the lamp. While the moon is hidden, $A_{pm} = A_m$ in Eq. (2) and the occulted area is the planet’s alone, so this is the shallowest part of the transit floor, not the deepest; the deepest stretches are those in which planet and moon are both on the lamp and apart. As the moon emerges from behind the planet, the total occulted area increases again before both planet and moon exit (Fig. 2). In an earlier version of this post I called the hidden-moon moment the deepest point of the transit, which contradicted both Eq. (2) and my own next sentence.
This second case permits a projected-area exercise under the uniform-disc approximation: if the relevant configurations are identified, students can estimate separate planet and moon areas from changes in the occulted area. In Fig. 2 the moon-alone step gives $A_m/A_s$, the hidden-moon plateau gives $A_p/A_s$, and the deepest stretch gives their sum, which serves as a check. A real light curve requires limb darkening, orbital geometry, finite exposure, and uncertainty to be modelled before inferring radii.
Video + Light Curve Together
The paper recommends recording a video of the experiment simultaneously with the light sensor measurement, from the perspective of the sensor (i.e., looking up at the lamp from below). This technique — which is also central to the transit method paper — is even more valuable here.
Without the video, the exomoon signatures in the light curve are easy to misread as noise or experimental error. With the video, students can advance frame by frame through the moments corresponding to the unusual features and see what the classroom apparatus was doing: the moon sliding in front of the planet, the moon emerging from behind the planet, the moon transiting alone at the start or end of the main event.
The paired record is intended to help interpretation by letting students test a feature against the visible apparatus. The article reports a design rationale, not a measured reduction in cognitive load.
Differentiation and Extensions
The paper presents the exomoon experiment as an extension after a transit unit. That is a pedagogical recommendation, not a measured learning-outcome claim.
For students who are comfortable with quantitative analysis, Eq. (2) allows a projected-area treatment: given the measured analogue curve and known lamp radius, they can estimate sphere radii under the stated assumptions. The depths measure angular sizes, and the spheres sit closer to the sensor than the lamp does: for small angles, the depth in Eq. (1) is $\delta \approx (r_p/d_p)^2 / (R_\mathrm{lamp}/d_\mathrm{lamp})^2$, where $d_p$ and $d_\mathrm{lamp}$ are distances from the sensor. The estimate $r_p \approx \sqrt{\delta}\, R_\mathrm{lamp}\, d_p / d_\mathrm{lamp}$ therefore needs both distances as well.
Possible further extensions:
- Noise floor investigation: systematically vary the moon’s size and determine the smallest moon still detectable. This connects directly to one part of the real astrophysical problem: small signals compete with stellar variability, instrumental systematics, and alternative models.
- Period ratio effects: vary the transit speed (and thus the effective period ratio between moon and planet) to see how the light curve changes.
- Sensor comparison: test different sensor types and compare their ability to resolve exomoon signatures. This turns the instrumental limitation into an explicit investigation.
For the deeper theoretical connections — transit timing variations, the David Kipping approach to exomoon detection — see the transit simulation post, which models these effects in a browser-based tool.
For the secondary school curriculum context and the Direct Imaging pre-experiment that typically precedes the transit unit, see Fremde Welten.
References
[1] Küpper, A., & Spicker, S. J. (2023). Ein Analogieexperiment zur Suche nach Exomonden. MNU Journal, 76(3), 223–227.
[2] Sato, M., & Asada, H. (2009). Effects of mutual transits by extrasolar planet–companion systems on light curves. Publications of the Astronomical Society of Japan, 61(4), L29–L34. DOI: 10.1093/pasj/61.4.L29
[3] Tusnski, L. R. M., & Valio, A. (2011). Transit model of planets with moon and ring systems. The Astrophysical Journal, 743(1), 97. DOI: 10.1088/0004-637X/743/1/97
[4] Heller, R. (2018). Detecting and characterizing exomoons and exorings. In H. J. Deeg & J. A. Belmonte (Eds.), Handbook of Exoplanets (pp. 835–851). Springer. DOI: 10.1007/978-3-319-55333-7_35. Preprint as “On the detection of extrasolar moons and rings”, arXiv:1701.04706.
[5] Teachey, A., & Kipping, D. M. (2018). Evidence for a large exomoon orbiting Kepler-1625b. Science Advances, 4(10), eaav1784. DOI: 10.1126/sciadv.aav1784
[6] Kipping, D., Bryson, S., Burke, C., Christiansen, J., Hardegree-Ullman, K., Quarles, B., Hansen, B., Szulágyi, J., & Teachey, A. (2022). An exomoon survey of 70 cool giant exoplanets and the new candidate Kepler-1708 b-i. Nature Astronomy, 6(3), 367–380. DOI: 10.1038/s41550-021-01539-1
[7] Heller, R., & Hippke, M. (2024). Large exomoons unlikely around Kepler-1625 b and Kepler-1708 b. Nature Astronomy, 8(2), 193–206. DOI: 10.1038/s41550-023-02148-w
[8] Kipping, D., Teachey, A., Yahalomi, D. A., Cassese, B., et al. (2025). Concerning the possible exomoons around Kepler-1625 b and Kepler-1708 b. Nature Astronomy, 9(6), 795–798. DOI: 10.1038/s41550-025-02547-1. The wording used here is from the preprint’s abstract, “A Reply to: Large exomoons unlikely around Kepler-1625 b and Kepler-1708 b”, arXiv:2401.10333.
[9] NASA. ExoMoons [media gallery]. Opened 24 October 2026. https://www.nasa.gov/universe/exoplanets/exomoons/
[10] Pérez-Rodrigo, A., & Rebollido, I. (2026). That’s no moon: Exomoon survival and detection limitations. Astronomy & Astrophysics, 714, A299. DOI: 10.1051/0004-6361/202661276. Preprint: arXiv:2609.07320.
[11] Küpper, A., Spicker, S. J., & Schadschneider, A. (2022). Analogieexperimente zur Transitmethode für den Physik- und Astronomieunterricht in der Sekundarstufe I. Astronomie+Raumfahrt im Unterricht, 59(188), 46–50.
[12] Texas Instruments. CC2650STK: SimpleLink Bluetooth low energy/multi-standard SensorTag [product page]. Opened 24 October 2026. https://www.ti.com/tool/CC2650STK
[13] Texas Instruments (2017). OPT3001 ambient light sensor (ALS). Data sheet SBOS681C. https://www.ti.com/lit/ds/symlink/opt3001.pdf
Changelog
- 2025-10-03: Updated the Tusnski & Valio (2011) reference to use article number 97, replacing the previous page range “1–16.”
- 2026-07-11: Qualified the habitability, confirmation, and classroom claims. The apparatus is an enlarged geometric analogy; it omits the noise, dynamics, and astrophysical inference problem of a real exomoon search.
- 2026-10-03: Corrected the reference for the MNU Journal paper to issue 76(3), pp. 223–227 (MNU Heft 3/2023), not 76(5).
- 2026-10-24: Corrected the Type 2 light curve: with the moon hidden behind the planet, Eq. (2) leaves the planet’s area alone occulted, the shallowest part of the transit floor; the post had called it the deepest point. Type 2 is now described as the moon passing behind the planet around mid-transit rather than “at the moment of maximum occultation”. In the video section, the moon emerges from behind the planet, not from its shadow, which a projected-disc analogy does not have. Described the post-transit dip more precisely (the moon trailing the planet across the lamp) and noted that a moon in front of the planet and one behind it give the same overlap. Added the overlap area, Eq. (3), the condition that Eq. (2) holds only while both discs are on the star, the duration of a mutual event with the sampling arithmetic for the 1.25 Hz sensor, and the distances a radius estimate in the classroom needs. Added two figures computed from Eqs. (2) and (3). Compared the classroom signal with a Moon-sized satellite and with the Kepler-1625 b-i dip, and summarised the dispute about Kepler-1625 b-i and Kepler-1708 b-i. Corrected the Heller (2018) chapter title to the published one (“Detecting and characterizing exomoons and exorings”; the old title was the preprint’s). Replaced “Every classroom treatment” in the summary with “usually”, as the paper’s own abstract has it (“oft”). The NASA page, previously described as a candidate-status overview, is now a media gallery and is cited only for its microlensing caption.