This post describes the work behind “Exoplanet Hunting in the Classroom: An Easy-to-Implement Experiment Based on Video-Aided Light Curve Analysis with Smartphones”, published in The Physics Teacher in 2024 (co-authored with Alexander Küpper) [[1]]. It also draws on the earlier German-language paper on analogy experiments for the transit method, published in Astronomie+Raumfahrt in 2022 [[2]].

Astronomy, device, and software documentation checked through 2026-10-24.


The Pedagogical Problem

Transit discoveries dominate the current confirmed-planet archive: on 24 October 2026, 4,785 of the 6,451 confirmed planets in the NASA Exoplanet Archive were listed as found by transit [[3]]. The fraction is time-dependent, and the method is biased toward planets whose orbits align with our line of sight. When a planet passes in front of its host star, it blocks part of the observed light. A sufficiently precise photometric measurement may record a dip whose shape depends on radius ratio, impact parameter, orbital motion, stellar limb darkening, exposure time, and noise. A flat bottom belongs to an idealisation, a uniformly bright disc crossed by a non-grazing planet (Fig. 1); it is not the definition of a transit.

This is conceptually accessible. The physics is essentially shadow casting — a topic covered in primary school — applied to an astronomically interesting situation. In our teaching design, the familiar geometry provided an accessible entry point. This article does not report a controlled learning-outcome or motivation study.

The problem is the implementation. How do you actually demonstrate this in a classroom?

Standard approaches divide into three categories, each with limitations:

  1. Simulations and database exercises: Students work with real data from Kepler or TESS, or use a software simulation. These are conceptually valid but remote from physical experience. There is no sensor, no measurement, no uncertainty to grapple with.

  2. Prefabricated kits: Commercial sensor systems such as PocketLab or PASCO can run the measurement; PocketLab published a transit lesson for its Voyager sensor in 2017 [[4]]. They work, but the hardware costs money: PASCO’s Wireless Light and Color Sensor (PS-3248), the successor to its discontinued Wireless Light Sensor, was listed at USD 99 on 24 October 2026 [[5]]. A school that buys a PASCO sensor also depends on PASCO’s software to read it. That software is free of charge: the Sparkvue app, and a Python library for the wireless sensors under a proprietary licence [[5]].

  3. DIY benchtop setups: Various published designs use phototransistors, Arduinos, or similar components with a benchtop light source. These are flexible and cheap but require component procurement, assembly, and some technical confidence from the teacher. The barrier to entry is real.

Our 2024 paper addressed a gap in the approaches it surveyed: an inexpensive, open-source student experiment using a compatible phone, while recognising that some devices require an external sensor. Smartphone transit analogues themselves predate it; the paper cites Barrera-Garrido’s “Analyzing planetary transits with a smartphone” in the same journal [[6]]. The 2024 paper’s contribution lies in what surrounds the phone measurement: a comparison of the equipment options, video-aided analysis of the light curve, and three levels of student independence.


The Smartphone Solution

Many Android smartphones include an ambient light sensor that phyphox [[7]] can access, but sensor availability, range, sampling, and optical placement vary by device. The phyphox sensor list says so plainly: “not every phone has the same set of sensors” [[8]]. Android reports the reading as illuminance in lux [[9]], the light arriving per unit area from every part of the lamp the sensor can see, so hiding part of the lamp’s apparent disc removes that part’s share.

The basic setup requires three things:

  • A light source whose short-term stability has been checked; LED drivers can flicker
  • An opaque sphere to act as the “planet” (a tennis ball, a ping-pong ball, anything with a defined circular silhouette)
  • A smartphone running phyphox, positioned beneath the lamp at a fixed distance and oriented so the light sensor faces upward

When the sphere is moved across the light path at a controlled height and speed, the light sensor records a dip whose shape reflects the apparatus geometry. It is a transit analogy, not a guarantee of a flat-bottomed astronomical light curve.

This is the core experiment. It works. The transit signal is clear enough to measure even with the modest precision of a phone’s ambient light sensor, provided the background illumination is controlled (dark room or at least consistent ambient light).


The iPhone Problem and a Workaround

The current phyphox FAQ states that its App Store build cannot access the iPhone’s physical ambient-light sensor: “On iPhones, there is a physical light sensor, but it cannot be accessed by our app” [[8]]. Apple does document ambient-light data through SensorKit, but only for apps holding an entitlement tied to a “preapproved research study” [[10]]. So a narrow research interface exists; it remains unavailable to this ordinary phyphox experiment.

The paper’s workaround is an external light sensor whose data can be brought into the measurement workflow. Compatibility, firmware, calibration, cost, and availability must be checked for the chosen device; the hardware the paper named in 2024 is not a timeless purchasing recommendation.

The external-sensor approach can remove one source of variation by using the same detector model at a fixed point. It produces comparable data only if the sensors, firmware, calibration, geometry, and acquisition settings have been checked.


Video-Aided Light Curve Analysis

The standard approach to a transit experiment is: measure the dip, calculate the planet-to-star radius ratio from the relative depth, done. This works and is pedagogically valid.

The paper introduces a complementary approach [[1]]: simultaneously recording a video of the “planet” passing in front of the “lamp”, and using the video frames to cross-reference the light curve data.

Why? Because the light curve from a real transit experiment does not look exactly like the idealised textbook version. There is noise. There is baseline drift. The “ingress” and “egress” phases — where the planet is partially in front of the star — can be hard to resolve at a smartphone sensor’s sampling rate and precision. The design problem is to connect the curve to the physical geometry that produced it.

Video-aided analysis addresses this directly. After synchronising the video and sensor clocks, students can compare the sphere’s recorded position with each region of the light curve. Seen from the sensor, the dip begins at first contact, when the sphere’s silhouette touches the edge of the lamp’s disc, and the ingress lasts until second contact, when the silhouette lies wholly inside it. Between second and third contact the hidden area is constant, so a uniformly bright lamp gives a flat bottom and a lamp brighter at its centre a rounded one. The egress, from third to fourth contact, mirrors the ingress (Fig. 1). The correspondence between geometry and photometry — which is the conceptual core of the transit method — becomes explicit.

Figure 1. The four contact points divide the light curve into baseline, ingress, floor and egress. A sphere of apparent radius $k = 0.25$ lamp radii crosses the centre of the lamp’s disc, drawn to scale and cropped; at constant speed the horizontal axis is time. Solid: a uniformly bright disc, floor at $1 - k^2$. Dashed: quadratic limb darkening with illustrative $u_1 = 0.40$, $u_2 = 0.25$, central depth $1.2\,k^2$. Computed from the overlap of two discs; not a measurement.

Contact times depend on the viewpoint, so the video should be filmed from close to the sensor, looking up at the lamp, as the exomoon post recommends for that experiment. Both records must also resolve the ingress, which for a central crossing at constant speed takes the fraction $k/(1+k)$ of the time from first to fourth contact, where $k$ is the sphere’s apparent radius in lamp radii. With $k = 0.25$ and an illustrative four seconds from first to fourth contact, the ingress lasts 0.8 s: 24 frames at 30 frames per second, but a single reading at 1.25 Hz. That was the rate of a TI SensorTag, the slow case in the exomoon experiments; the phone, Arduino and PASCO sensors used there sampled faster. A phone’s own readings are irregular in time, since Android reports the light sensor only when its value changes [[9]]. The two records also keep separate clocks. One way to align them is an event both capture, such as switching the lamp off and on before the run.

The depth itself is one minus the ratio of the floor, between second and third contact, to the baseline outside the transit. Dividing the scatter of each level by the square root of its number of readings gives its statistical uncertainty only if the readings are independent and the level is flat; change-triggered readings undercut the first assumption and a rounded floor the second, and baseline drift adds a systematic error that more readings do not remove.

In a teaching context, the video turns the error and noise in the light curve from an obstacle into an educational resource. Students can identify specific features of the curve and ask: what was happening in the physical experiment at that moment? The uncertainty is no longer an embarrassment. It is a diagnostic.


Scaffolding Levels

The paper distinguishes three implementation modes [[1]], corresponding to different levels of student independence:

Demonstration experiment: Teacher sets up and runs the apparatus. Students observe and discuss. Appropriate as an introduction to the concept before students engage with it independently.

Guided student experiment: Students follow a structured procedure, with specified setup, data collection protocol, and analysis worksheet. Appropriate for students who have not designed their own experiments and for lesson contexts where time is limited.

Open inquiry: Students are given the materials and a research question — “How does the depth of the transit dip depend on the size of the planet?” — and design their own procedure. Appropriate for upper secondary students with experience in experimental design, and for lesson contexts that explicitly address scientific method.

The materials for all three modes are described in the paper. The open-inquiry mode gives students more experimental decisions than the prescribed versions: they build a protocol rather than only following one. Whether that is suitable depends on prior experience, time, and the learning objective.


From the Classroom to the Telescope

A transit experiment with a lamp and a phone is, obviously, not the same as the photometry done by TESS or the James Webb Space Telescope. The planet-star radius ratios, $k = R_p/R_\star$, measurable in the classroom analogue are much larger than for most real exoplanets (Fig. 2). Measured by the uniform-disc depth $k^2$, a sphere a fifth to a half of the lamp’s apparent radius would dim it by 4–25%, Jupiter in front of the Sun by 1.06% and Earth by 84 ppm [[11]]. The median transiting planet in the archive has $k = 0.026$, a depth of about 650 ppm, and only about two dozen of 4,651 exceed $k = 0.2$ [[3]]. The signal-to-noise is worse. The lamp is not a star.

Figure 2. An illustrative classroom range of radius ratios lies in the far tail of the known transiting planets. Histogram: $k$ for 4,651 transiting planets in the NASA Exoplanet Archive’s composite table on 24 October 2026, in bins of 0.05 in $\log_{10} k$ [[3]]. Two lie off the axis: WD 1856+534 b, a giant planet seven times the radius of its white-dwarf host, and TOI-2049 b, whose tabulated ratio of 1.49 disagrees with its tabulated radii. Dotted: Earth and Jupiter in front of the Sun, from the IAU nominal radii [[11]]. Shaded: the assumed classroom range $k = 0.2$–$0.5$. Top axis: the uniform-disc depth $k^2$.

The qualitative measurement principle is shared: an occulter reduces detected light. For an opaque planet wholly inside a uniformly bright, unobscured stellar disc, the depth is exactly the ratio of projected areas, $\delta = (R_p/R_\star)^2 = k^2$, with $R_p$ and $R_\star$ the radii of planet and star. Limb darkening makes a star’s centre brighter than its average, so a central transit is deeper than $k^2$ (1.2 times, for the coefficients of Fig. 1) and its bottom rounded. Grazing geometry, spots, dilution, and the classroom lamp/sensor geometry break the simple relation in other ways. In the classroom, $k$ is a ratio of angular radii seen from the sensor: for small angles, $k \approx (r_p/d_p)/(R_\mathrm{lamp}/d_\mathrm{lamp})$, with $r_p$ and $R_\mathrm{lamp}$ the radii of sphere and lamp and $d_p$ and $d_\mathrm{lamp}$ their distances from the sensor. Students are practising light-curve measurement and model comparison, not recovering an astrophysically calibrated planet radius from any arbitrary lamp.

This connection to real research is not incidental to the pedagogy. It is central to it. The transit method works as a classroom experiment not because it is a good demonstration of some abstract physics principle but because it is a genuine slice of how contemporary science actually operates. The question the experiment answers — is there something out there? — is the same question the professional community is asking.

The simulation companion to this work — a browser-based model with documented limb-darkening, exomoon, and N-body paths — is described in this separate post. The simulation is the place to go when you want to explore parameter space; the physical experiment is the place to go when you want to understand what a measurement actually is.


Connection to the astro-lab

The transit experiment described here grew directly out of the astro-lab project at the University of Cologne, where Alexander Küpper and I had been developing smartphone-based analogy experiments for exoplanet detection since the COVID pivot in 2020. The astro-lab@home established the feasibility of the smartphone approach; the A+R 2022 paper on Analogieexperimente zur Transitmethode (“analogy experiments for the transit method”) [[2]] explored the design space more systematically; the TPT 2024 paper is the version written for an international teacher audience, with the comparative equipment table, the video-aided analysis technique, and the scaffolding levels made explicit.

The exomoon extension came before this paper, not after it. An earlier post describes our 2023 MNU Journal article, in which a small motor carries a moon sphere around the planet sphere: the moon adds small dips of its own, and the flux recovers briefly whenever one silhouette covers the other. It is an optical analogue, not a dynamically scaled model; transit timing variations appear only in the simulation. I previously described it as a later post about detecting the gravitational wobble that a moon induces in a planet’s transit, and both parts of that were wrong.

For the curriculum article that places the transit experiment in the NRW Sekundarstufe I (lower secondary) context — including the Direct Imaging pre-experiment — see Fremde Welten.


References

[1] Spicker, S. J., & Küpper, A. (2024). Exoplanet hunting in the classroom: An easy-to-implement experiment based on video-aided light curve analysis with smartphones. The Physics Teacher, 62(3), 226–229. DOI: 10.1119/5.0125305

[2] Küpper, A., Spicker, S. J., & Schadschneider, A. (2022). Analogieexperimente zur Transitmethode für den Physik- und Astronomieunterricht in der Sekundarstufe I [Analogy experiments on the transit method for physics and astronomy teaching in lower secondary school]. Astronomie+Raumfahrt im Unterricht, 59(188), 46–50.

[3] NASA Exoplanet Archive. Planetary Systems Composite Parameters (live database), queried 24 October 2026. https://exoplanetarchive.ipac.caltech.edu/

[4] Born, R. (2017). PocketLab Voyager: How to discover an exoplanet [lesson]. PocketLab. Opened 24 October 2026. https://archive.thepocketlab.com/educators/lesson/pocketlab-voyager-how-discover-exoplanet

[5] PASCO scientific. Wireless Light Sensor PS-3213 [product page, marked discontinued]; Wireless Light and Color Sensor PS-3248 [product page]; Sparkvue [software page]; pasco: PASCO Python API for Wireless Sensors [Python package]. Opened 24 October 2026. https://www.pasco.com/products/sensors/wireless/ps-3213; https://www.pasco.com/products/sensors/wireless/ps-3248; https://www.pasco.com/products/software/sparkvue; https://pypi.org/project/pasco/

[6] Barrera-Garrido, A. (2015). Analyzing planetary transits with a smartphone. The Physics Teacher, 53(3), 179–181. DOI: 10.1119/1.4908091

[7] Staacks, S., Hütz, S., Heinke, H., & Stampfer, C. (2018). Advanced tools for smartphone-based experiments: phyphox. Physics Education, 53(4), 045009. DOI: 10.1088/1361-6552/aac05e

[8] phyphox. Supported sensors and Frequently Asked Questions. Opened 24 October 2026. https://phyphox.org/sensors/; https://phyphox.org/faq/

[9] Android Open Source Project. Sensor types; Reporting modes. Opened 24 October 2026. https://source.android.com/docs/core/interaction/sensors/sensor-types; https://source.android.com/docs/core/interaction/sensors/report-modes

[10] Apple. SensorKit [developer documentation], including the com.apple.developer.sensorkit.reader.allow entitlement. Opened 24 October 2026. https://developer.apple.com/documentation/sensorkit

[11] Prša, A., Harmanec, P., Torres, G., et al. (2016). Nominal values for selected solar and planetary quantities: IAU 2015 Resolution B3. The Astronomical Journal, 152(2), 41. DOI: 10.3847/0004-6256/152/2/41


Changelog

  • 2025-10-03: Updated the DOI for Spicker & Küpper (2024) to the correct 10.1119/5.0125305.
  • 2026-07-11: Bounded the transit-shape and radius-depth claims to their approximation regimes; corrected Android and iPhone sensor availability; removed timeless hardware/reliability claims; and added the article pages and current documentation cutoff.
  • 2026-10-03: Corrected the reference for the German A+R 2022 paper (authors Küpper, Spicker & Schadschneider; title “Analogieexperimente zur Transitmethode für den Physik- und Astronomieunterricht in der Sekundarstufe I”; issue 4/2022, starting p. 46, per the issue’s table of contents), which had a wrong title, author list and issue number; removed a stray editorial remark from the iPhone paragraph.
  • 2026-10-24: Corrected the exomoon pointer: that post is earlier (September 2023), not later, and describes an occultation analogue, not a gravitational wobble. Replaced “the lamp’s light cone” and “full occultation of a central portion of the lamp” with the four contact points (Fig. 1). Stated that $\delta = k^2$ is exact only for an opaque disc inside a uniform one and that the classroom $k$ is a ratio of angular radii. Quantified “much larger than for most real exoplanets” with archive radius ratios and IAU radii (Fig. 2) and added the archive’s count of transit discoveries. Kits: replaced “purpose-built transit experiment setups” and “expensive, closed-source” with PocketLab’s 2017 lesson, PASCO’s current sensor price and free but proprietary-licensed software. Replaced “explicitly warns” with the phyphox quotation, quoted the phyphox iPhone answer and Apple’s SensorKit entitlement, and added that Android reports illuminance in lux, on change. Added the viewpoint, sampling, clock and uncertainty requirements of the video comparison, and Barrera-Garrido (2015) as prior work. Numbered and cited the references.