TV UMi: masses and radii of a marginal-contact binary from TESS

Two Sun-like stars that just touch, seen at a shallow tilt, get masses and radii to a few percent. We check the orbit, the speeds and the brightness with fibub tools.

Some pairs of stars orbit so close that their surfaces touch. These contact binaries whirl round each other in less than a day and share one outer layer, like a peanut. TV UMi, in the constellation of the Little Bear, is one of them, but it only just touches, and we see its orbit at a shallow tilt, so its eclipses are faint. A new study combines twenty sectors of TESS light curves with old, unused spectra to weigh and measure both stars to a few percent. This article shows how the radial velocity, the exoplanet mass calculator and the blackbody calculator let you check the orbit, the speeds and the light.

The paper in brief

The light from TV UMi rises and falls almost like a smooth wave every 0.4155 days, about 10 hours. A smooth wave can come from many things, so the team needed a second kind of data. They found radial velocities, the speeds of the stars towards and away from us, measured years ago at the David Dunlap Observatory in Canada and never analysed. Fitting the light and the speeds together gave an orbit tilted 48.3 degrees from face-on, masses of 1.254 and 0.929 times the Sun’s, and radii of 1.240 and 1.080 times the Sun’s. Both stars are a little hotter than the Sun, near 6,200 to 6,300 K.

The stars overfill their shared boundary by only about 2%: a “marginal” contact. A dark spot on one star drifts over time, and its movement matches small shifts in the eclipse times. The team classes TV UMi as a young W-subtype contact binary, about 2 billion years old. In a W-subtype system the smaller star is the hotter one, as here.

Checked with the tools

We used the paper’s period, tilt, masses, radii and temperatures, and fibub’s maths. We did not have the spectra, so the velocity fit in Step 2 uses simulated speeds and says so.

Step 1: how close are they? Kepler’s third law, with the two masses together (2.18 Suns) and the 0.4155-day period, puts the centres of the stars 3.04 solar radii apart. The two radii add up to 2.32 solar radii, 76% of that gap. Each star’s radius is 0.408 and 0.355 of the separation. The Roche lobe, the region each star’s gravity controls, is 0.405 and 0.353 of the separation for this mass ratio (Eggleton’s formula). So each star just overfills its lobe, by under 1% in radius, which fits the paper’s “marginal contact”.

Step 2: weighing the stars from their speeds. At the paper’s tilt, the masses predict speed swings (semi-amplitudes, K) of 117.6 km/s for the heavier star and 158.7 km/s for the lighter one. We made 24 pretend spectra over 40 nights with 8 km/s of scatter, and gave them to the Radial velocity fitter’s maths with only a search range for the period. It found a period of 0.41546 to 0.41555 days and swings of 118.5 ± 1.8 and 157.3 ± 2.4 km/s (Figure 1). From those two numbers alone, the masses come out as 0.515 and 0.388 Suns times sin³ i, a factor that hides how tilted the orbit is. Dividing by sin³ 48.28° = 0.416 gives 1.24 and 0.93 Suns, back to the paper’s masses. The speed ratio gives the mass ratio, 0.753, against the paper’s 0.741.

Simulated velocities and the fitted orbitsSimulated velocities and the fitted orbits-200-100010020000.20.40.60.81Orbital phaseSpeed towards or away from us (km/s)Primary (simulated)Secondary (simulated)fibub.com
Figure 1. Simulated velocities of the two stars (24 spectra, 8 km/s scatter, our assumptions) built from the paper's masses and tilt, with the orbits found by the Radial velocity fitter's maths. Simulated, not the observatory's spectra.

Step 3: why the tilt matters so much. Spectra only see motion along our line of sight. Figure 2 shows how the combined swing K₁ + K₂ would change with the tilt for these same stars: 370 km/s edge-on, but only 276 km/s at 48 degrees. Masses go as the cube of the speeds, so near 48 degrees a 1 degree error in the tilt moves the masses by about 4.7%. That is why the shallow, hard-to-model eclipses limit the precision, and why the paper’s few-percent masses are a careful result.

Total speed swing against the tilt of the orbitTotal speed swing against the tilt of the orbit010020030040020406080Inclination (degrees, 90 = edge-on)K₁ + K₂ (km/s)48.28° (paper)fibub.com
Figure 2. The two stars' combined speed swing against inclination, from the paper's masses and period with the Exoplanet mass calculator's maths. A shallow tilt hides much of the motion.

Step 4: the light. The Blackbody calculator gives a 6,152 K star of 1.240 solar radii a power output of 1.98 Suns, and a 6,336 K star of 1.080 solar radii 1.69 Suns. The paper gives 1.95 ± 0.51 and 1.68 ± 0.48, a match to within 2%. The smaller star is the hotter one, the W-subtype signature.

Check Paper Our tool result
Separation of the centres (in the fit) 3.04 solar radii
Radii against Roche lobes overcontact 2.1 ± 1.8% 0.408 vs 0.405 and 0.355 vs 0.353 of the separation
Speed swings at 48.28° (fitted to the spectra) 117.6 and 158.7 km/s predicted
Masses back from simulated speeds 1.254 and 0.929 Suns 1.24 and 0.93 Suns
Luminosities 1.95 and 1.68 Suns 1.98 and 1.69 Suns

How we made the graphs

The charts and every number above come from scripts/blog-figures/tv-umi-marginal-contact-binary-tess-masses.mjs in fibub’s public code. It calls the tools’ own functions:

  • fitOrbit, rv and binary from src/lib/sci/rvfit.ts, the engine of the radial velocity.
  • semiMajor and semiAmplitude from src/lib/sci/rvmass.ts, the maths behind the exoplanet mass calculator. The formulas are the same for a planet and a star.
  • star from src/lib/sci/blackbody.ts, the maths behind the blackbody calculator.

Where the inputs come from: the period, tilt, masses, radii, temperatures and luminosities are from the paper. The Roche lobe check uses Eggleton’s 1983 formula, worked in the script, not a fibub tool. The number of pretend spectra (24), their spread over 40 nights, the 8 km/s scatter and a systemic speed of zero are our own choices, labelled in the script.

How we test the maths

On every change to fibub, tests/sci.mjs checks these functions against known answers:

  • The orbit maths must give Jupiter’s pull on the Sun as 12.47 m/s, and 51 Pegasi b’s orbit as 0.053 au.
  • The Radial velocity fitter must recover a simulated eccentric binary (period 12.345 days, swing 23.4 km/s, eccentricity 0.42) within three times its quoted errors, and its mass function must match the textbook form 1.0361 × 10⁻⁷ (1 − e²)^1.5 K³ P.
  • A circular fit to 30 noisy velocities must recover a 4.2308-day period and a swing of 55.9 within three times its errors.
  • The blackbody maths must give the Sun exactly one solar luminosity and an absolute magnitude of 4.82.

Check it yourself

The speeds, on the exoplanet mass calculator:

  1. Choose Wobble from the mass.
  2. Type 1.254 in Star mass (Suns) and 0.415548 in Orbital period (days). Leave Eccentricity at 0.
  3. In Planet mass choose Jupiters and type 973 (the lighter star’s 0.929 Suns is about 973 Jupiters).
  4. Type 48.28 in Inclination (degrees, optional).
  5. Velocity semi-amplitude K reads about 117,500 m/s: the heavier star’s 117.6 km/s. Swap them (star 0.929 Suns, companion 1314 Jupiters) to get the lighter star’s 158.7 km/s.

The light, on the blackbody calculator: choose The temperature (K) in I know, type 6152, then 1.24 in Radius (Suns). It reads about 1.98 Suns. Try 6336 and 1.08 for about 1.69 Suns.

The orbit, on the radial velocity: paste time (days) and velocity (km/s) for one star, one measurement per line. Set Velocity unit to km/s (binary stars), tick Circular orbit, and type 0.3 and 2 in Shortest period to search and Longest period to search. With velocities of TV UMi you should find a period near 0.4155 days; the swing it finds is only the part along our line of sight, so divide the masses by sin³ i as in Step 2. We did not have the observatory’s velocities, so we could not run this step on the real data.

What researchers can use it for

  • Check a binary’s masses against its speeds, period and tilt before using them in a model.
  • Fit velocities from spectra without installing anything; the data stay on your computer.
  • See how the tilt limits the masses: try a few inclinations in the mass calculator and watch the speeds change.
  • Compare a star’s size and temperature with its brightness, as a quick check on a table of stellar parameters.
  • Teach how double stars are weighed, and why eclipses are so valuable for fixing the tilt.

The fine print

  • The velocities in Figure 1 are simulated from the paper’s numbers. They show the fitter can recover orbits like these, not what the observatory’s spectra contain.
  • In a contact binary the stars are not spheres. The radii here are “equal-volume” radii, and the Roche lobe check is only a rough guide.
  • The luminosities treat each star as a uniform black body; spots and the shared envelope make the real light more complex.
  • We assumed a circular orbit, as is normal for stars this close.

Sources

  • Čeki, A. and Latković, O. (2026), TV UMi: A Shallowly Eclipsing Marginal-Contact Binary, Galaxies 14(1), 9, doi:10.3390/galaxies14010009. Licence CC BY 3.0, checked on the journal page, 2026-11-16. Values used here, from the paper's table of results and text: period 0.4155480 days, mass ratio 0.7411 ± 0.0030, inclination 48.28 ± 0.59°, masses 1.254 ± 0.038 and 0.929 ± 0.028 solar, radii 1.240 ± 0.016 and 1.080 ± 0.015 solar, temperatures 6152 ± 200 K and 6336 ± 365 K, luminosities 1.95 ± 0.51 and 1.68 ± 0.48 solar, overcontact 2.1 ± 1.8%, age 2.02 ± 0.82 billion years.
  • Roche lobe size formula: Eggleton, P. P. (1983), Approximations to the radii of Roche lobes, ApJ 268, 368.

If you are an author or publisher and want a credit or a passage changed or removed, tell us and we will act quickly.

Tools in this article

Read next