Characterization of the NIRPS Guiding Camera for Close Visual Binaries

The problem

A large fraction of stars are in binary or multiple systems, and many of them are expected to have planets. To study those planets, we need to be sure whose light we are measuring.

NIRPS is a near-infrared spectrograph on the ESO 3.6 m telescope at La Silla, Chile, that searches for exoplanets with the radial-velocity method. Starlight reaches it through an optical fibre that sees only 0.4″ of sky, centred on the target star. If a companion star is close, part of its light can also fall into the fibre, mix with the target's spectrum and bias the measurement.

So I asked two questions: when can the guiding camera see the companion, and how much of its light gets into the fibre?

Seeing the star and the fibre at once

The light from the telescope reaches a mirror with a small hole in the middle. The target's light passes through the hole into the fibre. Everything around it is reflected to the guiding camera, which keeps the star centred on the hole.

That is how the camera can watch the star and the fibre at the same time: every image shows the star with a dark hole where the fibre is. The catch is that the camera only sees the light that does not go into the fibre, so it cannot measure the leak directly.

Removing the star's glow

Before searching for companions, I removed the star's own light so that anything companion-sized stands out. I used five filtering steps adapted from Wahhaj et al. (2013).

Fake companions and 95% completeness

To find out what the camera can detect, I injected artificial companions, made from the measured star image, in a spiral around the star: one at each separation. Each image then went through the same filtering, and a companion counts as found when a peak above the detection threshold appears at its position.

I rotated the spiral through 20 starting angles, 18° apart, so that every separation has 20 companions in different directions around the star. Then I made them all fainter step by step.

At each separation, the detection limit is the faintest brightness at which 19 of the 20 are still found: 95% completeness. Together, these limits form the contrast curve.

How faint can a companion be?

The fake companions give the contrast curve: how much fainter than the target a companion can be and still be detected, at each separation.

Close to the star, its halo hides companions. Beyond about 2.5″, the camera finds companions 9.6–9.8 magnitudes fainter than the target, which is around 7,000 to 8,000 times fainter. The curve closely follows a quicker estimate that uses only the noise level (5σ).

How much light leaks in?

To estimate the leak, I placed a shifted and scaled copy of the measured star image at each separation and brightness, and measured how much of its light falls into the fibre. The contamination drops steeply with separation and changes little beyond about 3″.

Companions above the dashed line should be seen by the guiding camera. Those below it can stay invisible while still contaminating the fibre, especially at small separations. So a non-detection in the guiding images is not enough: it has to be combined with a contamination estimate.