I was four years old when I pressed my eye to the eyepiece of a 14-inch telescope at the Carr Astronomical Observatory in Collingwood, Ontario. Saturn came into view with its rings perfectly inclined against the blackness of space. I drew what I saw and wrote about it in my best handwriting. That observation was published in the Royal Astronomical Society of Canada's Scope Magazine in 2014.

Eleven years later, I operate a research-grade remote observatory on a mountain in southern Spain - 7,000 kilometres from my bedroom in Toronto. It has its own International Astronomical Union (IAU) Minor Planet Center code R60. Every asteroid position I submit from it enters the global database of minor planets.

This is my story of how I built it: the engineering, failures, logistics, and problem-solving that made it real, and the science it enables.

Since 2021, all my observational research had been done on borrowed time. I wrote proposals for the Faulkes Telescope Project (FTP), the American Association of Variable Star Observers (AAVSO), and the Canadian Space Agency CSA). I coordinated with citizen scientists across four continents. I accumulated over 55 hours of observation on the Didymos binary asteroid system that contributed to a peer-reviewed paper in The Planetary Science Journal.

But borrowed telescope time comes with a limitation: you don't fully control it. Sessions are blocked in fixed windows. Targets are shared. And some of the hardest observations in my research - exoplanet transits - demand up to six continuous hours at a precise time, on a specific night, regardless of what else is scheduled. Similarly, windows to study binary asteroids when there a bright enough to be observed are limited to a few days.

I chose infrastructure. And since Toronto sits at Bortle class 9, the worst possible sky brightness, a backyard telescope was not an option. The observatory had to be remote, at a dark-sky site, and it had to be something I could build and operate as a high school student.

Before selecting any equipment, I wrote a research framework. It was a set of non-negotiable performance requirements that every aspect of my remote observatory from its location to power backups would be evaluated against. The most demanding target was millimagnitude photometry: the ability to detect brightness changes in stars as small as 0.001 magnitudes, or one-tenth of one percent. That precision is required to measure exoplanet transit depths and model asteroid rotation curves reliably.

I even considered the future data landscape. The Vera C. Rubin Observatory's LSST survey will begin generating millions of transient alerts per night. Many will require photometric follow-up from ground-based telescopes and I wanted to be in a good position to be able to be able to collaborate with other research scientists and observatories.

I evaluated hosting options across multiple continents — Australia, Chile, southern Spain, and the United States against a set of quantified criteria based on my experience of operating shared telescopes.

That last point mattered more than anything, and I am glad that I prioritized it. When you're operating a telescope remotely from 7,000 km away, the supply chain for replacement parts or something that did fit cannot be an afterthought, but a core reliability requirement to minimize telescope downtime. Sourcing equipment from a Spain-based vendor eliminated customs delays and meant that if a component failed, it could be replaced in days rather than weeks. 

I compared specifications, reviewed published test data, and mapped each component against the performance requirements before committing to anything. Below is what was selected and the engineering reasoning behind their selection. 

The Ritchey-Chrétien design uses two hyperbolic mirrors mathematically shaped to eliminate coma and field curvature simultaneously. These are the same optical configuration as the Hubble Space Telescope. For photometry, the star shape at the edges of the sensor matters as much as at the centre. A star that appears elongated or comet-shaped introduces systematic error into every brightness measurement. The RC design removes that error at the optical level.

Mount: Sky-Watcher EQ8-R Pro

An equatorial mount must rotate around a single axis aligned with Earth's rotation to track the sky. The EQ8-R is rated for 50 kg payloads with sub-arcsecond periodic error correction, which means that the mechanical imperfections in its gears are small enough to be corrected by the guiding software without introducing residual drift into long exposures. In fact the foundation of millimagnitude level photometry precision starts from choice of mount and its payload capacity.

Monochrome sensors are more photometrically sensitive than colour sensors because no light is sacrificed to the Bayer colour filter array on the sensor. Active cooling drops the sensor to below 0°C, suppressing thermal dark current which is the noise that accumulates in pixels over time and would contaminate faint-source measurements in multi-hour sessions which the telescope would be put to use.

A separate guide telescope introduces flexure - microscopic mechanical movement between the guide and science optical paths, which the guiding system cannot distinguish from actual tracking error. The OAG picks off light from inside the primary telescope's optical path, sharing the same tube. The guide camera and PHD2 software measure a guide star's position 15 times per second and send real-time corrections to the mount. The practical result: tracking stable to under 1 arcsecond for sessions of many hours. I was confident in this solution as I had first tried this configuration on my home telescope and it worked very well, but there was a steep learning curve on how to calibrate and operate it so that it is able to find a guide star even when sky field is sparse.

Rather than running a remote computer connected to a rack of separate USB hubs and controllers, the EAGLE integrates a full Windows PC, powered USB ports, and serial ports into a single unit mounted directly on the telescope. All device control runs through one remote session, reducing the number of physical connection points that can fail and reduces the weight on the mount. 

Learning: During the mounting process, I noticed the EAGLE was heating up a lot, which would impact the telescope mirrors. As a result, I was forced to move it to the telescope's tripod. 

Physical assembly of the telescope took place at AstroCamp on March 18, 2025, with the on-site team.

Three critical alignment procedures had to be completed correctly before the system could produce usable science data.

A Ritchey-Chrétien telescope is only as good as the alignment between its two mirrors. Misalignment by a fraction of a millimetre invisible to the eye produces asymmetric star profiles that invalidate photometric measurements. I used the laser-ring method: a collimation laser projects concentric rings onto the primary mirror, and the secondary mirror's tilt screws are adjusted iteratively until all rings are perfectly concentric. The effect is visible in defocused star images: a chaotic, off-centre donut transforms into a clean, symmetric ring. 

2. Polar Alignment: Tracking the Sky

An equatorial mount's rotation axis must point within arcseconds of the celestial pole. The first polar alignment attempt achieved arcminute-level accuracy which is sufficient for snapshots but limiting usable exposure times to 5–10 seconds before trailing became visible. A second attempt using the three-point TPoint routine in TheSkyX achieved arcsecond-level accuracy, reducing the pointing error by a factor of roughly 60. The telescope now tracks stably for at least three minutes unguided, and indefinitely with active guiding.

A raw astronomical image is not science data. It contains three categories of systematic noise that must be removed before any measurement is worth any scientific use. 

During one session, the guiding system repeatedly reported that it could not make sufficient corrections to hold a target. I worked through the problem systematically: recalibrated the guiding algorithm, re-verified polar alignment through software, and inspected every configuration parameter. Hours passed. Nothing worked.

The on-site team did a physical inspection and found a single loose screw in the camera adapter coupling. It was introducing micro-vibrations — motion so small it was invisible to the camera images, but detectable to the guide camera as uncorrectable drift. Five seconds of tightening with a screwdriver fixed what hours of remote software diagnostics could not reach.

The lesson was structural, not just practical: remote systems require a diagnostic framework that begins with physical inspection, not ends with it. I've since built a pre-session checklist that includes physical verification steps with the on-site team before I begin any software troubleshooting.  

First light occurred on June 12, 2025. The target was the Eagle Nebula — selected not for its scientific value but to evaluate tracking stability and star-shape uniformity across the full sensor field. The camera cooling system stabilised below 0°C as expected. The calibrated images showed clean, round stars to the corners of the 26-megapixel sensor.

To receive an official IAU Minor Planet Center observatory code, I submitted astrometric measurements of more than 10 Near-Earth Asteroids, each observed on at least two separate nights with 3–5 images per session, alongside full instrumentation specifications for independent verification. 

Code R60 means that any position measurement I submit is accepted into the same database that professional observatories contribute to. It marked the formal transition from building infrastructure to contributing to planetary defense. 
 

The first extended campaign targeted main-belt asteroid (2977) Chivilikhin: 7 nights, 25 total hours of photometry. By measuring how the asteroid's brightness fluctuates as it rotates — different facets of its irregular surface reflecting varying amounts of sunlight — I derived a rotation period of 6.257 ± 0.001 hours with a light curve amplitude of 1.030 magnitudes, fitting a clean fourth-order Fourier model. The data have been submitted to the Asteroid Lightcurve Database (ALCDEF).

Three exoplanet transit observations were accepted and published by the ExoClock Project, contributing to ephemeris refinement for ESA's upcoming Ariel space telescope: 

These results contribute to a coordinated international programme ensuring that when Ariel launches, its target ephemerides are accurate enough to schedule observations precisely. 

The observatory is now in full science operations. Near-term priorities include expanding the exoplanet transit programme across a broader target list for Ariel, characterising additional asteroid rotation periods and shape models for planetary defense, and beginning multi-filter photometry campaigns to extract colour information about asteroid surface compositions.

Longer term: binary asteroid candidates — pairs of asteroids orbiting each other — whose mutual eclipse events produce photometric signatures that only extended, dedicated baseline coverage can detect. That kind of observation was impossible with shared telescope time. It is possible now.

At four years old, I drew Saturn with my best handwriting. At fifteen, I built the instrument. What gets measured next is the open question.