Autonomous Astrophotography
Two synchronized imaging systems, one mount, and fully automated observing sequences.
Two optical systems, one mount
Two optical systems sharing one mount
The observatory carries two imaging systems: ARTEC 250 Pro for higher-resolution imaging and William Optics RedCat 51 III for wide-field acquisition. Each system has its own N.I.N.A. sequence and camera/accessory profile, but both depend on the same physical 10Micron GM2000 HPS mount and the same roll-off roof.
This creates a control challenge: the two sessions may be exposing at different filter lengths, but the mount cannot slew, dither, park or reopen independently for each telescope. The sequences therefore follow a master/slave design. ARTEC is the master for operations involving the mount and roof, while REDCAT is synchronised as the dependent sequence.
Automated preparation and centring
Preparing equipment before darkness
Both sequences include a Power & Connect phase tied to astronomical twilight rather than to a fixed civil clock time. This gives the system a seasonally appropriate start. The start phase can connect the weather source, Safety Monitor, power switches, camera, focuser, filter wheel and flat panel; ARTEC also coordinates the mount and roof dependencies.
The electrical controllers are used selectively: if the 10Micron mount is already powered, the sequence does not need to restart its power-on branch. Some devices are connected before the observing window, allowing camera cooling or other preparation to occur without wasting usable dark time.
The Home Assistant Overview mirrors the equipment grouping. Setup 1 exposes the 10Micron power and pulse controls, camera, flat panel, focuser and OTA fan; Setup 2 groups the second camera and flat-panel controls. These dashboard controls support manual operation and status checks, while N.I.N.A. coordinates the observing sequence.
Before the main acquisition phase can proceed, the Safety Monitor must report Safe, which requires acceptable weather and UPS/incoming-power status. The intended order is power/connect → check Safety Monitor → open roof → establish the mount position → prepare the optics → image.
Opening the roll-off roof and preparing the field
ARTEC handles the roof and mount. The sequence checks shutter status so it does not blindly issue the same movement command when the roof already appears open. It also contains steps for the powered accessories and flat cover, telescope Unpark, target slew and centring.
Centring is not based purely on the mount's nominal pointing. N.I.N.A. takes a frame for plate solving, determines the actual sky coordinates and corrects the pointing until the configured tolerance is achieved. Using a luminance filter for the solving exposure makes more stars available than a very narrow Hα filter would. The N.I.N.A. RedCat plate-solving profile selects LUM, ASTAP and a three-second exposure; this is a device-profile setting, distinct from an explicit filter change within the sequence.
Autofocus is a separate operation: it measures stellar sharpness over a series of focuser positions and fits a curve to estimate best focus. Merely configuring L for plate solving does not guarantee the filter needed by a preceding autofocus operation will be restored correctly after an Unsafe recovery.
Synchronization and meridian operations
Coordinating ARTEC and REDCAT
The sequences use a N.I.N.A. synchronised wait mechanism so the secondary camera can align with phases controlled by the master. In normal imaging this prevents a REDCAT step from assuming a mount/roof action is complete before ARTEC has finished it.
REDCAT evaluates Wait for Altitude before entering the synchronization barrier. This keeps the slave at the appropriate phase of the sequence and avoids a timing deadlock while ARTEC controls the shared mount.
During acquisition, a coordinated dithering step is also used. Instead of one camera changing the mount's pointing while the other independently exposes, the shared mechanism coordinates a dither with the active imaging sequences. The exact impact of a dither on an individual subframe depends on where that camera is in its exposure cycle and the sequencer's synchronization state.
Meridian flip, focus adjustments and synchronized return
At the appropriate meridian crossing, the master mount executes a meridian flip. The acquisition sequence then restores pointing and focus before exposures continue; the secondary sequence waits for the shared mount operation to finish.
After a meridian flip or reopening, both camera sequences must remain synchronized; target centring, science-filter selection and exposure settings are checked before continuing.
Acquisition, filters and focusing
Filters, exposures, focus and the acquisition loop
The photographic part is organised into ACQUIRE SET containers, with filter selection and exposure loops. Multiple filters can have different durations: RGB frames may last a few minutes, luminance frames several minutes and narrowband Hα frames around ten minutes. N.I.N.A. records images in corresponding filter/exposure folders.
The sequences also use autofocus triggers, including autofocus after filter changes. A problem observed with ARTEC was the difficulty of fitting a reliable autofocus curve through a 3 nm Hα filter, even with an extended autofocus exposure. With few detectable stars, N.I.N.A. can record insufficient or poor-quality autofocus points. In some sequence blocks, errors are configured to continue rather than terminate the entire night.
An alternative focus strategy uses luminance with calibrated filter-wheel offsets for Hα and colour filters. Offsets must be measured and tested: selecting L for plate solving does not, by itself, perform or restore the preceding autofocus operation.
Restoring the science filter after recovery
The problem was not that using luminance for focusing or solving is inherently wrong. It was that the sequence could resume an interrupted acquisition without an explicit guarantee that the scientific filter in effect before Unsafe had been restored. Selecting L manually before the recovery procedure changed the filter state that N.I.N.A. then saw.
A manual luminance-filter selection in the reopening path makes filter restoration important. The next scientific exposure must use the planned filter, not merely the filter selected for focusing or plate solving.
Unsafe interruption and reopening
Interrupted long exposures
The weather event can interrupt an exposure that is already in progress. The interrupted attempt may advance or affect a sequencer iteration without producing a valid LIGHT file. When the session later resumes, the number of completed N.I.N.A. iterations is therefore not necessarily the same as the number of actual saved frames.
Instead of adding file-count reconciliation to every Unsafe/Safe recovery path, the chosen design accepts the occasional missing frame and counts the real files at the end of the night. This avoids adding filesystem and network dependencies to an already complex safety-sensitive sequence.
Normal Unsafe: temporary closure and a recovery loop
The When Unsafe branch is designed as a temporary interruption, not automatically the final end of the night. Unsafe can be caused by weather or loss of incoming mains reported through the UPS. The master sequence parks the mount and closes the roof while REDCAT moves into its coordinated waiting logic.
The recovery process then requires 600 seconds of continuously Safe weather. If Unsafe returns, the stable-Safe qualification cannot simply be carried forward. Once qualified, an Once Safe path evaluates whether a reopening is still justified. The sequence contains a decision that prevents reopening too close to nautical dawn; the intended margin is approximately 30 minutes.
Where reopening is permitted, ARTEC first activates the laser mount-position check. The laser changes a Denkovi DI only when the 10Micron is parked; that DI activates a Denkovi relay that closes the electrical permissive circuit for the roof-motion relay. ARTEC then commands roof movement and deactivates the laser after travel. The separate permissive contacts remain in the electrical path, so the roof pulse cannot reach its motor controller if the laser has not detected the parked mount, even if N.I.N.A. or Home Assistant requests movement. ARTEC prepares equipment, restores a useful pointing, performs the necessary autofocus and target centring and synchronizes with REDCAT. The interrupted acquisition sequence can then continue in its planned order. It is precisely this set of return operations that differentiates a reversible weather interruption from the separate, hard emergency restart.
Normal completion, emergency and reports
Normal ending versus forced emergency ending
The normal Ending is part of the planned observing night. ARTEC has the responsibility for shared mount/roof shutdown, while REDCAT handles its dependent accessories. Flat-panel closure, camera and controller disconnections, electrical shutdown and other housekeeping are ordered so the observatory ends in a known state.
A separate emergency sequence can be launched by Home Assistant through the Windows server. Unlike a temporary When Unsafe branch, this forcibly replaces the running N.I.N.A. session. Its priority is to park the mount, activate the laser position check, obtain the electrical Denkovi permissive and close the roof, then perform accessory and power operations. The mount-position interlock remains in force even for detected internal rain: the telescope must be protected from a roof impact before the roof moves. The sequence is intended for cases such as detected internal rain, an exceptional controller-output condition or an Unsafe roof that N.I.N.A. failed to close normally.
This separation avoids sacrificing the remainder of a recoverable night merely because the weather became unsafe briefly. A true emergency, by contrast, deliberately gives priority to physical protection over continuation of astrophotography.
How the actual exposure report is produced
At the conclusion of the main ARTEC sequence—and also near the end of the emergency sequence—N.I.N.A. invokes a local reporting service. The reporting approach is based on the actual LIGHT files stored on the server, grouped by instrument, target, filter and exposure duration.
This matters because the resulting totals represent frames available for calibration and integration, not promises made by the sequence planner. ARTEC and REDCAT can have different counts even though both started the same planned exposure blocks.
Cross-checking physical and software states
An observing night can be evaluated with several independent records: N.I.N.A.'s sequence logs, the actual image files, weather events, the mount and roof driver logs, Denkovi/Shelly history and Home Assistant notifications. Comparing timestamps reveals whether a frame was interrupted by Unsafe, whether the mount was parked before closure and whether both cameras met their synchronization points.
The engineering goal is recoverable autonomy: normal weather changes temporarily pause the night, a stable period of safety permits reopening, exceptional failures lead to a hard emergency, and the morning report records what was truly acquired. Every new focus strategy or synchronization change should be tested against that full chain, not just against a single successful exposure.