When the skies are bright, plan for darkness

Gotten a bit quiet here, hasn’t it? Well, here in the UK, it’s wonderfully sunny and bright. We don’t get proper darkness, and the planets are in an awful position, so imaging deep-space objects is a bit of a non-starter, or at least challenging. We’ve also had a run of crap weather, just to drive the point home.

I’ve been using the time instead to plan out my next astro-related project (though I may well push the execution out to 2020, just to make sure I have the cash to get it done right) – a fully automated roll-off roof observatory. The logic behind this is simple – my next “improvements” to my imaging system that I can make are:

  1. Upgrade the camera – already got a pretty good camera, so this means something quite high-end (>£3-4k), and would just get me more sensitivity/bigger pixels/larger field of view
  2. Upgrade the telescope – already got a decent Newtonian so a meaningful upgrade means either a high-end Newtonian/R-C astrograph (£3-4k) or a decent large-aperture apochromatic refractor (£4-5k)
  3. Add a second telescope – to do planetary imaging I could add a SCT or long-focal-length scope of some other sort, but the planets will be too low for the next couple of years for serious imaging, and it’d still be £2-3k of investment
  4. Update my telescope’s other parts (focuser, focus controller) and invest in tools (collimation, etc) – more reasonable investment (£1-2k) but just gets me slightly better images – this is my favourite option if I don’t do the observatory this year
  5. Build an automated observatory – easily doubles the number of images I can capture with my existing kit, thus acting as a massive force multiplier for my previous investments – but £4-5k at least!

So the biggest “bang for buck” is definitely the observatory, but only if it is fully automated. I’ve lost track of the number of nights where the sky was beautiful and clear, the clouds nowhere to be seen, ground and ambient temperatures low enough to make seeing incredibly clear – and I’ve been packing away the telescope at midnight because I’ve got work tomorrow, despite the further 7 or 8 hours of imaging I could have. And then there’s all the “well, it might be good enough, but…” nights – nights where the forecast says it won’t be good enough, but you might get lucky; often this involves going out frequently to stare at the sky, setting up if I feel optimistic, and usually being disappointed – but often not.

With a fully automated and remotely driven set-up the setup time is nil, as is the tear-down time. With the scope set up permanently, with the camera and other components mounted, there’s much more scope (no pun intended) for tweaking and tuning in advance of an imaging night, and fine tuning on cold-but-cloudy nights that just isn’t possible when you’re stripping the whole thing down each night. Being able to work in the dry and the day has a lot of appeal.

System-wise, full automation is pretty simple – you need a box with relays to drive motors and read sensors, a proper cloud/rain sensor (hard-wired to the relay box, so if any computers fail there’s a pretty dumb box responsible for shutting the roof when it rains), and a system capable of automating the selection of targets (what’s good tonight?), acquisition of images (frame the target, autofocus, guide and image), and the observatory start-up/shut-down. I’m most of the way here – I need the relay box and auto-focuser. The rest is already ready – I’ve been using INDI/Ekos/KStars for a while which can do all of this. The main INDI instance for the observatory will run on a 1U server in the observatory, with an INDI server on a Raspberry Pi 4 strapped to the telescope doing actual image acquisition and telescope equipment control. This makes the pier-to-desk cables simple – 12V for power, USB for the mount, and an Ethernet cable for the rest, with just 12V and Ethernet onto the telescope itself.

Making a plan

So, the objectives of this build are:

  • Full automation – but at a minimum, a roll-off roof which can open and close under all circumstances for safety – so I can program the observatory to image opportunistically
  • Imaging-stable pier, with room to expand – just the one pier, but room to set up a second non-isolated pier for a small solar/planetary telescope (isolation is less critical for these applications)
  • “Warm” room with enough room for a server rack, desk, chair and a little storage – somewhere I can sit while setting up
  • Good visibility down to ~30 degrees everywhere
  • Strong enough to resist opportunistic forced entry and 100mph wind when closed

Beyond this – it’s basically a shed! So I’ve started by getting a bunch of books on shed design and construction and reading them. My day job at the moment is (mostly) telling people how to properly build a fibre optic network, so I know a reasonable amount about concrete, aggregates, rebar, admixtures and slab design. Making a good solid observatory is mostly about mass, just like in acoustic isolation design, and I’ll be using almost an entire ready-mix concrete truck worth of C40 low-moisture concrete to pour the base slab and the (isolated) pier. The framing and design of walls, floors and doors is all fairly simple, though benefits from careful planning to make sure all the services will work and the structure remains rot-and-rat free for a few decades.

Some basic renders of the general layout – working floor-up. Note the duct from pier to warm room to allow for cables to reach the telescope safely

The tricky bit is the roll-off roof – I need to keep this building rodent-proof and ideally near-airtight to aid in humidity/temperature control. I will use forced, filtered airflow for cooling with a positive pressure maintained to minimise dust ingress. Active cooling with the roof shut will help cool-down times and avoid any kit getting too hot in summer. This means the roof needs to seal well onto the frame when shut. I also need to be able to shut the roof at any time – that means any internal rafters need to be minimal or non-existent, so the telescope doesn’t have to be “pointed down” to let the roof pass. This means when the mount fails or is unsure of its position the roof can still shut safely to keep the rain out. The roof needs to roll back enough to give good visibility, so the whole thing has to roll onto rails that extend beyond the back of the warm room. To further improve visibility and keep rain off the rails, some of the side walls will be mounted on the roof so the walls “lower” as the roof rolls off. There’s a lot of complexity in this (and it has to be something I can build), so this is taking some time to work out.

I’ve started designing in detail in Autodesk Fusion 360 – while I’ve used Sketchup for this sort of thing in the past, Fusion 360 in Direct Modelling (non-parametric mode) is about as user-friendly and can produce much prettier outputs as well as decent engineering drawings.

An early rendering of the pier and shuttering for the initial concrete pour
An early drawing with some detail/section views to show the base layout and design – the deep, chunky base should help isolate the pier from surface vibrations/movement, and the really deep and heavy pier root should by virtue of being heavy do the rest

I’ve also reconstructed my current telescope and mount with photogrammetry so I can build a digital model and check the motion all works – I haven’t gotten around to tidying up the mesh into some simpler models, but it’s a great reference for getting the dimensions and motion right.

Location, location, location

The other question is where to put this – I dithered quite a bit and in the end took a lot of level photos around the garden at twilight with a Ricoh Theta S 360 degree camera at roughly my telescope’s aperture height. With the moon visible in each photo and knowing where and when I took the photos, I could align the photos to north with a fairly simple Python script which spat out a nice set of data for horizon plotting.

Plotting horizons straight out of images. Probably should release the code for this…

It turns out there’s only a few places where I don’t enjoy visibility to 30 degrees pretty much everywhere, so I decided to plug the panorama for my favoured location into Stellarium – this turns out to just involve having a panorama with a transparent sky and a small .ini file to set north properly.

My observatory’s home, Stellarium-ready
… and loaded into Stellarium, so I can see how things will look – spending all the time with Photoshop’s background eraser to get the trees properly semitransparent makes a big impact on the visuals of this (though in summer they’re somewhat more opaque!)

The chosen location makes power and network connectivity simple enough – with 25 metres of mains cable and single-mode fibre I can connect to proper mains and Ethernet, only one switch hop away from my storage arrays.

Security is a concern – that field is adjacent to a footpath, though set back from the road, and there have been break-ins in the area. Other than making the building fairly secure against “opportunistic” crooks – reinforcing the door, lack of windows, and a solid lock – there’s not a lot that can be done. PIR sensors externally won’t work due to the abundant wildlife, so a combination of internal sensors and an alarm to make a racket if someone does force the door or climb in through the open roof will have to do. CCTV around the perimeter might work but could work just as well as an attractant as a deterrent, and wildlife would probably again make alarming impossible. I’m also planning on using a worm geared or lead screw based roof mechanism, which should be very hard to force open.

Making plans

I took the view early on in this that I wanted to build this myself. I’m still not 100% sure about this, but I think it’s a reasonable project and something I should be able to do! I am budgeting for some help, though, and will have to hire kit in regardless – a mini digger for the groundwork, compactor to pack down aggregates, concrete vibrator to settle concrete in the forms, etc.

I also need planning permission. I started with a footprint that wouldn’t normally need it, so long as the building isn’t tall – but I’m in a conservation area, which means “permitted development” doesn’t really apply. I’m not concerned about getting planning permission – it’s a small building in an otherwise empty field (except for a shed we’re going to remove) and will blend in just fine. Having to go through planning permission also means I can relax around some of the limits that I’d otherwise be avoiding.

Working through the material costs there’s easily £2k, maybe £3k of materials – labour would be another £1-2k atop that if not more. That’s quite an investment, and I’m really keen to make sure that everything about this is right – giving up power to a third party feels risky. It may be that when I get the design done I sit down with some local builders that I trust and see what they say.

The first step remains the plan and design, which is taking time – but I think time invested here is time well spent. I may not start until later in the year, or even early next year – one more winter without it wouldn’t be the end of the world. It’s going to be a fun project if I can get the plan right!

MORE DOMES

Fans of domes will be wondering why I haven’t just dropped £3k on a nice big Pulsar/insert-vendor-here dome. The answer is simple:

  • It’s not £3k, it’s £7k by the time you’ve automated it
  • It’s impossible to insulate the roof nicely – you end up slapping neoprene sheets up with glue just to stop condensation build-up raining on your scope
  • They’re relatively small and uncomfortable to work in unless you get big ones which are even more money
  • They only allow for a single telescope
  • They’re definitely harder to get through conservation area planning permission committees

I’ve looked at a few other dome designs and while there’s some good contenders they all have similar problems. I did consider making a “clever” geodesic dome – something I could build pretty cheaply but which would still have decent wind resistance – but automation remains the problem. Ground-level domes (where the whole structure rotates, rather than using a rotating section on a cylinder) make the construction simpler, but the bearing and rotation mechanism have to cope with increased gravity load and all of the wind loading. Cylinder-style observatories have similar problems.

The round/dodecahedral designs of these structures also make literally everything harder. Want to bolt a light to a wall? It’s not flat, so if you want it level/flat you now get to make a bracket… weatherproofing, insulation, and more all get more complicated. Having four flat walls which never move makes life simple – mounting insulation, cable entry glands, coolers, dehumidifiers, fans/filters, lights, shelves, etc is all so much simpler.

So – no dome here for now.

And another thing…

While we’re building a light-shielded box in a quiet location with power and networking, what else could we do? I’m also going to include infrastructure to support a small ground-level dish and motors for radioastronomy, as well as some mounts for meteor spotting cameras, an all-sky camera, and a weather station. I won’t have all this on day one, but putting a little extra concrete in now is way easier than doing it again later, and it means I can put in cable ducts to make wiring it up simpler. The cost of the pads, etc is tiny and turns those future projects from a pain into something much simpler.