The Exterior…

  1. Groundbreaking is one of my favorite parts of the project…..second only to completion. And be aware that you will probably never actually ‘complete’ an observatory so it’s an elusive emotion. There is always one more bit to add, new idea to try or thing to improve. In contrast, the construction has a very definitive beginning……typically starting at the very bottom and involving shovels or excavation equipment. Station 1 was no exception. The first step was to mark out the locations of the two footings.
    The observatory has a 15’W x 12’L footprint. Each of the 2 sides of the 12′ length is a triple 2 x 12 x 12 girder. Each girder attaches rigidly at one end to the existing porch gable end using a HUC212-3TF Simpson String Tie hangar (each able to support over 2 tons) and the other end sits atop a 4″ lally column (each able to support over 9 tons). Each of the lally columns sit atop a 16″W x 48″D footing. The position of these footings is where this phase began.
    To get the position of the footer centers I dropped a plumb-bob from the center of the girder joist hangars down to the ground….I connected these two points with a line. This will be one side of a square. Then I drew a line from each these 2 end points (exactly perpendicular to the first line) out 11 feet. The end points of the 2 11′ lines are the center points for the 2 footings. Verify this by confirming the corner to corner measurements (across the square’s center) are equal. The 3/4/5 ‘triple rule’ is also a good way to verify. I left a foot above grade on my footings as well to be sure that if a lawnmower every went astray it would hit the massive footing instead of a thin lally. I did the dig (and after hole inspection) the mix, pour and moved on.

  2. After the footings cured for a couple days, we nailed the girder hangars to the existing gable end. Next a lally went up to the same height as the hangar bottom. The lally needed to be cut which was quick work with an abrasive metal cutoff wheel in the circular saw. The height of the lallys was such that the observatory floor would be at the same level as the existing floor in the ‘attic’ space over the porch. Once sanding, the girder was placed on the top with the crown side up and the process repeated on the other side. Two long 1 x 2 strips were nailed underneath the girders in an ‘x’ pattern to prevent racking. A 3/4/5 check was made here as well. A length of 2 x 6 is laid atop the girders to provide a good nailing surface for the framing above. Lallys were painted white with Rustoleum paint. At this point the weight bearing members are in place and the rest of the structure can safely be built on top.

  3. Floor joists went up quickly. 2 x 8s were used to keep consistent with the existing floor the observatory was connecting to. These were placed 16″ on center and started 16″ off the outermost existing one…again to maintain conformity and consistency of the overall structure. Be sure to place these crown side up. Once all joists were in place, the 3/4″ T&G CDX subfloor went down. Seams were staggered and screws were placed. I made sure to extend half of the center 4×8 sheet into the existing porch attic so the obs structure is rigidly attached to the existing house.
    At this point the platform is fixed so a first look at the horizons from the final elevation is appropriate. The perch provides a ~340 degree view with a small obstruction at NNW that rises up halfway to Polaris. Small enough that most of the sky being blocked at one point of the night will come into view later. The ecliptic however is fully accessible and the horizon is low with only 2 small light domes at the east and south west.

  5. Now the roof framing begins. Plumb / seat cuts can simply be traced off the existing rafters eliminating the need for calculations. The 2 x 12 ridge beam is raised and with a pair of 2 x 10 rafters at each end. The remaining ones are all filled in and face nailed into the ends of the floor joists forming a rigid triangle truss at every 16″ interval. The floor surface to the bottom of the ridge is about 6.5′. This means walking down the center is fine but bending is needed to get into the flanks. However, ALL the area under the dome is wide open with the floor surface to dome peak height at about 12.5′!
  6. With the main framing now complete, the roof deck is laid. This is 1/2″ OSB placed with staggered seams. The ends of the ply are slid to the center of the first rafter pair. This leaves a few inches of gap to the existing shingles on the porch roof (more on this later). Roof continuity is maintained across the porch/observatory transition. A 12″ overhang is left past the last rafter for the eaves at the other end. A long 2 x 6 rake board is placed under the ply at the gable end (12″ out) down each slope. This is screwed down into from the top of the ply. Ice & water shield is rolled across the entire roof perimeter and 30lb felt (with a 3″ overlap) over the entire roof deck. Drip edge is placed across the bottom of the roof and up the eaves. Lastly the shingle courses go down. I was lucky to find the same shingle from the same manufacturer in the same color….but I can still detect something is still a shade off….perhaps the existing shingles had some color bake off. Regardless, you would never notice unless you really studied it.

    So at this point we have a really nice 12′ standard gable roof extension. No signs of anything more interesting. However, from here forward, the build has no reference. There are no books on construction techniques and practices for raising a cylinder up out of a triangular prism. No kits from Home Depot or Lowes and no yellow page services. To go on from here is really a leap of faith. A leap that if miscalculated, could be a very expensive mistake. Here is where the SketchUp model turns into a life line.


  8. This next step required a bit of mental fortitude. Slicing huge holes into a perfectly good roof forces you to convince yourself it’s the right thing to do OR to just roll the dice and accept all the risk… the end I’m not sure which way I went. Despite my checking and re-checking of the cut there was still a nagging feeling of uncertainty and pending doom.But I’m getting ahead of myself…..let me give you some rationale on the goal here.I considered 3 options for the dome support structure, and with each, the water proofing needs, the resulting rigidity and strength of the frame and the space it consumed.
    The options were:

    1. Build a full framed (top ring to base) cylinder on the floor of the obs…and then later build the roof around it. Think of a cardboard toilet paper tube standing vertical. The issues with this approach is that the rafters and roofing would be a pain to conform around a cylinder. I also didn’t like flashing against it to prevent leaking. Having the framing going down to the floor also took precious sq footage from within the obs.

    3. Finish the roofing structure (framing and decking), cut the opening and then build in the full framed cylinder (top ring to base) on the floor. Flash it and then shingle. This eliminates the complexity on confirming the rafters to the cylinder but we still have the issues of flashing against it and the over-consumption of floor space.

    5. Finish the complete roof including shingles, cut the opening and then build just the upper part of the cylinder and sit it ON the roof. This means no conforming the roof to the cylinder. The need for flashing is eliminated and the framing within the obs is drastically reduced. It does introduce an added difficulty though….the fabrication of complex wall stud cuts….cuts that aren’t only at varying angles were they meet the roof (your typical 30, 45, etc degree cuts) but are also off-axis to the plane of the board. These were hard to figure out but fortunately there weren’t that many to deal with. This option also requires a waterproof sill for the cylinder wall….probably the most aggravating part that is discussed below.

    After all the pro/con considerations I opted for C.
    In all 3 cases above the roof opening is an ellipse. An ellipse is the parabolic curve created when intersecting an angled plane (the roof) into a cylinder (the dome wall). Whole ellipses are simple to draw but in this design we were’t that lucky. The whole ellipse is actually folded at the ridge resulting in two half ellipses….one down each side of the roof. This made things much harder.

    I approached the ellipse calculations a few different ways to be sure this opening was right. I had only a 1″ +/- tolerance on the cut.

    1. Mathematically – I sketched out a cross section of the roof and cylinder, and using Pythagoras, calculated all the key measurements to determine the lowest point of the ellipse (which is also the length from the center of the ellipse to the furthest point out). Using the length above I modeled a geometrical conic projection in Sketchup with the major and minor axes to get the exact length of the arcs. I also verified these dimensions using the standard ellipse formula
    2. Since it was near impossible to physically draw the 1/2 ellipse down each roof slope, I decided to cut out templates to later be used as a guide to scribe the arcs onto the roof. The templates were easy to prepare. I simply laid out 2 sheets of plywood end-to-end. The butt joint would be the semi-minor axis of the ellipse. Using the ‘string method’ I traced out the full ellipse onto the plywood and cut out the two pieces. This gave me 1/2 an ellipse that I would use to guide my cuts into the roof.

    4. I also verified these templates before using. I raised the actual dome top ring onto a plank supported by ladders. I then positioned each template under the ring at 45* butting the straight sides. Then, using a plumb-bob, I verified the ends of each ellipse were directly under the outside edge of the ring.

    Ok…so we have our template! I shuffled it up the ladder and lined up the straight edge with the ridge of the roof. I also centered the template to the mid-point length along the ridge to assure the hole would be in the exact middle of the new structure. I tacked it down and scribed the arc onto the singles. I flipped it over onto the other roof slope and repeated the process. Then, one last deep breath, and I plunged in the saw. Quite an interesting shape at the conclusion of it all. Last step was to shore up the cut ends of the rafters with 2×6 kneewall studs. The kneewall studs were nailed into the obs floor at the bottom and the top was cut exactly where it met the underside of the roof. They were then side nailed into the rafters. This will add support to the rafters and the dome/cylinder wall that will sit atop them.

  9. Now the real tricky part begins. With the opening now present, the cylinder wall would be constructed like any other wall as far as basic components….a sill plate, vertical studs and finished with a top plate. A couple challenges were immediately evident:
    • The sill plate should not be wood or any other product susceptible to moisture. Immediately TREX came to mind….but how to form it into an arc?
    • The studs (as mentioned above) would all have different base cuts angles as well as an off-axis cut for most

    I spent a lot of time thinking about the sill plate. This would be a very likely point of failure if not properly planned. Wind would regularly drive rain into the roof/wall union. Snow would also collect there and slowly melt. The sill plates could easily be cut from Azek sheets…but this would be very expensive. I returned to TREX idea and came across some heating/bending techniques. Initially this seemed quick and easy and perfect…but as is often the case, quick and easy became complex and expensive. A heating jig needed to be created from joined sonotube. Then the plank would be inserted, both ends capped and a 60k BTU kerosene heater fired into the center via HVAC tee duct. I ran this setup for hours at the recommended 245* but never got any flexibility. I ran it up to 300* and 12″ or so of the center (by the heater) started to give a little but that was it. At higher temperatures it started burning. On Youtube they demonstrated the process flawlessly. I tried and tried but couldn’t reproduce it. I even built a plywood coffin as one guy did…but the result was the same. Fortunately I was able to return most of the components to Lowes so the experiment didn’t incur any cost. It was apparent that the only way to achieve a TREX arc was to cut the plank into small segments, arrange them on a plywood arc and screw into place. A few hours later we had 4 arcs fabricated (two for each sill on each side of the roof). They were then screwed in place on the roof along the outside of the opening.

    Next were the 2 x 6 studs. These needed to be flush with the outside of the sill so the cylinder sheathing would lay flat against the wall all the way down to the shingles. I chose to position them so the wide side was parallel to the sill (as opposed to perpendicular like most wall studs). This provided more nailing space for the sheathing later. I used 7 studs on each side starting at the bottom which was an easy 45* cut on the bottom (all studs were flush cut on the top as a plywood top plate will sit here). Moving out the studs became harder to figure. There wasn’t a technique I was aware of to easily determine these unusual cuts so we used an angle finder and got pretty good results.

    In parallel with the stud work we cut a bunch of plywood arcs from 3/4″ OSB to form the top plate. The rings were drawn by placing the dome mounting ring down onto the sheet and simply tracing it out. We cut out 4 arcs to make the 1 round top plate. There will be two plates stacked on top of the studs so we have 8 arcs total. The arc seems were overlapped for strength. The metal mounting ring (with a “T” shaped cross section) will rest on top of, and inside, this wooden plate. Like before, this plate will also sit flush with the outside of the studs so the sheathing sits flat. With the top plate in place the cylinder framing is complete. The finished shingle roof extends up under the TREX sill and beyond it for about an inch. Silicone sealer will be run on both sides of the sill making wind/water/snow penetration impossible.

  10. Again we arrive at another decision point. Do I rent a knuckle boom capable of lifting the dome the required 20′ into place (cost=$300) or do I setup a tall ladder and push the dome from the bottom while my brother pulls an attached rope from the top (cost = $0). There are two labels sometimes applied to me that I cannot dispute…. cheap and stubborn. With that, the ladder and rope were put into commission. This seemed to be a low risk idea at first. The thought was “if it doesn’t work out we can just lower it back down”. So up it went. I was on the inside pushing up…picture a large white turtle….and my brother pulling rope fist over fist above me. It started well.
    Then, suddenly, the shutter opened up and over my back making a huge bang as it crashed to a stop jolting the whole dome and the ladder on which I stood. Things got serious now, press on or lower down? Since we were 1/2 way there the decision was made to push on. At about this time my wife arrived home, and with a loud, concerned voice said I was crazy and I was gonna kill myself…..more motivation to get er done. Progress was slower and harder and the 300 pound dome continued to resist. Finally one end was making its way over the metal base ring. It was now that my brother and I had first eye contact and we both realized there was very little more either of us could give to the task.
    Just then, in the most unbelievable demonstration of ‘perfect timing’, my neighbor rushed in like a hero and saved the day. With not a second to spare, he pushed hard on the bottom of the dome rescuing my brother and I from certain collapse. Then, from the ladder beneath me, he single handedly gave it the heave it needed to overcome the tipping point, which it did, coming to rest exactly on the marks. “See honey, I told you this was no big deal” I said with the little breath I had remaining.


  11. With the big work done I was looking forward to the easier task of ‘skinning’ the dome. Again there were options to consider.T11 was a product some colleagues used to build their observatories….but having this on a house in the past I knew it would quickly rot from any standing moisture. Additionally, it probably wouldn’t easily bend the required 30* per board and It would also require periodic painting.Baked enamel aluminum was a strong candidate as well. It came in 4×8 sheets for about $100 each (+ shipping) for 14 gauge…which is about a 16th of an inch. I liked the idea of a smooth glossy finish that would last a long time. This is exactly the same product as typical soffit flashing….just in large sheets.But slumming around Lowes one day I came across something called FRP…Fiber Reinforced Plastic….$33 for a 4×8 10th of an inch sheet and in stock! FRP is simple to work with, it’s very flexible, cuts like butter with aviation tin snips and has a 50 year life expectancy. It’s used often in marine applications so it performs well around moisture. I grabbed 4 sheets and a box of white screws. I measured down from the bottom of the mounting ring to the shingles every couple inches along the cylinder sill. I transposed these lengths to the FRP and traced an arc. I cut it out and up it went. I chose textured side out as it had a glossy finish and better matched the dome. I overlapped each section by 6″ and placed full sheets at the highest parts of the wall and pieced in at the ends at it approached 0. This gave a nice symmetry. When all the FRP was up, I applied a can of PlastiKote (T-45-6PK) “Acrylic Clear General Purpose Premium Enamel” to prevent the FRP from yellowing under the sun’s UV rays.

  12. The last bit of work to get everything closed up (so work can start inside) was the gable end wall. This was framed with 2 x 6s. It was a simple matter of dropping a sill plate across the far end of the floor (flush with the last rafter) and then standing up the studs, 16″ apart on top of it.Finally…something that (from the outside at least) looked like an observatory. Queue the sense of accomplishment.

  14. Now for another nail biter….cutting out the central 8′ of my observatory ridge beam. Theoretically the 2 complete pairs of rafters at each end are plenty to maintain the shape of the roof. As well, the partial rafters are doing their job transferring load vertically (via the kneewall studs) onto the floor and joists below. And the plywood roof deck will assure there is no shift along the direction of the ridge. Repeating this over and over, and eventually convincing myself, I again took the unnatural action of destroying what I just built….without any certainty of the results. But the cutout fell and everything held up. The structure didn’t seem to miss all that ‘extra’ wood. As a precaution I rigged a cable-turnbuckle and two eyehooks (one into each ridge beam end) which I keep in place when the observatory is not in use.

  15. One particular aspect of this build bugged me from the beginning and I was never quite satisfied with my few ideas on how to address it. The new and old roofs do not overlap or interleave. They are completely separate and independent of each other. The shingles butt together cleanly, and it is hard to notice any seam, but in fact there is one. Weather being crazy (and seemingly more so lately) I struggled with a way to make this positively watertight. Since the existing roof is well nailed and fused from the sun, there wasn’t a non-disruptive way to get something underneath it to use to bridge the gap. So kept thinking how to ‘cover’ the gap from the top. I considered a long piece of aluminum or other light metal with a “T” shaped cross-section that could be dropped into the gap and extend a couple inches over either side of it. This would mean an obvious, few inch wide, transition….which I was trying to avoid. So under shingles was ruled out and on top of shingles had no good answer so I wondered could I deal with any weather penetration after it came through. This turned out to be a much better option. I ripped a 3″ PVC pipe in half lengthwise ending up with two gutter like chutes. I pushed them up to meet the inside/bottom surface of the shingles and put some long screws underneath the gutter (and into the rafter next to it) to keep it there. I did the same down the other side of the roof. Now, anything that falls through the crack, will be caught and quickly exited from the observatory (the end of the pipes hang above the vinyl soffits). Out of curiosity I inspect these gutters from time to time after storms and have yet to see any moisture there. But it’s good to know we are prepared just in case.