Chapter 4 - Fuselage Bulkheads
In chapter 4, the first actual aircraft parts are constructed: The fuselage bulkheads. F22, to which the canard will one day attach; F28, the instrument panel; the front seatback; the landing gear bulkheads that the fixed gear attaches to; and the temporary and permanent firewall components.
Before building anything, I copied the chapter 4 bulkheads into CAD and modified them to fit the revised fuselage shape I am building (see Design for details). Once modified, the drawings were e-mailed to a local print shop, and the result can be seen on the right. The pages on top are the drawings that come with the plans and are only big enough to depict a quarter of the instrument panel each. For many parts, only one side is shown, you have to copy / mirror for the other side. My modified drawings are printed in full, and F22, F28, Instrument Panel, Firewall, and cut-outs for the seatback all fit on the one huge sheet.

Per-plans, the seatback is the first bulkhead to build. However many builders recommend starting on F22 first, as it is easier.  I did F22, then the instrument panel, then the seatback. I'm not sure why the seatback is considered more difficult, I found it to be the easiest. Maybe it would not have been so if I had done it first?

F22 & F28
This is also my first time cutting PVC-200, the high-density foam core. This stuff is TOUGH! I had previously cut the lower density PVC core for the seatback with a knife, but struggled to cut much deeper than about 1mm into the high-density foam! I ended up using a jigsaw to rough cut the foam, then sanded (36grit) to the final dimensions.
F22 also has a doubler, which is simple in shape, except that the bottom edges taper down to zero to form a smooth transition to the main piece. Again, I sanded these tapers. This foam is tough stuff, it took awhile to get everything sanded to the precise shape (which was conveniently marked in the foam with 1mm surface cuts!).
I cut F28 at the same time.

I applied micro to fill the surface first on the seatback. The lower density foam has larger surface voids and took a fair amount of micro to fill. The high-density foam for F22 and F28, however, took very little, due to the very small surface voids. Make sense. Since I am hard-shelling, this was all done in advance, left to cure, then sanded with 36grit to ensure good adhesion of the following glass layups.

Time to glass the first side. I had recently 'perfected' my vacuum system on test pieces, having had trouble achieving good vacuum and / or drawing out excessive epoxy.  I made only one change before using it here, I connected a different vacuum port, one with twice the port area, which I figured would help compensate for any leaks. As I would come to fully understand, vacuum bagging is very, VERY sensitive to your setup and exact method!

The I first laid up the side with the doubler, as the other side has a large number of plies, and so would be trickier as a 'first' layup. The layup went OK, without too much fussing with keeping the glass straight and so on.  Getting the subsequent glass layers right once they are sticking to the underlying layers was a pain, especially in the narrow areas where the glass overlaps from the doubler down the taper to the main piece. Lack of experience, I guess.
Then I added peel-ply overall, paper towels to soak up excess epoxy, the plastic vacuum 'bag', and on top some weight to keep it all flat and ensure the doubler was firmly pressed on. Vacuum on, walk away and wait.
I waited overnight. The next morning I returned to find the workshop a bit foggy; the vacuum pump had dripped oil onto the motor, which had vapourized off to fill the air with oil vapour. Nuts, I'll have to work on that.
I removed the weights, plastic, then epoxy-laden paper towels and peel-ply. This revealed a horror scene below; White, air-filled layups.
Mild disappointment does not describe how I felt at that moment.
The vacuum had drawn most of the epoxy out of the layups, leaving junk, epoxy starved glass below. 
The temperatures overnight had been hovering not far above freezing, and the epoxy had cured hard, but brittle, needing more heat to develop strength. Thanks to the brittle cure, it turned out it was relatively easy to peel the layup off the foam, including separating the doubler.
The foam was undamaged, a few minutes sanding restored it to layup ready status. The glass, partially cured and epoxy starved, was easily peeled apart and folded.
I put the glass removed from F28 (yes I ruined that at the same time) and folded in half in my epoxy cabinet to finish curing in the 30c heat. The next day it had indeed cured to the expected strength, so no problem with the epoxy or mixing.
The glass removed from F22 I stapled to the wall. A reminder not to mess up, or make unproven changes to my process on flightworthy parts!

Before making a second attempt, I constructed the heat tent (see workshop page for details) to ensure I could get adequate temperatures to get a full-cure even during winter months. This was not the major problem the first time around, but I did need this solution if I was going to continue doing layups in cold weather. You may have read on another page on the site that the epoxy I am using, ADR-246TG, needs a bit more heat to cure properly, and mild post-curing to develop full strength. Post-curing will come later when I have larger assemblies, but I still need to get a proper cure in the meantime. 

I decided to abandon the vacuum-bagging for now. I had spent a lot of time getting it ready and a procedure worked out to use it, only to discover than an apparently minor change can cause it to ruin parts. I'll come back to that later, but for now, I want to make progress on actual parts, and there are other ways of obtaining almost equally good fibre-resin ratios.

Attempt 2: Same again, but this time no vacuum. I applied peel-ply overall and then 125mu plastic, through which I squeegeed out excess epoxy. Then weights on, heat-tent up, and overnight cure.

The weights are 20l containers. I filled them with about 8kg of water each (1l of water = 1kg) in this instance. I obtained 4 of these containers for $5, and I can choose how much weight each has by varying the amount of water inside. I place a smooth, flat piece of wood between the weights and the layup to ensure a flat, even cure.

This time, the result was much better. I had drawn a little bit of air into the layup over the doubler, most likely from removing and re-adjusting the weights I had put on it, causing it to compress - expand (drawing in the air) then compress again. Note to self: don't do that. But, it isn't much, and it's within specified limits.

I didn't knife-trim at the green stage. These days everyone claims it is easier to get out the multi-tool after complete cure and trim then. Hmm, yeah that does work but mine didn't cut 'like butter'! I'm not sure if it was the blades I was using (quite possible) or that the toughened epoxy I am using just doesn't cut as easily as more common epoxies. It was particularly laborious cutting the thick layups on the other side of F22!
Next, I did the other side, with the gazillion layers of glass built up on the sides. I took plenty of time and care to get the layers on the sides just right, though there was slight variation in the dimensions of the pieces, so there is a slight taper resulting at the edge of these pieces. No problem.

This time I applied heat (using my wife's hair drier) through the plastic while rolling and squeegeeing out excess epoxy. It worked a lot better, the epoxy flowed to the edges more readily, and presumably more of it. Unfortunately the plastic I am using didn't much like the heat, and expanded and caused some creases and wrinkles. Minor ridges on the layup resulted (they are just epoxy, and can be sanded away). Finding affordable plastic sheet in a suitable thickness that isn't also pre-creased everywhere has been very difficult!
After cure, I removed the peel-ply, and somehow in the process caught an edge of dry UNI overhanging the edge of the foam. It peeled the bundle of fibres out, right up through to the top of the part!
Once again, the feeling was not mild disappointment.
I took the problem to the online builder's community, and the suggestion was that, as it was just a narrow strip of glass off the top layer, in an area where there will be lots more glass added anyway, that I should just fill the groove with flox and sand it smooth and level with the surrounding glass. I did, and it's fine. Just annoying. I'm discovering new and novel ways to muck things up!

I made F28 at the same time as F22, and it experienced the same woes with the first layup. It turned out well in the end as well. It's just much smaller than the other bulkheads, and tends to be easily forgotten. :-)
Instrument Panel
With F22 and F28 finally complete, I started work on the instrument panel.
F22 was almost stock, just a bit wider. The instrument panel is a bigger departure from the plans, however.
The leg-holes are 1" taller in addition to being wider. The actual instrument panel area is increased by making the raised switch-panel area almost full-width. This gives enough room for a 10" EFIS one day. I'm tall enough that I'll be able to see over it fine, and my wife can use an extra cushion as needed!
Finally, I took some inspiration from Dennis Passey and angled the instrument panel back 12deg, so that the instruments will be seen more 'square on' by the pilot. Commercial airliners have this, so it must be a good idea! I didn't cut the top off and re-fit it on an angle, I made mine with the angle built-in, using a simple jig made with some spare MDF I had lying around. 

The first instrument panel layup seemed to go well, however, after cure it became apparent that I had been over-zealous in squeegeeing out excess epoxy, and I had a number of air bubbles and dry areas that were out of spec. I elected to peel the layup off and try again.
This time, with the heat tent in place, the glass had cured properly, and it was a much more difficult job separating it from the foam. With the aid of a scraper pushing between foam and glass and two bloodied hands later (it turns out that glass is sharp) I had it off, with just minor gouges in the foam to repair.
I repeated the layup, and this time was more cautious removing epoxy. Still a couple of minor air bubbles, but well within spec. I suspect these bubbles were connected with bubbles in the plastic on top, which may have helped lift the glass slightly.
I turned the heat up in the heat tent, and added additional heat with a hairdryer (my wife is tiring of this misuse of her equipment by now) running, bringing the temperature up over 50c. It cured in just 2 or 3 hours, at which point I turned it over and laid up the other side. No plastic layer this time, I just squeegeed over the peel-ply. This seemed to work just fine, and avoiding excess pressure seems to avoid introducing air. It also means that creases and wrinkles in the plastic can't full up on the glass.
Both sides were completed and cured in half a day, no problem. Maybe I'm finally getting the hang of this stuff!
You can see to the left how readily the ADR246 wets out (and runs on angled parts). No need to stipple it through the glass, even with 2 plies of BID and one of UNI.  Just spread it around, adding more as needed to achieve a full wet-out. Once the glass is all wetted out, I add the peel-ply on top, and squeegee it down allowing it to wet out using excess epoxy from below. A bit of heat with the 'heat gun' (hairdryer was cheaper, had the same heat output, and has a neat tip to spread the heated air out flatter) continues to assist the peel-ply wetting. I add a little more epoxy to fully wet the peel-ply as needed, making it transparent so I can see any air bubbles to work out of the layup.

Note in the second image to the left that the temperature sensor is being supported just above the layup by a plastic clamp. This feeds the temperature controller to which the heat-tent heater is connected, providing a steady temperate of my choice at the layup. In this case, I've also left the hairdryer running inside to raise the temperature further, thus achieving over 50c.

I started knife-trimming once partially cured with these layups. With a controlled curing temperature, it is easy to predict when I need to come back to trim. This greatly reduced the amount of wear on my Dremel cutting wheels!

At this point, the plans have you create stiffeners / wire channels on the back of the IP. This is reportedly quite tricky to achieve, and a number of builders have developed methods to make this easier. 
I noticed that many builders later cut these and in some cases the whole panel above the leg holes, rendering the effort creating those channels somewhat redundant.
Instead, I added just straight stiffeners, with no lip to form a channel. If I really need them later, it should be trivial to form a U-channel and glue it exactly where I need it.
 The top stiffener I also moved to the top of the panel. This leaves plenty of uninterrupted area for a big EFIS one day. I figure that if it gets in the way there, I can just trim it off later.
I microed the foam for the stiffeners in place, held in what will be the horizontal plane with lots of clamps, and supported by wood from the angle jig (which were therefore cut to just the right angle for the horizontal stiffeners).
Once cured, I added a micro fillet before adding the glass. I created a 'pre-preg' by wetting out the glass on foil, covering with plastic and squeegeeing / rolling out excess epoxy. The entire sandwich was then cut into the desired size strips. After removing the plastic, the glass was then applied to the foam (which I had micro-filled before attaching to the IP). The layups are quite slippery when applied in this way! I had to peel off and re-apply a couple of times when I first did this, as I was inadvertently sliding the glass down and off the foam!
I did 1 ply BID on the horizontals and 2 ply BID on the verticals. I created micro fillets where the stiffeners meet, allowing the glass to wrap around the corners. Nice and strong. 
I put peel-ply strips along the glass edge where it ends on the back of the panel, but failed to add enough epoxy to create a properly smooth transition. The pre-pregs didn't have excess epoxy to contribute, so I should have 'painted' and bit of extra onto the peel ply.

I also added a bit of micro around the 'inside' of the cut-out areas of the IP and F22 to 'protect' the exposed foam in these areas. It only took a very small amount, and I had some excess micro from other procedures anyway.

​After trimming the extra glass extending above the stiffeners, I sanded the edges smooth. IP complete!
Front Seatback
This is a much simpler shape to create than the previous bulkheads and is a much lower density foam, so much easier to cut! No power tools used here. I cut the foam to size with a knife, then trimmed the 45deg angles top and bottom by running it through the bandsaw. Easy.
I had decided before starting to hard-shell all foam parts, unless impractical. The first foam I did this with was the front seatback, and I was initially surprised and just how much micro is needed to fill the surface! Better lightweight micro than heavy epoxy, that's for sure! 
I filled one side, then gradually filled the flip side with excess micro left over when working on the other bulkheads. I think I half-filled that side with 'spare' micro.

I marked reference lines to ensure I was achieving precisely 45deg with the BID layups. Yeah, I like to be really precise. 
The first side layup worked fine, the largest area of glass I had laid up to this point. The peel-ply had subtle wrinkles in it when I was done, so there is a slight wrinkle texture to the surface, which I sanded down a bit. That's one less gram of weight on the airframe! On that note, when I 36-grit sanded the seatback after hard-shelling, I removed over 6 grams! Yes, I actually measured this!

After the first side was cured, the cut-out templates were used to draw cut-lines. These cuts need to be at 45deg, and I was quite paranoid about cutting 45deg in the correct direction! It worked out fine. The foam crumbles slightly sometimes, but I cut slightly undersize and sanded to perfect shape except for the middle cut-outs which will be finalized after fuselage assembly, per-plans. This was probably excessive, as even the outside edges are likely to be trimmed a bit during the fuselage assembly. 
After making the cut-outs, I used a wire-wheel on the Dremel to remove the corner of foam for the flox corners. This also prepped the glass surface nicely, no foam or micro on the surface to reduce the bond strength to the flox. The plans don't specify how big the flox corner should be, so I used a bit of TLAR engineering here (TLAR = That Looks About Right), and kept them small. Flox is heavy, remember? They are smaller than I have seen on some other builder's websites, though they are not often shown.
I also 'touched up' the micro hardshell before the second side layup. You can see it's lacking a bit in the photo to the left, plus a bit more foam was opened up sanding the corner the glass needs to lay over. A couple of grams saved here?

The second side layup went smoothly, I was careful to try to avoid mini-wrinkles in the peel ply, and they were reduced, if not eliminated.
The flox corners seem quite strong, so it seems my small corners were adequate. Good.
Trimmed and sanded the edges, and that was that. Seatback complete, except the holes for the aileron torque tubes. I read somewhere that it is best to leave these to later.



Chapter 4 Summary
I am waiting to build the permanent firewall until fuselage assembly, as there are some possible changes that may not be apparent until then. I did cut a temporary firewall, but I subsequently changed the planned shape. It should still work fine as it is only temporary for use in assembly operations.

You may also note that I drew centrelines on the parts. This will help with assembly later. Some of these lines bled a bit when hit by epoxy. Oh well, they'll be painted one day and you'll never know.

So until I come back to the permanent firewall, that's it for chapter 4. 
I weighed the pieces and compared to weights reported by nine other builders who are building stock Cozys. Despite my parts being larger, I am only 2% heavier on average for these parts, and for all bulkheads except the seatback, I am lighter than some stock builders! I am very pleased with that. Some of that can be accounted for in the slightly lighter foam core, but there is still more of it than on stock aircraft.

I had the privilege of having David Orr (Beagle, a very experienced canard builder) visit after I finished the bulkheads. He remarked that the parts look very good, and were very precise. Too precise. I need to be less particular and pick up some speed! Thanks, Beagle, I appreciate the feedback!

Below are my part weights. THESE ARE NOT STOCK SIZE so will not be comparable to yours, unless you are using my modified plans, which you aren't.

  • F22 - 985g / 2.17lb
  • F28 - 442g / 0.75lb
  • IP    - 1833g / 4.04lb
  • SB   - 2680g / 5.91lb