Frameless construction, for flat-pack disassembly
Modular construction, for easy assembly or modification
Drill jig design for repeatability
QC (quality checking) during construction
Creating a dovetail template
Forces versus joint strengths
This project creates a 6-sided enclosure providing 8 cubic meters of internal space, and explored:
We needed an outdoor enclosure for 2 very energetic youthful ferrets, as the warm weather was upon us and we could no longer stand the sight of their plaintive faces pressed up against the living room window enviously espying the cats running amok in the vegetable patch. I would obviously love to give them free range, but this world doesn't really care for little lives as it ought and since ferrets can be only semi-domesticated they'd be missing and then dead within a week of free range living for sure. Plus our neighbours have chickens whom I don't think should be introduced to ferrets. So we needed a compound for them.
Ferrets though are both diggers and climbers, so it isn't just a matter of fencing an area of the garden off; we're talking total incarceration. The enclosure required not just walls, but flooring and ceiling too – in essence more a shed type structure than a simple compound (albeit skinned with wire mesh).
I also need to ensure that anything large I construct can be readily dismantled for transport, as life is a little more nomadic than once it was for us... well that's the romantic way of saying we rent our living accommodation these days. This means I wanted to avoid building overly large components. The most rigid (i.e. difficult to transport) component of any shed like construction is the framework, or carcass, on which the walls are hung. So I made a design that didn't need one, from a series of interlocking panels that would support each other.
Here's a view showing how three panels come together to form one corner of the construction, which can be fastened to each other using coach bolts for easy disassembly when required:
But there's a problem with this arrangement; the panels are all of an equal size with a significant material thickness which, as we add more panels leads to gaps:
We could of course make the floor and ceiling panels a little shorter to account for this. That though would make both the construction of the panels and the assembly of the enclosure more complex. If panels are to be different sizes the tooling set-up would need changing (to cut different lengths) which can too easily lead to error. It also means the right panels need to be picked at each step of the assembly, which is bound to cause frustration. Furthermore it would mean the design was less flexible – I couldn't, for example, decide to make a larger compound without a ceiling by using the ceiling panels as wall panels. This latter point is not critical for the ferret enclosure as I'm not going to suddenly decide I don't care if they run away and so deprive them of a ceiling. But I do envision the need to create more shed-like constructions, in all manner of arrangements, and I want a design basis for those rather than having to create a bespoke design every time I came to make a large box-type construction.
All of this means the panels have to be identical, and need to be constructed in a way that allows them to be assembled in any orientation.
The first challenge is to discover how to arrange a set of identical panels so that the gaps are eliminated, or at least reduced to something acceptable. Each panel is 2m by 1m and there are 12 of them to create a 2m cubed enclosure. In the simple arrangement (above) the wall panels all sit around the outside of the floor panel, creating the observed gap. However we can instead place the last wall panel so that it rides atop of the floor panel, as so:
Of course this panel now stands proud of the others at the top of the enclosure, but that is of no matter. The ceiling panel will ride atop of the first three wall panels and will then abut with this fourth, risen, panel.
We have not however completely eliminated the gap. Looking now behind the structure as it currently stands we can see there is one small 'missing' corner:
But you know what? I just couldn't give a damn... I suppose there are applications, or people, where such would be a real annoyance – but actually, I like how this gap reveals the nature of the construction.
With the top fitted to these panels the enclosure assembly is half complete:
The assembly then continues such that opposing panels are always off-set; the one side sitting flush with the floor panel, the opposing side riding atop the floor panel, such that the full assembly looks like this:
Here I have colour coded the panels, those which rise up from the starting position are shown darker. This reflects how I painted the final panels, in order to emphasise the nature of the construction; making it clear that two open (three sided) cubes have been twisted together and interlocked.
Interlocking has been achieved with coach bolts driven through complementary anchor holes pre-drilled into the panels. Because the panels can be used in any orientation (up-right or flat) the anchor holes are always drilled in perpendicular pairs so that they can be anchored to another panel either side to side or at a right-angle. However, panels may or may not be offset with respect to their neighbours; where a panel has been off-set single anchor holes would not align. Therefore, where ever a perpendicular-pair of anchor holes is required a second pair of laterally offset holes are also drilled.
The amount of off-set (if any) is always known as it is simply the thickness of the panels. Laterally off-set pairs of anchor holes are therefore off-set by the thickness of the panels (in this case, 45mm). All of this means that instead of drilling a single hole at an anchor point, each anchor point requires 4 holes to be drilled – which is a bit of an overhead in construction but is, I (clearly) believe worth the trouble in terms of ending up with a highly flexible design basis for future constructions and in simplifying the assembly stage. In this case I had to drill 288 anchor point holes instead of 72. So I bought a cheap drill press stand and set up a jig that would make the job accurate and easily repeatable.
In this design each panel contains a centre strut to add rigidity. These also provide convenient structural support to any added internal features (such as shelving). The structure is quite tall though (2m) and a single additional level at the mid-point seems a little miserly. So I designed a second panel type with twin struts at the one-third and two-third heights of the panel. By mixing these with the original mono-strutted panels I could provide convenient structural support to the interior at three different levels, as in the final design for the enclosure:
The internal fixings are bespoke to the specific requirements of the ferret enclosure and are not part of the flexible modular shed-type construct design pattern; they perhaps do however demonstrate the flexibility of the construct. Note that since some of the panels are off-set to achieve the enclosure, the internal fixings sometimes require spacer pieces if they are to be level, which I don't see as a terrible heartache.
I have also designated one of the panels as the enclosure's doorway, which means it's neighbours will be adorned with additional hardware – latch and hinge brackets – which will need to be bourne in mind on any re-assembly. I could have chosen any of the upright panels as the door, but I elected to select one that rode atop a floor panel, rather than running along the side. This gives a little ground clearance for the door's operation.
8 off twin-strut bearing long-edge pieces
16 off mono-strut bearing long-edge pieces
24 off dovetailed short-end pieces
16 off reinforcing struts
However the design remains strong as the whole construct (excluding the bespoke internal fixings) is made from only 4 different shapes of piece:
It is imperative that the panels exhibit symmetry, so that they can be used in any orientation. The short-end pieces are nominally anchored at the centre. This means that one of the perpendicular-pairs of anchor holes are drilled 22.5mm to the left of centre (477.5mm from the left end) and the second perpendicular-pair are drilled at 22.5mm to the right of centre (477.5mm from the right end). With the drill press jig set-up it's a simple matter of drilling the first perpendicular-pair of anchor holes and then turning the work-piece around to drill the second pair. It can be a pain establishing a jig for a power tool that guarantees such repeatability, and failing to do so is probably the number one cause of shonky results (actually, I may have made that up, I have no reference data on the matter – but you get the point). There are 96 holes in total to drill in to the dovetailed end pieces, all of which can be accomplished with a single jig set-up, with a stop block set 477.5mm from the drilling point. It is therefore well worth the effort to ensure the jig is ultimately stable.
After cutting all the pieces to the right lengths (24 x 2m and 40 x 1m) I laid them out along a straight edge to double check the accuracy of the cuts.
Two of the pieces needed an extra 2mm cutting off as they were slightly over sized. I had (initially) used an Erwin quick-grip to hold the stop-block in place on the mitre saw jig and I remembered that during the cutting I nudged the block and had to re-set it. It must have been then that I cut two pieces slightly too long. When I made my drill jig I fixed the stop-block in place with screws, and I think that's the way to go with these things really.
What's more, the same jig set-up allows the outer sets of anchor holes to be drilled in the long uprights, 477.5mm from either end (another 96 holes). By simply moving the stop block to 522.5mm the inner sets of anchor holes can be drilled into the long uprights. That's a total of 288 holes drilled with just one jig set-up (including one movement of the stop-block).
This repetition of operation is a real strength of any modular design. There are also 88 half-depth lateral cuts required for the lap joints of the panel struts, plus 60 chop cuts to make the pieces the right lengths. These were made using the compound mitre saw in concert with a similar jig arrangement (cutting bed, fence and stop-block). The use of an accurate and stable jig meant that none of these cuts required any markup (except to establish and prove the jig with the first piece). In fact just over half (53%) of the joinery here was made in this manner
I don't have a dado blade (or table saw) for cutting the lap joints, so despite the foregoing I still had to hand-chisel 32 laps in the uprights and hand-cut 32 end-laps on the reinforcing struts. I also had to hand cut and chisel all 48 corner dovetail joints of the panels. That was a total of 80 chiselled joints and 312 hand-made cuts to complete the joinery – optimising the accuracy and repeatability of the power tools made an otherwise daunting project quite manageable.
While joining-up one panel from the pre-cut and pre-drilled lengths I noted that the greatest time spent was in marking-up the dovetails. Since there are 4 dovetails on each of the 12 panels it seemed sensible to spend time simplifying that task. So I made a cardboard hood that would slide onto the end of a piece of the 2”x2” and provide me with an aperture in the shape of the dovetail that I could just draw around.
Template in use
I covered the cardboard hood completely in a plastic (somewhat colourful) parcel tape so that it would weather and keep its shape for the length of service that the project would require.
The cardstock was 1250micron and proved to be durable enough once the plastic tape was added.
Once all of the panels were made I did an initial assembly so I could be sure that it would stand-up (this meant taking it apart later to paint and skin each panel; it took me a few days to gather the resolve to do so once it had been proudly erected and was standing so solid!)
In the assembly I arranged the floor panels so that the panel struts faced 'upwards' – ie. The half-laps cut from the centre of the long edges faced up, like so:
This means that downward forces on the centre strut causes the half-lap joint of the long (2m) edges to close up around the end-laps of the centre strut. If the panel was the other way up such forces would tease the joint apart and could cause the long beams to snap.
Using similar logic, I arranged the roof panels so that the cut-out half-laps also faced upwards. The weight of the roof panels, and anything that I suspend from them (or any cats running across the top), provide a downward force that acts to close up the half-lap joint.
With regards the wall panels, I arranged the cut-out half-laps to face inwards. Because the frames all support each other any bowing that occurs is most likely to be outwards not inwards. If a wall panel does suffer such a force it will again act to close the lap-joint up. (I wish this were true, some of my wall panels happen to have ended up facing the other way, despite the fact I clearly told them which way round to stand...)
The panels are also dovetailed at the corners. Because most of the panels are uprights in the enclosure it seemed to me that the majority of the forces in the final construction would be acting so as to pull the long uprights away from the short edges. Dovetails are strongest (most resistant to forces) where the pins pull against the tails, so I put the tails on the short edges and the pins on the long edges. In a different arrangement of panels it might be better to arrange the dovetail joints differently, but actually it's probably not worth sweating over.
Another thing not to sweat over is the relative arrangement of the floor and ceiling panels. It seemed to make sense to orientate these at a 90 degree offset so that the overall construction was less like two halves bolted together. I'm not sure at all if this will make any real difference but I had a choice to make so I made a choice!
Perhaps more important was deciding how to orientate the floor panels themselves. Our garden has a distinct slope to it, rising at the rear and falling on its approach to the cottage, but relatively flat laterally (left to right). The weakest aspect of the whole design is those long 2m edge pieces with the half-laps cut into them. If they are poorly supported they will surely snap from the weight of a person (although probably not from the weight of the ferrets, just yet...). So I laid the floor panels with the long edges running left to right across the garden, meaning that each long piece was at least on ground level with itself. Obviously I added supports to level the whole construction up, but since there is no hard-standing for the enclosure it seemed best to take the lay of the land into account.
None of these decisions about orientation were made at the drawing board, in fact you may spot contradictions to what I am saying in the 3D models presented earlier. Rather, during assembly I just kept asking myself 'which way round is best for this piece'. It would probably have all been fine if I'd just connected everything up, but it feels better to me to question everything; that way when something eventually goes wrong I should at least have some idea of why.
Plus, you simply can't predefine perfection – leastways not in one lifetime. Certainly not in one project (you know what, there might be another Venn diagram coming up soon – akin to that which I added to my notes on woodworking and blunders). The act of doing is the far greater teacher than the inact of contemplation. Not to denigrate the one or the other. Both matter, seek balance. But how? Do we contemplate and design around that which we know, and then build and struggle against that which we discover we do not know? Not really. Designs and plans are of limited help with regards what we already know. If we knew everything we'd simply crack on without any planning. It's precisely because we don't know things that we create plans and designs. We research, we read, we contemplate and we plan – we plan around the things we have learnt we don't know. It helps. It really helps. But then one way or another the design and the plan lets us down. Things go awry and then we really learn something. Why? Because there's always that slice of stuff we didn't realise we didn't know. Until we came to do it.
Which of course means that some stuff we thought we knew, we actually didn't; but then we secretly suspect that we don't know some of the stuff we do know as well as we ought to know it. And of course we hope that some of the stuff we don't know we don't know won't turn out to be all that important, but then some of it will be... It all seems a little complex which is why people buy other people's plans (or, ahem, make a donation to folk who have good plans...), or else just leap in and hope it will all turnout fine and not shonky. Or expensive.
Sure, feel free to crack on and just build stuff if you're some kind of master craftsman that's been doing this for years – but when there's anything novel at all in a project then a plan or design is essential. Just don't fall into the trap of thinking you can plan your way out of ignorance; lest the planning take a lifetime or more. Accept that there's stuff you don't even know you don't know, and instead of trying to plan for that adopt an approach to the work that deals with it.
Perhaps a picture will help (you're getting one anyway as I find Venn diagrams simply irresistible):
In the example of the ferret enclosure, I knew full well how I was to make the 2m x 1m dovetailed frames so I didn't bother creating a materials cut-list (like I did for The Bookcase), I just jumped right in and chopped up 24 two meter and 40 one meter lengths of 2"x2".
But then, I knew the accuracy of the anchor points would be critical so I paid close attention to those in the 3D design model I created before any drilling activity.
Furthermore, I had no idea what the lay of the land would be before assembling the enclosure as I needed to see the thing to determine the best placement for it... so I didn't plan or design for that, but I did carefully question all of my decisions about how I would lay the panels down and join them up.
Planning is good (I should say that twice really, or rather you should read that again) – but it is never perfect; it can't be. So taking time to think about what does and doesn't need to be planned for, AND what cannot be planned for, is the best starting point for any project.
Here's more views of The Construct and the ferrets once finished.
Now it took me 5 days to do all the painting so although I've probably already made this article too long I am going to include a picture of the painting process:
But turning attention now to the inside of The Construct, you can see that the entry vestibule provides oversight of all that lays within:
The Construct is a cubic Platonic solid and as such has six faces. In order to orientate this discussion a regular and shared nomenclature is necessary.
Door-side (the face on which the door is sited)
Ceiling (the uppermost, skyward facing, face)
Floor (the bottommost, least gravity defying, face)
Each face shall be designated thus:
At the right-on-entry uppermost back side you will see the upper mezzanine, or l'observatoire bleu (The Blue Observatory, or The Blue Room) from which the grand vista of the North Yorkshire Moors National Park can be drunk in; or if you're a ferret, where you can scrabble in the sand pit til your heart is content. The sheer height of The Blue Observatory allows sand to be scattered far and wide by the serious minded ferret.
Broad-beam white declivity leading to:
The first floor Grand Ballroom stretching the full length of the accomodations from the door side all the way back to the back-side, sporting a sprung parquetted dance floor hailing straight out of the roaring 20's and providing the narrow-beam white declivity down through the gymnasium and on to the groundfloor.
As stated, central to the groundfloor is the large structural 'hyper-cube' climbing frame in the gymnasium section. Left-on-entry back side provides toiletry facilities, since any discerning ferret will choose this particular, hardest of all corners to clean, corner to deficate; and indeed they do. Left-on-entry doorside finds the indoor pool (not shown) for cooling offf after an energetic tumble down the Hyper-Cube centre piece. Right-on-entry back side finds the blue declivity, leading up to the lower mezzanine.
On the lower mezzanine you will find the master bedroom, with its crennelated entrance. The crennelations speak to the architectural mores of the time, more than to the disintergration of the reclaimed wood used to build the sheltered accomodations.
The Construct is finished throughout in a fine galvanised steel mesh (specified, procurred and installed by Kristiin and her hammer-tacker, from Cockahoop Containment Solutions Inc.).
The Construct was first commissioned by and for Daniel James Hunt and Daisy Phillippa-Jane Morrison, who remain in residence at this time.
You should also see here a little video made on moving in day (from Cockahoop Video Productions).