Gorgon lights by Ian Devenney

A few months ago, I was in a restaurant in the annex that has some of the nicer lighting fixtures made with standard copper plumbing fittings. There seem to be a lot of these floating around, but they tend to be inelegant, messy (not in a good way), and, on the whole, boring. I thought perhaps I would design my own version of this type of chandelier.

I liked the copper, but I wanted organically curving lines with smooth profiles, instead of the square and lumpy fittings from the hardware store. I set to work bending some copper refrigeration pipe and worked out a system to bend nice curves over a plywood form, machined with a cove in the edge.

Then I began designing a custom-spun socket cover that would work with a 2" keyless candelabra base.

I drew the cross sections that I knew I needed, and then lofted them in Rhino. Here is the result.

The canopy is an assembly of two nesting plates, also spun copper, that was CNC water-jet cut to admit the copper refrigeration tube.

Luckily, 7/16 brass tube fits nicely inside the 1/2" copper refrigeration pipe, and I machined some custom fittings on a metal lathe to admit the 1/8 IPS nipples commonly used in lighting applications.

The whole assembly is then soldered together. The first iterations of this design used plumbing solder, but it shows as a faint line at the join between the tube and the socket cover. Sadly, low-temp solder cannot be plated to hide the joint. Subsequent iterations use a much higher temperature silver braze, which can be copper electro-plated almost entirely hide the joint. There is still a crisp line at the joint, but no change in color.

The electro-plating is a fantastic process that I hope to go into in more detail later. But essentially I bought a pre-made plating solution, dipped the end of each "arm" into the solution and ran a current through the solution.

After primary assembly and plating, the parts are brushed and then sprayed with a durable metal lacquer to prevent oxidization of the copper. These lights are also available unfinished.

Then the lamp is assembled with the canopy and the sockets and wire are installed.

The last step is the certification by the ESA. This is a poorly understood process for a lot of new designers, and it is hard to find solid information. I would suggest trying to get in touch with your local inspector to discuss the requirements before he/she makes the trip at $190/h.

Lights must have proper labeling and grounding. In this case it looks like this:


Gryphon Wingback by Ian Devenney

What if Tom Dixon's wingback bore the child of BDDW's wingback? And what if that child turned out to be suave, audacious, and a little bit kooky?

I think that child would look vaguely like this.

See if you don't agree:

Of course it isn't upholstered yet... but you get the idea. A little brighter in temperament than daddy, and more graceful. More ostentatious than mommy, but still a little old-fashioned.

Recommended upholstery: Herringbone Stripe by Paul Smith.

Decomposition #2 by Ian Devenney

I've had some good reviews and interest in the first Decomposition table, and I've been playing with some ideas for the next iteration of the design. I think it is now time to get moving.
Firstly, I wanted to better capture the sensation of movement and falling of the blocks, so in this version I will be attaching some of the blocks to appear that they are falling by making cut at compound angles and joining them to the main structure with a Festool domino machine and some creative clamping.


Second, I wanted to make the table a bit wider. Looks like I'm going with something in the neighborhood of 34-36" (up from the 19" width of the first table). This allows me to bring the cascade up to the level of the main table-top on both sides, to avoid seeing the flat line of cascading blocks from the side. The first table never felt balanced when viewing directly from the front, either.

Third, the new base will be made from two plates of UV bonded glass, instead of steel. This will enhance the illusion of a floating table top.

The first step was to mock-up the shape using pink foam and spray adhesive. Pictured above. Next, I will measure all the "steps" down the cascade at each row and draw them in Rhino 3D so that I can tweak the dimensions and output my instructions for assembly.

I've also pulled out some material (quite a bit in fact) - about 65
board-feet of 2" x 2" x 16" walnut, or 145 sticks (see below). The first thing I've done is to paint one end of each stick, so that I can keep track of their orientation while I machine them. In the previous table, I kept them in the same orientation, which led to some minor magnification of error due to the sticks being slightly out of square. The issue was rectified in subsequent machining tasks, but it made things a bit more tricky. Here, I will rotate every other stick during the initial glue-up to cancel-out any kind of error.

Check back for updates on the progress of this new design!

Hawthorne patio table in Process by Ian Devenney

A little while ago I was walking with my mother at her cottage property outside of Kingston - a lovely spot.


Here's a shot of the pond, and another of my mother.

We were clearing some trails and I found a sapling with some really interesting texture. I have been interested in working with aluminum castings for some time now and it seemed like this sapling would be a perfect shape for some table legs. When I showed it to arborist Joseph Gillingham, he identified it at black hawthorne.

The casting process used for making the legs is aluminum sand-casting. In this process, sand with added binders is packed around the form or "plug" of the object one wants to create in metal up to a middle "parting line". Then, a release agent is put down and the other half of the plug is also packed in sand. Once the binders harden, the two sections are separated and the plug removed. When the two parts are re-assembled there is a hollow cavity into which the molten metal is poured.

In this case, I wanted an integral mounting plate as part of the casting, so I simply added a plate of 1/2" MDF to the bottom of the sapling, and filled in a hollow with bondo, an auto-body filler. The natural hollow could not be cast in sand, because it lay on the natural parting line of the plug.
The completed plug looks like this, with the finished casting beside it:

When the casting comes from the foundry (I use Skara foundry in Mississauga), it is still fairly rough, and the parting line can be seen in a thin line of protruding metal.
Finishing the castings is a labour-intensive job that involves removing the parting line and smoothing out the mounting plate with an angle grinder and a flap-disk.

Next, the casting is sand-blasted with 80 grit sand in a blasting cabinet attached to the air compressor.

Then, it is sanded by hand to polish the high spots, while still leaving the dull texture of the low spots.

In order to prevent corrosion over time, the legs were anodized clear. Aluminum doesn't really rust, but it does oxidize to a rough, dull grey colour. I wanted to keep the brightness of the polished areas and keep it easy to clean, so the anodizing step is crucial.

Unfortunately, in this case, the anodize tends to even out the tones of the leg, leaving it much more uniform than before. In order to bring out the texture of the wood grain and bark, I rubbed the legs down with a black India ink wash. It took a little bit of trial and error to get the mixture of ink to water right, but I think the overall effect is successful.

For the first iteration of the Hawthorne Series, I wanted to make a table with a stone top.

Granite and marble are probably the most commonly thought-of options, but they both have a hard, cold feeling to them. The aluminum legs are organic in shape, but the castings give the design enough "coolness" already. I knew I wanted a mostly black stone with grey accents to mirror the colour of the legs. Soapstone was the natural solution.

Many people think of soapstone as a grey, soft material used for carving. It is true that soapstone is softer than the other architectural stones, because of its high talc content but there are harder varieties in use for kitchen counter-tops etc. The upside of all this is that it has a luxuriously soft hand, and doesn't feel as cold to the touch as other stones do. And when oiled, it has a wonderful deep black colour. Unlike granite, soapstone is naturally non-porous, so it will never stain and sealing it is not necessary. If it does scratch, light sanding with mineral oil is all that is necessary. Ultimately, it will develop a patina, so it would not be the best choice for those who demand a pristine and flawless surface. Personally, though, I prefer the kind of elegance that patinated surfaces have. Wear and tear reads as an intentional aspect to the design, as opposed to "damage".

For a top to go with the Hawthorne legs, it seemed a perfect fit. The organic quality of the legs would do well with the elegance of worn-in soapstone.

So back to the constru
ction details, the stone is 1.25" thick, but still requires continuous support to make sure it won't crack. For the support, I used 1/4" thick stainless steel plate that was CNC plasma cut to reduce weight, to give a handle when moving the table, and to give an extra visual feature if one were to look underneath the table.

Future iterations of the Hawthorne table will feature cast-concrete tops, though I am happy to produce the design with any other kind of natural stone, too.

Decomposition Table: production by Ian Devenney

Here is the order of operations I used:

1) Joint stock square (in this case, 2" square, 16" long sticks of walnut)
2) Rip to final thickness (1.5") on the table saw in one dimension only
3) Glue panels (3 of them) from the stock
4) Plane panel to thickness across the grain (I know you're not really supposed to to this)
5) Rip panel into five 2.5" wide strips, and rotate to reveal end-grain blocks
6) Layout end-grain strips in book-matched pattern
7) Trim to length according to location of the cascade
8) Glue strips in a staggered formation into the final table-top
9) Flatten bottom of table with router sled


10) Sand bottom of table to 220-grit
11) Map out length of blocks needed for each line in the cascade formation (ensuring 2.5" of glue surface area between each block).
12) Cut the blocks slightly over-size (20-30 thousandths of an inch over)
13) Hand sand the individual blocks on the front and back faces to 220 grit.
14) Dry-fit blocks together, and apply paste wax on the inside corners (to enable removal of glue squeeze-out)
15) Glue entire row of cascade together

16) Hand sand whole row to the proper thickness (also flattens the surface)
17) Dry-fit whole row onto the table top and wax surfaces of adjoining inside corners.
18) Install row, position with clamps
19) Repeat steps 12-18 for each of the rows in the table.
20) Have coffee
21) carefully scrape glue and wax residue from inside corners
22) Flatten top of table with router sled

23) sand table to 320 grit
24) Apply Danish oil finish
25) find more glue residue revealed by the finish.
26) scrape it off, re-apply finish
27) Measure exact height of cascade, order CNC cut steel plate to match.
28) Mortise underside of table with router jig and bushings for mounting the legs
29) Mortise underside of table for steel-tube stiffeners

decomp mortise.JPG

30) Beg your friend Marina Guglielmi to weld the plates together

decomp leg weld.JPG

31) Beg your friend Aaron Gough to help you cut slotted holes in the tube on his milling machine.
32) Mark out mounting holes for the legs and tubes
33) Install insert nuts for 1/4-20 stainless machine screws.
34) Get busy with another project for several weeks, only to find that the whole thing has shrunk and the steel doesn't fit anymore.
35) Repeat steps 28 & 29.
36) Install UHMWPE strips to bottom edge of steel plate with pop-rivets (this is really several steps in one...)

37) Apply final finish, clean, photograph!
38) Yay, you're done!

Which is all to say that it took a little while.

Decomposition Table: first steps by Ian Devenney

The decomposition coffee table came out of my desire to create a complex and labour-intensive design as a showpiece, rather than a design that could be easily put into larger scale production.
I'm not sure where the idea came from entirely, but I say it is based on the idea of a digital image "decomposing", or losing resolution, to reveal its component pixels. I was looking for the juxtaposition of electronic and organic motifs in a single object and the sensation of movement captured in a solid material.
The first step after coming up with a basic concept was to make a half-scale model with some blocks of poplar:

I liked the general concept, but it wasn't quite resolved so I went on to make two more models out of blue foam at full-scale. Most people who saw those models were either confused as to what I was trying to do, or skeptical that it could be made well. I won't show them here because they were basically a mess of spray adhesive and chemically-melted foam.

The foam was a good starting place for my first digital drawings of the table. At this point, my intention was to make a full-sized dining table with a separate base, so the cascade only went part-way to the floor.

I drew the tabletop in Rhino-3D by creating each block individually and dragging it into position:

I sent this image to my friend, architect Scott Barker, and he suggested that the cascade seemed too jumbled, and could benefit from being smoothed out. I agreed, and modified the drawing:

As you can see, I also started working on the base design. I felt that it needed to be somehow connected to the table, as opposed to just something slapped on top (or bottom, rather). Here was one of my ideas that would use CNC cut steel plate in a profile that mirrored the profile of the cascade when viewed from the end. It would also allow economical use of the material, because both sides of the cut-out were used on opposite ends of the table.
Ultimately, this version seemed too busy and it has been relegated to the cognitive recycling bin.

I did like sections of the cascade. And in my musings over the base design, I really wanted the wood to reach the ground as one of the legs. But it didn't seem realistic to build it as a single piece for a full-height table due to the weight. I thought about numerous options for having a knock-down system for the wooden cascade so the table could be moved in multiple parts but I wasn't sure how to make those connections seamless.

I'm sure there is a way, but the next step was to make a smaller version as practice. No doubt there would be unanticipated challenges in production that would need to get worked out before moving to a full-size table.

This is the drawing that guided my practice: