Wednesday, May 25, 2016

Slip Casting Update

This post is an update to my earlier post on slip casting 3D printed forms. The 3D print is on the left. The first glazed/fired sample is on the right. They shrink (about 11-12%)!


I've had a few of the pieces fired. The first step after firing is to rinse them off. The first firing leaves some residue and you want a clean surface for the glaze to adhere well.

The next stage is to put some wax on the bottom of the pieces so they don't get any glaze there which would stick them to the kiln shelf:

Next it's time to experiment with the glazing. First, this is how the glazes are stored - 5 gallon buckets with a toilet brush to stir them up. The solids settle out and need to be mixed back in - it takes a few minutes for some that haven't been used in a while:

The vases get dipped using tongs. The vase in the rear has two dips which overlap. So there will be three different colors. The vase in the front is a solid color from a single dip. Once the glaze has dried it's easy to smooth out the bumps and drips you see. You can also fill in the areas where the tongs hit by just rubbing adjacent glaze into the holes.

Waxing, dipping, brushing, drizzling...

You'd never guess what colors they'll be based on the picture below:

Here's the first one done:

More to come as these get fired...

New Cottle Boards

I made some more cottle boards for mold making. These are larger and have an improved design. No more screw holes. Instead I rabbeted the edge to accommodate the clamp block. In this way the connection is Poplar to particle board rather then Poplar to Melamine. Thus the glue adheres just fine and there's no need for screws. Here are the boards prior to glue up showing the rabbets:

Clamped up:

Two sets are 12" high and one is 16" high. 12" wide each. Great for taller pieces. In addition I made two more at 8" high.

Thursday, May 12, 2016

Slip Casting 3D Printed Forms

I've been taking a course in slip casting at the Ann Arbor Art Center. My goal with the class was to combine this traditional method of making ceramic forms with newer digital techniques of production. This blog post documents what I've learned so far and some of the tools and techniques used. 

The basic process involves pouring a plaster mold of a form, then pouring in liquid clay known as slip to create a ceramic version of the objects. The ceramic molded part is fired to create a durable, usable object.

Making Cottle Boards

This method uses something called cottle boards to make the plaster mold. I wanted to make the boards from Melamine surfaced particle board. The Melamine surface doesn't stick to the plaster so the forms release from the mold very easily.

I bought a 3/4" 4'x8' sheet for $38. I cut it up on the table saw. I milled a 6' poplar 2x6 into 1.5"x1.5" strips which got crosscut to the width of the boards to make clamping blocks.

I drilled holes to attach the clamp block to the Melamine and screwed them on. The drill holes, which will be exposed to the wet plaster, were sealed with an adhesive sealant and were covered in plastic caps. 

The edges of the boards were sealed with a stain resistant shellac.

I made quite a few of these because I wanted to donate some to the Art Center. I kept a few sets for myself and my studio - West Huron Sculptors.

One Part Mold

As a very simple first test I made a one part mold. I used a basic plastic bathroom cup as the form. Here are the cottle boards clamped together. You can see how the configuration lets you vary the size of the mold. The boards sit on another piece of Melamine.

I wanted about 1" around the entire part. The edges need to first be sealed with clay. In this case I used water based clay. I've learned that using oil based clay is a much cleaner method as it is easily removed from the Melamine and plaster mold.

Next the object to be cast needs to be placed in the form. It is secured to the bottom with a layer of clay. This keeps the objects from floating up as the plaster is poured in.

We used this plaster which was purchased from Rovin Ceramics in Ann Arbor, MI. It gets mixed in a ratio of about 5000 grams of plaster to one gallon of water. Another way to compute the amounts is 3 parts plaster to 2 parts water, by volume.

The resulting mold, with the prototype still in place:

The mold has been cleaned up a bit with a rasp. The first pour of slip is in place:

After 20 minutes or so I poured out the excess slip:

After the slip set up a bit more I trimmed the top edge with a plastic scraper:

Here compressed air is used to create a separation of the cast from the mold. Once the edges are free it's easy to turn the mold over and bang it onto a table to cause the cast to drop out.

Gee the process works. Craftsmanship score: C-. Inspiring form score: F.

Two Part Mold

More complex forms can be made by using a mold which comes apart to release the object. For the two part mold I decided to 3D print an object. The key to this is that the form must have no undercuts and should fall away from the edges of the form. However, I did not want a simple form - I wanted something with some geometric complexity. This is what I came up with:

The idea was a twisting, tapered form designed in a way so there would be no undercuts after the parting line of the mold was introduced. Here are the cross section curves - you can see there are no undercuts relative a parting line which could be made. Actually a straight line doesn't work - but the parting "line" is actually a surface and can move subtly to prevent the undercuts. This was modeled with only two curves, then twisted 45 degrees. More cross sections were automatically generated to improve the fidelity of the mesh used for printing.

I had it printed at Shapeways in their cheapest plastic. It is 7" high and 4.3" square at the top. Here's the printed result which took one week to get delivered to me:

The surface has tiny horizontal ridges from the layering done in the printing process. These translate to the mold and appear on the piece. By the time the glaze is on you don't see it at all however.

Time to make the molds. It starts with laying a clay bed down to hide the bottom half of the prototype. In this way only the top is exposed to the first mold.

Here's the prototype embedded to the parting line. Clay seals all the seams.

After the plaster has been poured and things have cooled the clay can be stripped off. That completes one half of the mold:

Now to make the second mold. Here the cottle boards surround the first section of the mold. The second portion is poured over it. A coating of mold release (Murphy's Oil Soap solution works, so does a light coat of WD-40) is essential to keep the two halves from bonding. Essential. 

Yup, I forgot the mold release. Fuckity-fuck-me! The two sides were completely stuck together. I had to axe/chisel the new mold off. Surprisingly, and happily, I was able to at least save the first mold:

Okay, back on track. The plaster is poured AGAIN and allowed to set. After the heat dissipates the forms can be removed.

Here's the 3D printed prototype being removed from the molds. Plastic doesn't stick to plaster so that was easy to remove.

The Pour

Next step is to pour in the slip. The forms are held together with a tightly stretched inner tube. Clay plugs any areas which would leak - in this case it's the upper area where the fit is chipped away:

The slip box and bucket mixing method:

Here the mold is filled - the slip I used was purchased from Rovin Ceramics. It is the Laguna Clay Company NS4 Dover Liquid Casting Slip. After pouring in the slip... you wait - in this case about 45 minutes because the plaster was still fairly wet which slows the water absorption. When the mold is drier it's only about 20 minutes.

It's important to drag the slip beyond the edge of the mold opening; this holds it against the mold wall as it solidifies. It's also important to keep the mold full. I usually top of the mold 3 or 4 more times as it dries.

After the wall thickness builds to the desired level the excess slip can be poured out for reuse. Below the molds have been separated.

The form needs to be carefully pulled from the other mold half. Again, using compressed air is extremely helpful. You can see the seam which is cleaned up with the clay tools and a brush:

The top of the form requires more trimming/shaping to clean it up. I usually just drag the piece, once leather hard, over some flat 80 grit sandpaper to smooth it out. Then scrape on a subtle chamfer along the inside and outside edges.

Any ridges from where the molds meet can be scraped out:

Things are smoothed over with 80 grit sandpaper. That's generally very aggressive paper but it works well because using horizontal strokes matches the layering of the 3D print. So the seam cleanup seems cohesive.

See this post for some results: Slip Casting Update.

Interlock Joint Table

I decided to make a table using an interlocking joint I came across. The joint is designed in such a way that once the parts are assembled it appears they penetrate one another without a visible seam.

This is not a new joinery idea, nor a new table design, but it was fun coming to understand the joint and making it on the router.

I actually made the table twice. Between the first run and the second I changed the overall height, the proportions, and the method of fixturing the stock to the router table. I actually really didn't like the first one. It was too small and fragile looking. It was also an overly complex way of holding the stock and required a fair amount of hand work to clean it up. The second version was wider, taller, and much easier to secure to the table. Although that too had its disadvantages.

Table Version 1

The table has three wood members. I wanted each to read as separate so I used three different wood species. The woods I chose were Walnut, Ash, and Mahogany. Walnut because it is darker than the others and the grain is evident but not overpowering. Ash because it has a very prevalent grain pattern, and Mahogany because the color is different and the grain is very subtle.

I found the wood in the off cuts section of my lumber dealer here in Ann Arbor (B&B Heartwoods). They were cheap, but not exactly sized correctly. I planed them down to be a uniform thickness, which I wanted to maximize. This resulted in stock 1.16" thick - oooookay so be it!

Parts and Sequence of Assembly

Here are the parts - geometrically identical except for the mortise in the middle:

Here are some images of the sequence of assembly. The mortise of the red accommodates the section of the green:

This is a key image. You can see how the blue member has a gap in the front while the green member is positioned with a gap for the back.

The view is reversed below. You can see how the blue member can slide past the green one. The gap on the other side of the blue member slips over the green one.

The green is slid back, closing over the blue, and hiding the mortise in the red.

For those to fit invisibly (or nearly so, it's definitely not totally invisible) the stock needs to be planed uniformly and measured accurately for the 3D modeling and toolpath programming. The parts obviously need to be cut accurately. I aimed for a gap of no larger than 0.01" (0.25mm).

I cut them on my 3-axis CNC router. Because the parts need to be milled from both sides I had the flip the part half way through cutting. If set up correctly this is absolutely no problem. I used two dowel holes, centered on the part to align the part again after the flip. I used a 1/4" upshear endmill to cut them into the spoilboard and through the workpiece. These were two separate toolpaths using the same geometry - otherwise the depth of cut was too great. The tool cut length is only 1.25" but in this case the stock was 1.16" thick and I need to go at least 0.5" into the spoilboard.

Here you can see the 1/4" x 1-1/4" dowel pins and corresponding holes in the stock:


The three tools I used for routing were:
  • 1/4" Upshear Endmill
  • 1/2" Upshear Endmill
  • 1/2" Upshear Ball Endmill

Part Milling

Here's the sequence of cutting:
  1. Bore into spoilboard for the metal alignment pins 
  2. Secure the workpiece to spoilboard with two screws
  3. Bore through the workpiece for the alignment pins
  4. Mill the mortise in the center of the part
  5. Contour the outline using thick tabs to secure the part into the stock
  6. Rough out the chamfered edge (include tabs)
  7. Finish the chamfered edge (include tabs)
  8. Insert dowel pins into the spoilboard
  9. Flip the work
  10. Secure the workpiece again with screws
  11. Contour the outline with tabs (same toolpath as above)
  12. Rough the chamfer (same toolpath with no tabs needed on this side)
  13. Finish the chamfer (same toolpath above, again no tabs)
Here's the first part I cut, mid way through the top side operations. The mortise has been cut, the outer contour has been cut - with tabs to hold the part in place in the stock. The roughing of the chamfer is done as well. Also note the holes for the dowel pins along the centerline of the part. The holes are milled but they dowels aren't in place yet. You can see a dowel pin near the left bottom of the image. That one is used to line of the stock straight on the router bed. Additionally note the screw holes (upper left, lower right) which secure the stock to the spoilboard. Dowel pins alone aren't gong to work well enough when using an upshear bit which tends to lift the workpiece from the table.

Next up is the ball end milling to smooth the chamfered surface. You can see the tabs in this image. They hold the workpiece in the blank during the cutting. They only hit the chamfer from this side. The other side is cleanly milled. After the part is fully cut those will be smooth out with a sharp chisel and scraper.

Here's the roughing pass after the flip - the centered dowel pin is visible. These provide a very accurate way to realign the part after the flip.

Here are all three parts after finished routing:

The parts are removed from the stock block using the bandsaw and then a chisel is used to clean off the tab material from the bevel.

Next a chisel is used to square up the corners where the cylindrical router bit could not reach. This allowed me to keep the crisp edges I wanted to the chamfer. It would have been possible to round the edges of the chamfers to match the mortise but that's not the look I wanted. If I were making these tables for production I would certainly do that. Or even make the mortises completely round at the ends and the chamfers become a round over.

This version was a lot of toolpaths and a lot of manual cleanup.

Table Version 2

So in this next version I tried to simplify everything. I wanted to use smaller sized stock without so much waste, and no tabs which had to be manually cleaned up. In terms of the fixturing I've tried lots of things in the past, usually using a vacuum table. This time I tried screwing right through the part and making the hole part of the design.

Here's the 3D model with the chamfered through holes:

3/8" Bolts. A hole is bored into the table, narrower than the hole in the stock, so the threads bite into the spoilboard.

Here's a part, bolted down, after milling one side:

Here's the part flipped, after completing the roughing pass. A ball end mill chamfers the surface. A final contour pass using the side of the ball end tool finished the milling.

Milled parts, after chiseling out the corners:

The hole gets a chamfer by hand:

The finished table:

After having seen the built project my conclusion is - the holes are fantastically lame.  :) I think I'm going to have to do some more work on this to make them appear more integral to the piece.

Because the original holes were centered on the end arcs, to cover the hole location and be centered now, the new holes have to be quite large as seen on the green piece. Doesn't look good. Same for centering them on the hold holes (red). So I think the best (least worst) bet are 1" holes over the old location (blue).

Slots better echo the overall form - those below are 2" long and 3.75" long (which matches the width):