A Twisted Form

[Mark J]

Well in the vane of "Make no little plans" I’ve started another project and I thought I would try a journal.   I’m not promising this project is even going to work; it’s definitely new ground for me.  I have gotten successfully through several steps, but there are a lot of rivers yet to cross and no guarantees of getting to the other side.  

Here is the destination.  I don't have a title for the piece, yet, that will come after (if) it is in hand.  Please ignore the surface lines,  that’s just an artifact of SketchUP

I am basically using the three sided bowl technique that I have worked out to turn the bowl, although there will be some evolution.  But before I can begin to turn I have to make up the blank and that is the process I am engaged in now.

I have actually been working on this project since last fall.  A lot of thinking.  At the outset I envisioned using a twisted block of wood, but how to make it?  My first thought was a large solid block cut down to the twisted form  perhaps using CNC.  But I couldn’t immediately see how to even draw what I was thinking about so I could evaluate dimensions and aesthetic. 

The block itself is not easy to draw in SketchUp, but a lot of computer time and some tips from RonnW later I figured out how to approximate a cube with twisting sides by stacking up thin rectangles and rotating each successive layer by a small degree:

Then it occurred to me that I could make the block up using sheets of veneer.  And these  did not have to be square, I could cut them down to reflect the target shape and thus reduce the amount of wood that would need to be turned away and control the width of the pillars.

With a drawing of the blank done I then had to figure out how to model the bowl itself.  That process is best explained by the picture.

After stacking and rotating the veneer stars in SketchUp, I draw a cross-section of the bowl.  This is wrapped into 3D with the ‘follow me’ tool and then superimposed on the turning blank.  Using ‘intersect faces’ the two models are joined.  These are complex drawings so the intersect faces operation can take several minutes to hours.  Now all that remains is to delete each line segment that is not a member of both the original models.  It is a laborious process.  I’m getting faster, but if I’m making up a blank and a bowl design, the modeling has been taking a good two days. 

So with a satisfactory plan it remains to figure a way to actually stack and glue the veneer stars with one degree of rotation between sheets.  I spent a little time on this one, but it occurred to me that if the points of the stars were perfect (or nearly so) then each point would be separated from the adjacent point by a fixed distance and that this space could be filled by a pin of specific diameter just tall enough to separate two sheets. I did some more modeling in SketchUp using just one set of star points.

How big a pin?  Well I knew there was a geometric solution, but honestly at this point in my life what I remembered about trigonometry was that it exists.  But after a little remedial study I came up with:

So I put it in a spread sheet and iterated to a solution that yielded a “purchasable” pin diameter of 5/64” and square size of 9.161”.   Two sets of three pins positioned diagonally 12.955" apart would in theory position two veneer stars with 1 degree of rotation.   I also figured for smaller and larger diameters.

This assumes that the jig was constructed with a high degree of precision, i.e. not in my shop, this would need CNC.  Further more cutting out the veneer stars would also require precision and also would be better done on a CNC router.   

I was asking around at my woodworking club if anyone had a CNC machine and heard about a makers space in a nearby suburb, “Make It Here” owned by Mike Haines.  Some investigation and I found that for about $100 I could get trained on the Shop Bot CNC router and have ad lib use for one month.  So I signed up. 

Mike turned out to be worth the price of admission himself.  He understood my project and gave me good advice on how to proceed at multiple junctures.  He was also there to help with the Shop Bot, and kept me from making a lot of mistakes (I still managed to squeeze in some).   Still, I was surprised at how easy it was to learn to use Vcarve and the Shop Bot, which is in part due to Mike’s well written instructions and his real time support.  Make It Here also has everything else you would want in a complete wood shop as well as a complete metal shop.  It’s worth looking around your own area to see if something like this exists. 

Since the positioning jig is plastic I need clamping cauls.  I decided to use ¾” plywood and go overkill with two on each side because the quality of the glue joint between each veneer star is critical.  While I could have cut the plywood on the bandsaw, it was a good opportunity to try out Vcarve and the Shop Bot on a simple task. 

Success, 14” circles.  I’m a little farther along than this, but I’ll leave it here for the moment.  I’ll post some more saga shortly. 

[pkinneb]

[Chestnut]

[Byrdie]

This is going to be interesting to watch.

Just wondering out loud ... since each of the 'veneer layers" of the "star" are the same shape, could this have been cut or routed into a single, long block of wood and then the layers sliced from that on a band saw?

[Mark J]

I supose you could.  It would be like making your own veneer, but from a block with a cross sectional shape other than a rectangle.

[wtnhighlander]

This is 3D printing.

The hard way.

I can't wait to see the result!

[Coop]

This is going to be one helluva deal! Good luck bud! 

[Mark J]

Now for the positioning jig.  I had acquired 2’ x 2’ PVC ceiling tiles to cut out the jig, but this material proved to be too soft.  Fortunately I was able to buy some acrylic sheet from Mike.  Then he offered to cut my jig out on the Epilog laser.  This was really generous of him as I had not registered for or taken training on this machine, so Mike cut out the parts for me.  They came out perfect!  Perfect as in it made the CNC router look crude by comparison.  I gotta get trained on that laser.

The jig has a bottom plate with the pin sets for medium and small stars as well as a top plate with no pins.  We also cut out templates for the large and small stars.

The jig needs 5/64” pins (exactly).  Searching the net I was able to find rod in various metals and diameters, but would have had to buy feet of it and cut off tiny bits.  Discussing this with Mike he suggested I look for steel dowel pins.  I was able to source these from McMaster Carr in the exact diameter and lengths I needed.  Pins in hand, and what do you know, they fit perfectly in the laser cut jig.  And even better the templates fit perfectly between the pins.  

So now the penultimate step in the setup for this project before I even get to see if this blank will turn, it’s time to cut out the veneer stars.  This has proved to be a troublesome step.  First my veneer flitch is about 9 feet long while the Shop Bot table is 8 feet, so this has to be done in two sections.  With the Shop Bot, the material is affixed to the spoil board using a pin nailer shooting tiny plastic nails into the waste areas.  There are 11 sheets of 1/16” veneer and while the total thickness is within the capabilities of the pin nailer, Mike thought it would be prudent to divide the flitch into 5 and 6 sheet batches.  

This proved to be good advice for another reason.  Maybe it's because I bought the veneer 6 weeks ago, but it won’t lay flat, it’s waffled.  Not much, but enough and it is surprising how much resistance to flattening a 1/16” sheet of wood can muster, much less 5 or 6 sheets working together.  

Typically one would program the nail locations into Vcarve, and then have Shop Bot mark the spots on the material.  After completing your cuts Shop Bot can then be sent back to the nail locations and drill them out.  Unfortunately this does not work if you material won’t lay flat.  I was able to use the templates to find the safe areas for the pins.  I'll have to slip a pull saw under the veneer and saw through the pins when I'm done cutting.

Unfortunately the waffle woes don’t stop there.  The material needs to be flat to be cut out accurately, and precision in this project counts.  I worked this problem for a couple of hours trying various ideas.  The best I could come up with was to place two small sections of ¾” plywood on top of the veneer with a 2 inch space in between.  Then kneeling on one piece of ply and pressing down on the other with my left arm I shot the pins (and then some more pins) into the veneer and spoil board in between the two plywood squares.  And still there is a just perceptible waffle.  It’s the natural elasticity of plastic nails and wood, and well the ornery nature of the Wood Spirits.  

The good news is the sample stars I tried fit between the pins just fine even when I press down.   So I’m counting that a success.  Now we have to see how the glue up goes.

[Ronn W]

I had no idea that I was getting you in to something like this.  WOW!  Good porgress so far.  Good luck.

[Mark J]

So here are the stars of our show, in order of appearance.

I am maintaining the flitch order, so it's 11 stars from position one, then 11 from position two, etc.  I don't think this will matter very much if at all, but it won't hurt.  This is quarter sawn Khaya (African mahogany) which I choose for its uniform grain pattern.  I had some losses while cutting out and chopping out the stars, but not too much.  More with the small stars, which I think reflects the longer arms.  I made 121 and I think I will use 90 - 100, so I have some wiggle room.  By the way I purchased the veneer from Certainly Wood. They had a good selection of thicker veneers and great customer service particularly for my small order, so a free product placement for their company.  By the way the stars get sandwiched between some pieces of plywood while waiting their turn on stage so as to keep them flat.  

So here is my version of a book press.  

I'm gluing up the first two stars here.  The spring clamps are to hold up the the parallel clamp heads out of the way when setting up.  There are two pieces of 3/4 ply on each side as cauls, and the acrylic jig plates and stars are in between.  It's all working well, ...but it's a little rascally.  

So first it's a time sensitive process hence the wrist watch.  I am using Titebond Quick and Thick.  The other choice I considered was Titebond II.  Both have relatively short in clamp times which allows me to get the next star on to the stack more quickly, but they also have short open times, so you need to get things in clamp quickly.  No time for chit chat or taking pictures.  The chief difference between the two glues is that Quick & Thick is (wait for it) much thicker.  It also dries clear as opposed to orange which is why I am using it here.  Incidentally I should mention that Q&T is significantly thicker.  It's spreadable, but it is noticeably more difficult and I do want to get complete coverage of the star.  On the plus side you get less squeeze out.  

So that leads to the next problem.  What to do with the glue spreader (roller) in between stars.  I don't' want to do a full water clean up in between stars, yet it cures just as fast on the roller as anywhere else.  I spoke to the Titebond folks about this and they suggested swishing it in a cup of water, then wiping it off.  That's my current plan; seems to be working.   

The stars fit into the jig perfectly, I couldn't ask for better.  But I am surprised about what happens next.  I fit the first star into the jig and roll on the glue completely covering the surface and the star curls up.  Now I understand the science behind this (it's the Wood Spirits,  the glue tickles).  It's easy to flatten out when I lay down and position the second star which also fits the jig perfectly.  Clamp up and we're good.  Now twenty minutes later release the clamps and repeat the operation with star #3, which also fits perfectly, and curls and flattens easily, but low and behold, the glue up of the first two stars, which had previously fit perfectly, now does not fit, and while sitting out of clamps it has taken the opposite curl.  Did I mention rascally.  Clearly the fibers are swelling from the water in the adhesive, but I would have thought this negligible*.   And perhaps it would be in other circumstances, it only looks to be about a1/64", but I built this jig to 4 decimal places of precision so 1/64" is a lot.  

Still it's just veneer, so cajole it in place.  This 1/16" veneer is behaving a lot more like a piece of lumber than a piece of paper.  Under full clamping pressure that corner will not pop into place between the pins.  

*Negligible definition:   Adjective.  A factor or parameter that you should have paid heed to but didn't and are now trying to justify having ignored.  

Here's a shot showing three stars out of clamps and taking the reverse curl.  

 

I now have a grand total of half a dozen stars glued up, so making progress.  

OK, that was a picture of 5.

If I was going to write down all the ways this project could go a crapper it would run a page and a half, I mean if I single spaced.  Might have been better if I hadn't told you folks what I was doing, but I still think it will work.  

 

[pkinneb]

This is so far beyond my capabilities...not going to lie though I am enjoying following along. Thanks for taking the time to share!

[Chestnut]

i'M STILL FOLLOWING ALONG INTENTLY EATING MY POPCORN.

bah caps lock.... #@*$ it i'm not retyping that.

[JohnG]

I'm glad someone has the skill, creativity, and energy for projects like this because I sure don't!

[Byrdie]

Still think you should've cut the basic shape out of a board (considering grain) and then sliced the wafers on a bandsaw, maybe sanding them a bit after slicing.  I don't think that would've avoided the problems you're having in glue up now but the edge grain would've looked interesting.  Not sure that using the CNC added much.

On the other hand, whatever the method I'm still anxiously awaiting the outcome.  Don't give up - might be some tweaking involved in the final form but it'll be worth the trip and make the next time more interesting.

[Mark J]

Thinking about it some, what I wish I'd done was rotate the stars 45 degrees when cutting out the veneer.  That way the grain would run diagonally from one point and one set of pins to the opposite corner.  This dimension would have been stable in length and the curling would have been perpendicular to this line, too.  If I ever do this again....

[Mark J]

I haven't been able to focus as much time on this as I would want, but I am up to 15 stars.  However, this morning there was something unexpected.  The last glue line has turned black.  I posted this under Finishing as I thought it would get better exposure.   

Otherwise I'm up to 16 stars:

[Mark J]

Well no answer to the black glue question, but no recurrence, either.  Not worried about it.  I am sure that it will be gone with the final shaping and sanding.

So at this point I have about 25 stars glued up, but I've run out of glue.  Off to the store later today.  I set a timer after every star is placed in clamps, but It's still taking longer than I thought it would mostly because I wander off and don't hear it ring.  There are some slight variations in the sizes of the steps between each slice and I did manage to glue one star on turned 90 degrees.  You knew stuff like that would happen, but I've chosen a very bland (read that forgiving) grain pattern so in the end I don't think these will show much.  But I won't know till I get there.  

I am rethinking the design--within the constraints of the blank I am making up. There's something about it that I think could be better.  That process will take a couple of days, but so will the glue up.  

[Mark J]

This going a bit glacial, but 32 stars.  At 36 I switch to the smaller ones.

[wtnhighlander]

I wonder, Mark - would it have made the actual turning part go faster if you had cut away some of the central waste from the stars that contain the hollow of the bowl?

[Mark J]

You could program the CNC router to remove a central circle from the first x number of stars.  This would be similar to drilling out the waste from the blank with a Forstner bit.  Make the hole the right size and you could use it mount the chuck jaws.

[Mark J]

A lot of distractions these last few days, but I'm up to

[Byrdie]

Oh great.  Now my YouTube search is going to be full of that ...

[Mark J]

Well if it makes you feel better we've left that number behind.  56.  Only 42 more to go.  Apparently I was delusional,  I thought this glue up would take a week or so, but in reality on a good day I'm getting 5 or 6 stars glued up. 

I am a little concerned that some of the stars may not be adequately bonded.  The problem is with that tiny bit of swelling the veneer does when wet with glue.  The tip of the star on one end always overrides the pins.  Up til now that has not been a significant problem as the press forced the layers together.  Now the growing block has become so thick as to resist the clamps and leave a significant gap.  Whether or not this matters will depend on whether this section ends up being turned away or remains part of the final design, but I'm going to have to inspect for these and try to get some adhesive in there (hence the question about low viscosity epoxy in "Finishes").  

I am going to start nipping the tips off with a knife to make sure the block lays flat against the next star.  

[Mark J]

So this turns out to be an interesting problem.  And I just gotta stop and say, I hate interesting problems.  I mean if you're gonna have a problem couldn't it be just a boring I've solved this a hundred times problem?  No, it's gotta be "interesting". 

So the stack has developed a rocker bottom.  I don't know if this is because the tips of the stars have had some upward stress in the jig and this is a cumulative result, or if this might be that the glue layer is somehow thicker in the body of the star than the arms, but there is just enough curvature that I can't really get the next star to make good contact along all four arms regardless of the pins.  

Not sure how to address this issue so I'm going to let the glue up sit for a bit while I think. 

I wish I had a thickness planer (but I don't), assuming the top is flat I could just shave off a little from the bottom and return to flat.  I could probably find a club member who could help me.    I could place a thin shim under each arm, but I would probably have to abandon the alignment pins.  Any thoughts?

[Mark J]

In between glue sessions I have been re-working the design for the last couple of days.  In this version the pillars or bands rise up a little more steeply and roll into a tighter curve at the top.  The basin then droops or hangs down a bit.  The basin itself is the same diameter as before, but is a little fuller than the previous version.   I'm not sure whether or not I will take the time for another model, but I can incorporate some subtle changes without remodeling.  Do you like the earlier or later version better?  Any (preferably small) changes?