Saturday, November 11, 2017

Building a Canoe Paddle

Aside from building things, my main recreational activity is canoeing.  I paddle mostly on rivers that are shallow and rocky, so I split a lot of paddles.  I typically buy the Bending Branches Loon model.  It's cheapish and light, feels good in my hands, and I can buy them locally.  I really wanted to make my own paddle though, so that's what I did.  I teach high school wood shop, and one of the great benefits is that I have a really awesome wood shop all to myself in the summer.

The first thing I did was to watch the How it's Made video on canoe paddles.  It turns out this is Grey Owl's facility, and I think they are making the Voyageur in this video.  I mostly did it the way they do it, with some adjustments to suit my tool availability.


I also watched this video from Sanborn Canoe Co.  I mostly didn't do what they do, but I really like the way their paddles turn out, and I would like to incorporate more of their techniques in my future builds.


First, I glued up the blade halves.  This looks like it's going to end up as a single glued piece of wood, but there's no glue between the two middle pieces of walnut.  For my blade I used walnut closest to the shaft, then cherry, then some more walnut, then maple on the outsides for impact resistance.  My paddles get beat up on the sides of the blade the worst.
Gluing the blade halves
 While that was drying, I glued the shaft.  It is made of basswood for light weight and stability, with a walnut stringer down the middle.  The 1/16" walnut sliver was the hardest part of the entire woodworking project.  At the time I didn't have a drum sander, so I had to cut it perfectly on the table saw.  I really dislike cutting thin things on the table saw (and the only piece of walnut that I had at the time was warped enough that planing it flat would have made it nearly non-existent).  Now that I have a drum sander I would have cut this piece on the band saw and sanded it flat, but it worked out OK anyway.  I glued it all against the table to keep it straight.
Gluing the shaft
 After a few hours I glued it all together, including a few pieces for the handle.  I intentionally left the handle long on the blade end for two reasons.  The first is that my thickness planer snipes badly, and I wanted it to snipe that leftover part instead of the blade.  The second will become apparent when we get to the CNC shaping of the blade.
Gluing it all together
 I made sure the handle matched the wood on the blade.
Future handle
 After it dried for a few hours I ran it through the planer, taking off only what I needed to get it flat.
Thickness planer action
 At this point I got really excited.  If I had to do it again, I would have used single pieces of cherry on the blade instead of the two pieces glued together, or put a maple sliver between the pieces.  Not a giant deal, but next time.
Looking good
 Here's the handle.
Handle
 The next step was to cut the profile, but I didn't cut the end of the blade.  This is for when I shape the blade on the CNC, after I flip it over it will still lay flat.  Pictures below.  Note that the shaft is thinner than the shaft glue-up.  It is 1" wide, and I planed the whole thing to 1.25" thick, to match the handle of the Bending Branches Loon that I currently use and like.
Mostly cut out on the bandsaw

Cutting straight lines on the shaft was hard for me
 Next I routered the whole thing, both sides, with a half inch round-over bit  This was mostly to shape the shaft and handle, and to provide a neat transition between the shape of the shaft and the upcoming blade taper.  I wish I had a quarter-oval bit that would make the shaft oval instead of a rectangle with rounded corners, but I've never seen one.  I could use a custom shaper blade like in the How It's Made video, but it's probably not worth it at this point.  I would really love to hand shape the shaft with drawkinves and spokeshaves, but while I have both of these tools, I'm not yet adept enough at sharpening the blades to make them work well.
Beginnings of a nice shape to the walnut piece

Basswood burns badly when routering its endgrain, apparently
 Before I was a wood shop teacher I was a drafter, and I teach some computer aided drafting classes too.  I used these skills to model my paddle blade in Autodesk Inventor, then use the (awesome) Inventor HSM to make the g-code to shape the taper of the paddle.  I would like to make the taper a more complex shape and add a small spine with the shaft transition, but I decided to keep it shaped like the Loon that I know works well for me.  I have read that a spine can cause flutter if not shaped correctly, but mostly I'm lazy and I just wanted to get this done.
On the CNC!
 Here is a shot that shows the main reason I kept the blade too long.  If I hadn't, when I flipped it over to cut the other side it would have been difficult to hold to the table.  If I didn't have a CNC, I would have cut it on a bandsaw, on its side.  I doubt I would have been able to do as good of a job as the robot though.  Again, someday I hope to be able to hand shape all of this with planes, chisels, and spokeshaves.  They say, "To a man with a hammer, every problem looks like a nail."  My hammer is my CNC.
Still with the leftover on
 At this point it looked like a paddle, and I was very excited.  There was a LOT of sanding to do, and I was envious of the folks in the How It's Made video's giant belt sanders.
Looks like a paddle!
 I was very excited about the way the wood joints curved, and spent a lot of time making sure the transition between the shaft's 1/2" router shape and the blade taper looked good.
Walnut brings out a nice shape
 I have a 2" diameter oscillating spindle sander that I used to shape the finger and palm part of the handle.  I think it turned out particularly nice.  I was very happy that I spent the time to get that sliver of walnut down the center cut and glued in the handle.  It brings out the shape nicely.
The handle
 My next step was to give it some tip protection.  I beat the daylight out of the blade tips of my paddles, and if I had left this paddle as is, it would split and dent, especially where the soft basswood is exposed in the middle.  I decided to make an ash tip guard, and I wanted it to be connected with a tenon and lots of glue surface.  It took a long time to plan, but I decided on a two-part tip with a two-step glue-up.  Here is the first half of the tip after being cut on the CNC.
Half of the ash tip guard
 Then I cut the mating profile onto the blade.  The walnut strip down the middle made positioning the bit in the middle of the blade much easier when I started the program.
Cutting half the tenon
 Next I glued the two pieces together.  They are being clamped up and down, and also along the length of the shaft.  I built a clamping point out of 2x4s to pull against.
Will the gluing never end?
 Then I cut the other half of the tenon on the CNC.
2nd half of the tenon
 At this point I shaped the end of the blade to its final shape.  This would help guide me in knowing where to cut the second half off at, as the tip of the blade wood would be hidden after the next gluing.
Final blade shape
 this is the CNC of the other half of the tip guard.  Note that it does not have a step, so the two pieces of ash will not be glued along the centerline of the blade, but rather offset along one side of the tenon.
More tip guard wood

Here's how it will fit.
 More gluing.
Gluing never ends
 While I was waiting for the last of the glue to dry, I CNC cut this piece out of a 2x4.  It has an elliptical profile, and I used it with sandpaper to give my shaft a more pleasing profile.  It's like a poor man's shaper.
Shaft elliptification
 At this point I could hardly wait any more.
 After the glue had "dried"
Impatient woodworking
 Finally it was time for the big hand sander for final shaping.  Things went fairly quickly.
Belt sander ftw
 Next time I want to make the middle lump pointy-er.
It ended up looking pretty good

Nice.

You can't really see the glue line between the two pieces of ash
 I needed to fiberglass the sides of the blade for strength, and I have a laser cutter, so why not cut the fiberglass on the laser?  Lots of reasons, it turns out, but I didn't know them at this time.  Don't do this.  The edges burn ever so slightly.  Not everywhere, but enough places that it took some time to cut out the burned parts with scissors, which sort of made the laser cutting pointless.  That wasn't the main reason though.  More below.  Also, the fiberglass you buy at Lowes is 6 oz. fiberglass, which holds way too much resin and adds weight.  4 oz. would have been better, or even less.
Laser cut fiberglass cloth.
 Speaking of lasers, I used the engraver to burn my name and a logo into the basswood, which turned out nice.  You have to trick this unit into thinking the front door is closed with magnets.
"Lasers"
 I mostly canoe the Buffalo National River in Arkansas, and it's thanks to the hard work of Neil Compton (along with many others too) that we have this treasure.  They fought for years before I was ever born to save it from being dammed.  He is no longer with us, but his canoe is on display in Bentonville and it has this triangle pattern panted on it.  I decided to engrave it into my paddle as an homage to him.
Thanks Neil!
 Everything had gone really well, and worked on the first try, up until this point.  We are now two days into fabricating this paddle, but the rest of the job would take a week and a half.  I want to express at this point how much I dislike the process of fiberglassing.  It's important, however, for the lifespan of the paddle.  You can make the blade much thinner if you strengthen it with fiberglass, which saves weight and allows it to slice through the water easier.  I decided to use epoxy resin instead of the more "normal" polyester resin because it's supposed to stick to wood better.  I had used Bondo polyester resin (which I bought at Lowes) to repair a paddle once, and it worked, but indeed it delaminated from the wood in several places after time and abuse.  The epoxy had almost no odor, which was great, as polyester resin smells truly awful.  I used Evercoat because I was able to buy it locally, although it was a hassle finding it.  Nobody stocks epoxy resin, it seems.  Epoxy resin costs WAY more than polyester resin.  Since completion I have been recommended to use U.S. Composites 635 thin epoxy resin, and it is much cheaper than most I have found.
 Here is what NOT to do.  I put the cloth on the face of the paddle, then poured the resin over it, like I saw on a YouTube video.  I should have read the comments though because after I went back when this day was over and watched it again they all said not to do this.  The cloth will float on the resin, which makes a space under it full of resin which adds weight but not strength, and will make the face look lumpy and not smooth.
Don't do this

Ugh.  Lumpy.
 What you SHOULD do is paint down a thin coat of resin, roll the fiberglass cloth onto it to stick it down, then add a bit more resin to fill the cloth.  That's what I did on the other side and it worked great.  Here's the problem with the laser cut cloth though, and that's that it should have overlapped the edges so that I could have cleanly sanded them down when the epoxy hardened.  Instead I got this annoying ridge all the way around, near the edges, that took a ton of time and resin to fill.
Laser cut fiberglass equals misery
I was also distressed to find that the epoxy resin darkened the basswood significantly.  I lost a lot of the contrast in the wood that I liked, but I guess I should have expected it, as it ended up similar in tone to my current basswood paddles.  I'm not sure what I'm going to do about that in the future.
White wood is now tan wood
 It took a lot of sanding and adding more epoxy, then sanding more before things started looking good again.  Every coat took 24 hours before it could be sanded too.  Ugh.
Before the final epoxy coats
 I ended up having major fish-eye problems with my epoxy, and I never did figure out why.  Eventually I just stopped adding coats and brushed a final coat of Minwax water based Spar Urethane over everything for UV protection.  It turned out all right I guess, but it was way more work than it should have been.
Finally!
 I didn't weigh it before the fiberglass, but it sure did feel heavier afterwards.  One pound thirteen ounces isn't terrible, but it's heavier than I wanted.  I could have thinned the blade more, I think, and that's where you want the weight reduced.
A tad on the heavy side
 Overall I'm happy with the way it turned out.  I'm going to try to use and abuse it like a normal paddle.  It's been a fairly dry late summer and fall paddling season though, so even though it's November now, I still haven't had a chance to try it out.


Since I built this one, I mostly completed another paddle, made mostly of western red cedar, but the CNC gave up in the middle and plunged a giant gouge right down the middle of my blade.  Another reason to switch to hand tools?  We have a new control board and everything seems to be back to normal.  I have also purchased the book Canoe Paddles: A Complete Guide to Making Your Own, and I would recommend it.

Thursday, January 14, 2016

Creating Involute Bevel Gears in Autodesk Inventor Using the Zweerink-Snider Process

In my high school CAD II class we are 3D modeling radio controlled cars, 3D printing the parts, and racing them.  Everybody is super pumped and things are going well, but our drivetrain options are somewhat limited.  We currently use pulleys press fit to the output shaft of our motors and rubber bands to transfer power to the rear wheels, but rubber bands slip badly, break often, and are very inefficient, wasting a lot of power in the form of friction.  Some students discovered Inventor’s ability to make spur gears with a true involute tooth profile, and their results were noticeably superior to rubber bands.  We assumed that we could use that knowledge to make involute bevel gears as well, but Inventor lacked the “export tooth shape” feature on the bevel gear generator that made accurate spur gears possible.  Inventor can generate several types of gears, but they are all simplified for visualization purposes only, and are nearly worthless for 3D printing or CNC machining.  Only on the spur gear generator does it have the option to export a true involute tooth shape, which can be used to make a rapid prototyped, functional gear.  How to transfer the involute shape of a spur gear into a bevel gear design was a problem that seemed simple at first, but turned out to be extremely difficult to figure out.  With the help of one of my students who was also taking trigonometry, we were able to come up with a process that generates working bevel gears using the exported tooth shape from the spur gear generator. 
Here is how we did it.

My 3D printer is an Afinia H480, which I recommend heartily to all teachers, and I use ABS filament for strength.  I have found that the finest functional teeth that I can reliably print have a module of 1mm.  This means that if the gear has a diameter of 24mm, it will have 24 teeth.  Lego gears have a module of 1mm.

Open a new assembly file and save it.  Open the Design tab, and click Spur Gear.  Expand all expanders, to the right, down, and then the” <<” next to the cancel button.  In the bottom section change input type to “Number of Teeth” and size type to “Module”.  Change Design Guide to “Center Distance”.  Now you can enter the number of teeth you want on each of the two gears.  A pressure angle of 20 degrees works fine.  Keep your helix angle at 0.  Set your module to 1mm.  It doesn’t matter what your facewidth is for what we’re doing.  After you have your data entered (Module and Number of Teeth), you can hit calculate.  I’ve never had it tell me my gears would work.  It’s always “Calculation indicates design failure!”  Ignore this.  Click “OK” and accept the failure again in a pop up box. 

Now you should have a couple of gears on your screen.  These gears are not ready to be used.  If you zoom in you can see that they overlap with interference.  We  need to right-click on one of them and choose “Export Tooth Shape”.  We will have to do the rest of this procedure twice unless your gears have the same tooth count as each other.  Once for the pinion (the smaller of the two gears) and once for the gear (the larger of the two gears).  Use “Normal” backlash, and choose the largest value it will let you enter.  For 1mm module gears, it seems to be about .006”.  This backlash will keep the gears from interfering with each other with an imprecise 3D print.  Click “OK”.

Now Inventor will take you to an .ipt part, which will be a cylinder with one of the spaces between the teeth on a sketch on the end surface.  If you were making a spur gear you would make a cutting extrusion of that space, then do a 3D circular array of it in the amount of teeth you entered in the gear calculator.  We, though, are going to delete the extrusion, but leave the pinned Sketch1.  Edit that sketch, and delete all of the construction circles.  Next do a 2D circular array of the tooth cut, then trim the outer circle so that the sketch shows the actual gear profile all the way around.  Click “Finish Sketch”.  Now make a new sketch on a plane perpendicular to that gear sketch.  I use the YZ plane.  Draw a centerline to the right, from the origin, which should be the center of the gear.  Make it pretty long.  Now draw a construction line straight down, from the origin, at least as long as the gear radius and then a little bit more.  Now draw a solid line from the origin, down and to the right, a little bit longer than the gear radius.  The angle between this line and the “down-from-the-origin” construction line will be 90 minus the inverse tangent (aka arctan) of the gear ratio divided by two.  If I have 32 teeth on my gear, and 8 teeth on my pinion, my gear ratio is 4.  The inverse tangent of 4 is about 76.  90-76=14.  Half of 14 is 7.  My angle from vertical for that line will be 7. 


On the Windows calculator, use the “Inv” button to make tan into inverse tan (tan^-1).
Next, draw a line from the end of that line to the other end of the centerline.  Then, from that end (end opposite of the origin) draw another line to the angled line sort of close to the end further from the origin.  Make that line a construction line.  It should look like this:


Next, you will place an angle dimension between the centerline and the construction line.  This angle will be 90 minus the inverse tangent of your gear ratio, which in this case will be 14. 

At this point you will go back to your assembly file, right click on the gears, and click “Edit Using Design Accelerator”.  This will bring up the spur gear component generator.  Click the notepad in the upper right corner.  This will bring up a tab in your internet browser which lists the gear parameters.  We are interested in Pitch Diameter and Outside Diameter for the gear we are working on.

Back in the sketch, place a diameter dimension from the centerline to the lower corner of our triangle, and make the dimension the Outside Diameter, plus .001”.  The Outside Diameter is rounded in the chart, and if our soon-to-be-revolvoed part isn’t bigger than the gear sketch that we are going to project on it, it won’t work, so don’t forget to add that .001”.  Now, for the lower point of the construction line, place a diameter dimension from the centerline, and make the value the Pitch Diameter.  Everything should be purple now, and the sketch is ready to finish, so click finish sketch.


Next, use the revolve command to revolve the triangle along the centerline.



Our next step is to get a point at the far tip of the cone, so we need to make an axis through the cone, and place a point at the place the cone and the axis intersect.  Click axis and click the cone to make an axis, then click the down arrow next to the point button and choose Intersection of Plane/Surface and Line.  Then choose the cone and the new axis, and there’s your point. 

Next click Start 3D sketch (below Start 2D Sketch), then click Project to Surface.  The Faces is the near side of the cone, and the Curves is the gear sketch.  This will project a flat sketch onto the curve of the near cone.  The effect of this is that the teeth are taller because they are now on the hypotenuse.  Without this step the teeth would be too short when lofted to the point at the far end of the cone.  It doesn’t make as much difference on the small gear, but on the larger gear the difference is significant.  Anyway, now we use the loft command to loft the projected 3D sketch to the point on the far cone.  You will use the Cut or Intersect button in the Loft command to make this happen.  I am not sure why, but one or the other button works, and the other one doesn’t, but it doesn’t seem to be consistent.  Just try them both and find out which one works.   At this point you have a functional bevel gear, but not a practical one.

You will need to make a plane with a sketch and a cutting extrusion to cut your gear off at the facewidth you need.  Don’t forget to use your Slice Graphics (F7) button.  

Then pop a hole in the gear for your shaft.


The steps are exactly the same for the other gear, but you need to remember that because the inverse tangent of 4 is 76, on the smaller gear we used 90-76=14 for our angle.  On the other gear we will be using 76 as our angle.  Why is that?  The “pitch cone” of two gears with the same number of teeth will be 45 degrees.  No matter what ratio of teeth they have, the angles of the pitch cones must add up to 90 if the shafts are at a 90 degree angle.  The angled construction line in the revolved cone sketch was our pitch cone line.