I found a small riding lawnmower differential on amazon for only $50, so going to try building a 6 leg version that will hopefully actually work when completed.
Box or C construction on linkages - Instead of I beam construction (where parallelism and distance of two sides completly depends on the center portion being a precise size with perfectly perpendicular edges), will instead just box in the linkages with thinner plywood. This will improve tolerances and avoid extensive sanding / hand fitting I had to do previously
Reduce scale - Everything will be scaled to 80% the size of previous version. Instead of the main triangles being two 1/2" thick plywood pieces glued together (which took significant amount of time to clamp each one) the main triangles will just be a single 3/4" piece of plywood, reducing construction time
Fewer joints in metal frame - Metal frame will be simpler because only two arms instead of four will support each set of 3 legs (will depend on the rigiditly of 1/2" main axle for supporting middle leg). Use angle to connect the size to reduce chance of main part of frame warping during welding
Simple feet - Instead of bolted on feet, will simply wrap the rubber stair tread around the bottom of the plywood
Triangles are strong - Instead of the cantilever plywood arm design used previously that broke in multiple places, will stick with stronger triangular approach
Differential - The main sprocket will be on a differential out of a riding lawnmower, this diff will have keyed straight couplings
Things that will be the same:
CNC cut plywood for the legs, metal square tube for frame
Delrin bushings and 1/2" axles everywhere
Weld in crankshaft, the crankshaft will be constructed in the same way (bearings pressed into tube, then crankshaft sides welded onto rods going through bearings in place)
To connect the diff output shafts to the crankshaft will just use straight keyed couplings, which required turning down the output shafts to 1/2" and adding the keyway
Started off using the 3 jaw, but the jaws are so bell mouthed that I couldn’t hold it with only 1" of stickout
When cutting the 1/8" keyway I think I was way too conservative on the feedrate, causing the bit to simply rub instead of cut. I destroyed a couple HSS bits before I found a carbide one that was able to complete the cuts
This is the inside of the diff, note the precision ground pin that goes into the end of each of the output shafts, thats what avoids there needing to be bearings in the diff housing itself
The entire assembly will be just under 3 feet across. I tried to make the legs as close together as possible, here there is 1.2" of clearance between the closest parts of each leg
Got all the pieces for the frame machined, there is a lot more repetitive work with 6 legs. Parting and then boring out all of the bearing holders took awhile, with each of the 8 tube pieces needing to be taken in and out of chuck 5 times (part, bore one end, bore other end, chamfer one end, chamfer other end)
When cutting out the plywood leg pieces on the CNC router will also cut out an assembly jig to hold each of the 3 pieces in the correct alignment for welding
Also picked up some actual baltic birch plywood to use, waiting on smaller diameter delrin rod (going from 1.5" diameter to 1" diameter) before cutting the pieces out of the plywood, need the precise dimension to get a 5-10 thousandths interference fit
One thing I learned making mk1 bushing is to not trust that the cheapo 1/16" amazon parting tool blade is parallel to the body, so this time I made sure to align the blade itself using a test indicator
Got a 0.501" reamer to use, the .501 minus pin would have been a press fit, the .500 minus pin fit with no slop
The 1018 cold finished rod that will be the axles everywhere is 0.4990-0.4995"
Did some test cuts to dial in the press fits and accuracy of the bit I got for the CNC router, I will slow the feed way down and run a seperate 10 thou cleanup pass to cut it to final dimensions, aiming for the interpolated holes to be +/- 0.0005", which may be a bit optimistic.
Going with a 1 thousandths press fit for the axle to plywood (which just has to keep axle from sliding out left/right), and 10 thousandths press fit for the bushing into the plywood.
I was surprised to see that the delrin inner diameter actually shrinks by ~2 thousandths when its pressed into a 10 thousandths interference fit, will have to run reamer back through each of the 72 bushings after they are pressed in
Next step will be cutting out all the parts from 5’x5’ and a 4’x5’ sheets, going to be cutting some extras (enough part for 8 legs instead of the 6 needed) just in case
Took only about 30 minutes to run the CNC router, I probably should have slowed down the RPM for the small holes where it was only cutting 10 thousandths as a separate last pass with a slower plunge ramp feed rate (cutting a 0.498" diameter hole with a 0.375" cutter), could see some discoloration on the mdf after the first 10 holes and on the bit from overheating, (was never any smoldering or smoke)
Had a close call while using the CNC router, I was using a 3/8" downcut endmill (Onsrud 57-320) at 100 inches / minute and 18k spindle and 3/8" depth of cut and stalled the machine.
What amazed me was that
The 3/4" 4’x8’ sheet did not move as the machine was pushing into it as hard as it could with a stopped endmill
Did not damage the end mill at all (I had it choked up in collet as much as possible)
Was able to rehome the machine, reset depths, and then resume with slower speeds and feeds and complete job with no issues (avoiding trashing a large sheet of plywood, I hit pause instead of the estop when it started bogging down)
I thought I was running at a conservative feedrate by doing half the manufacturer recommends (as far as I can tell they recommend 200 inches / minute at 18k spindle speed and full diameter depth of cut) but I guess the spindle just didn’t have enough horsepower for that much removal through birch plywood
Here is where it stalled, after I had already cut lots of inner parts, I guess it just never ramped up to full feed rate for the smaller parts
Cutting out the ~100 plywood pieces was very fast on the CNC router, hand sanding each piece is very slow. After some deburring I got the delrin bushings pressed into all the pieces, it ended up being about a 10-13 thousandths press fit which seems very secure. After arbor pressing in the mk1 bushings I just went with a hammer to insert the 72 bushings.
To make it easy to start hammering the bushings in I made an external debur/chamfer tool out of some scrap 1.5" aluminum rod I had laying around. My initial plan was to mill some flutes into the aluminum, but then I realized that just super gluing some bits of sandpaper into the funnel works.
The interference fit of bushings actually shrunk the delrin a couple thousandths, so I reamed them out again (0.501"). To speed up the 72 bushings to ream I just let the initial taper auto center then held the piece by hand. (First I checked adjustment of bridgeport head to be sure it was perpendicular to table, saw about 1.5 thousandths off over ~6 inches and was able to improve it a bit)
Most of the axles pieces are fine at the rough cut (+/- 1/16") length, but I needed to make the crankshaft pieces a more precise length. Came up with a much better process than I used for mk1 pieces. First I turned one end of each piece on the lathe, then I used the bridgeport and a stop to use an endmill to cut the other end at a consistent precise length
I was actually suprised that the finish looked much better with the endmill than the power feed facing cut I did on the lathe. (Using the random cutting insert that has seen who knows what abuse probably didn’t help). Here are some examples after a quick chamfer in the lathe, half of them will be drilled and tapped later
Was about to draw up a tool to make to press in / out the friction fit axles instead of hammering them when I stumpled across double headed c clamps on amazon, I had no idea these were a thing, exactly what I need to press out the 1/2" axle easily (need something smaller and cheaper than typical ball joint service tool / c frame press)
Any application where one needs absolute perpendicularity referenced from one side? I can those being useful for numerous metalworking applications where the lack of linear clamping wouldn’t be an issue nor would the potential to press the workpieces through the gap be an issue unless a really mean monkey was doing the clamping.
I can think of some family members who likely have zero idea these exist, who will probably be getting some for Xmas. (If I can remember that far ahead)
You could always plan ahead and buy them now, put them in a “safe place,” then be unable to recall that place when the holiday season approaches. Finally, you can rediscover them two years later in the back of a closet.
Wanted to polish up the steel rod axle pieces so got some light grey scotchbrite and was very impressed at how easily it improved the look of the axles with just a quick spin in a hand drill
Will use the axle pieces I cut recently as alignment pins in the welding jig that was cut on the cnc router at the same time as the other plywood pieces, getting close to constructing the frame
I’m also glad I went with the make 2 extra of everything approach, so that I have spares if I mess up at a step without having to go back to the beginning. Here I broke a tap in one of the crankshaft pieces, but I had another so I just had to repeat this one op
Here is the clamp I’m currently using to quickly press out axles. It works much better than my previous hold 4 things (clamp, delrin, strut, smaller rod) at once approach, but even just pressing out 6 my wrist got tired from all the twisting. Would be a good idea to take the wooden handle off and weld on a nut so I can use a drill to drive it in / out
The other good test result so far is the lateral rigidity of a leg, clamping the foot end to my table it’s only the flex in the plywood that allows the opposite end to move 1/4" inch. One of the design goals was making it as narrow as possible, so that means each leg is depending entirely the strength of the 3/4" plywood for lateral rigidity. As long as it’s ridden on fairly flat ground and you don’t try to take quick turns my hope is the lateral load will be minimal
You can drill a diametric hole through the handle of your clamp to insert a t handle for turning the handle with some mechanical advantage. You can then take it out when you don’t need it.
Have all the wood pieces glued together, I may spray some poly on all of them, but I would need to buy a larger heater to do that in my garage this week.
3d printing spacers worked out great while glueing up struts, all the joints fit over each other with a nice 10-32 thousandths clearance, much unlike mk1 where I had to hand sand each joint to proper clearance.
)
