Archive for December, 2008


Machine tool withdrawal

Dec 25, 2008 in Stuff

It has been 75 hours since I arrived back in Atlanta.

I think I’m hallucinating about building stuff.

There’s nothing to do here.

Someone save me.

LOLriokart Update 8: Did I mean later? I mean now.

Dec 21, 2008 in LOLrio Kart, Project Build Reports



in 13 hours. There was much Mountain Dew involved.

I got the gears, shafts, and bearings in Friday afternoon, to my surprise, as the mail desk folk usually take off early. At first, I wanted to leave it as an exercise in Atlanta. Take the last weekend off to relax, etc.

Then I came to my senses. Building IS relaxing, after all, so I crossed the Antarctic continent to MITERS, where I spent the next day watching piles, both of snow and of metal chips, grow in height.

So let’s begin.

Here’s the raw materials rundown on Saturday at 7AM. Two 21 tooth, 12 pitch spur gears. Some lengths of precision-ground 3/8″ shafting. A handful of bushings. Four large bearings. Spur-gear-on-a-stick (pinion rod). The 3/4″ shafting and large sprocket came from the parts bin.

First order of business was roughly machining all the raw parts to final dimensions. Since roughing and finishing operations require different tools and attention spans, I figured it was better to do this now than repeat the process for each part.

Cutting the hubs off the spur gears.

Yes, I know I’m supposed to use a collet chuck for these things to avoid grunging the teeth, but I found that the tips of the chuck jaws fit between two gear teeth nicely. This positive mechanical contact meant I could shove a little harder on the parting tool and have nothing blow up.

I needed 6 little shafts. To cut them one-by-one would have been unnecessarily repetitive, so I stacked the raw shaft stock on the horizontal bandsaw and clamped them down with a block of wood. Without the wood,  at least one shaft would not clamp properly due to a variety of misalignments and tolerances. This way, I could cut 3 at once.

I wasn’t as lucky when cutting the spur gear rod into smaller chunks, since I only had one spur gear rod.

It’s giant sprocket time. The outside jaws for the chuck are worn pretty heavily, so it’s a bit of a trick actually getting something to run true, especially at large diameters, such as a sprocket hub.

Solution: Squish the sprocket between a lightly-tightened chuck and a live center, then briefly power on. This causes everything to fall into alignment with the center. Then the jaws are sequentially tightened and the center removed.

Then cram a shiny, new solid carbide boring bar  into the tool holder, kick the spindle into the highest speed, and take .075″ deep passes with reckless abandon until the .625″ bore becomes a 1.625″ bore. While it didn’t take long, I did try to justify to myself the purchasing of a CNC lathe for MITERS.

I got tired of dodging flaming-hot curls after a while, so I set this chip shield up using two magnets and a piece of grungy acrylic.

With one end of the assembly mostly done, it was time to cut a chunk of one of my giant billets and finish the other side.

The horizontal bandsaw cuts almost as straight as the outside jaws on the lathe are true (i.e. not very). Since I had to have some kind of reference surface, I planed off the cut end on the mill first. The V-block rests on a parallel, and keeps the round more vertical than it would have been otherwise.

To do it quickly, I used the mysterious not-indexable-but-not-single-piece-either facing tool. What the hell IS this thing? Does anyone know?

With the round stock shaved down, it was back to the lathe for final dimensioning.

Finally dimensioned. Since aluminum is soft and carbide is awesome, I was able to pull off .125″ wide cuts at top speed while cramming as hard as I could, and have it turn out gorgeous AND precise.

Of course, cutting the part off was going to be the exciting part. Invariably it is when everything goes horribly wrong because something flexes and the blade jams. To make this worse, it’s a large diameter part on a 50 year old, student-abused machine. With an aluminum toolpost!

I scoured the Intergoogles to find out about cutter geometries that would mitigate this problem, and ended up carefully grinding a sort of U-shape to the tip of the parting blade. This curls up the material as it is being cut, and the sward comes out in neat little rolls instead of getting caught in the groove. It turns (LUL PUN) out that parting tools for heavy production use something like this to quickly separate part from stock.

Aluminum donut being machined. I had it on the indexing fixture, but didn’t use the indexing part of it, since all my holes were done in Cartesian coordinates anyway. A spot with a stub drill, a 1/4″ clearance drill, and crafty counterboring with a 3/8″ endmill later, the output carrier was done.

Shiny finished and semi-finished parts.  The second picture is shiny enough such that it may become a candidate for the site banner.

Pinions bored out and bushings installed. Since I had a capable boring tool now, I just set it to the right diameter and (again) crammed as hard as I could, on the highest speed.

8 flawless victories.

Repeating the hole pattern of the aluminum donut on the toothy steel  donut.  The sprocket didn’t actually fit on anything, so I had to pull out the clamp kit and strap it directly to the table, spaced by buffer wood.

With all the holes and miscellanea done, it was time for a test fit.

It works! Provided that I keep the output gears off to the side, the rotation is as expected, and there is no binding.

So now it was time to make the output shafts. Simple enough – they’re sections of 3/4″ keyed shaft with two grooves.

After digging through the pile of random lathe tools we inherited, I found a custom-ground groove cutter with a .075″ wide tip. A trip to the grinder turned it into a .05″ wide tip which fit the retaining rings I planned to use.

I used to despise retaining rings. Hated them, hated their concept. Mostly because I couldn’t get them off, while trying to grunge a cool part from some old crufy machine. But all that changed as soon as I got ahold of some good retaining ring pliers. Now I find that they’re the most low profile way to make sure something stays where you want it.

The shafting has keyways, so I needed a keyway cut in the gears. So, off to broach the gear bore…

Hey! That’s not a broach! That’s a 3/16″ carbide endmill plunge-cutting a weird half-circle thing!

Yep. We don’t have keyway broaches at MITERS, and I couldn’t think of a way to simulate a shaper-planer machine. So, in a moment of 5AM engineering brilliance, I found a solution in a spare 3/16″ round lathe tool blank.

The half-circle is where the keyway would be if I actually broached the bore. Instead, the 3/16″ rod acts kind of like a pin, and kind of like a key. Half of it sits in the square keyway on the shaft, and the other shaft sits inside the semicircle in the gear.

It’s a terminal case of round-peg-in-square-hole.

A retaining ring keeps everything in place.

I’ll probably reconsider the engineering merits of (read: throw loctite upon)  this assembly after returning from Atlanta. It should, however, not explode outright. The keyway and semicircle may mush a bit, but I don’t think anything will ever catastrophically fail.

And with the gears completed, it was 5AM Sunday morning. Time for the final assembly.

Yeah, I know, uncountersunk flathead screws. I couldn’t find standard cap screws of the proper length, and the only ones I had which were not too long or too short were the flatheads.

I could just countersink everything and be done with it, but that would involve taking it apart again. So, I’ll just get some normal screws later.

It works! Turning one shaft causes the other shaft to rotate in the opposite direction (in a visual sense, which is the correct result). Attempting to turn both shafts towards or away from me (which is actually a physical opposite rotation) makes the whole thing rotate. The motion is smooth and there is little backlash.

One downside is that I used the cheapest, shittiest ball bearings McMaster had. These actually come with some pre-wobble due to their nonprecise nature. So, unfortunately, I do have to use some outboard bearings if I don’t want grinding gears. Alternatively, I could get some real bearings. Something rated for power transmission paraphenalia, not handcarts.

Another shot of the gearing. This thing is enormous, and steel. To go all the way, I could have made the aluminum endcap from a steel round I had, but aluminum goes quicker on the machinery. Total weight is probably around 9 or 10 pounds.

Now to finish the kart so I can actually use it…

LOLriokart Update 7: Slowly but Surely

Dec 19, 2008 in LOLrio Kart, Project Build Reports

More than a month has passed since the last LOLriokart update. Does it move yet?!

No. I really haven’t had that much time to sit down and perform the design work that is necessary to generate parts. Since I’m practically doing all of this on-the-fly, I need to consider how future parts (i.e. the ones I’m making in 5 minutes) are going to fit with existing ones, and how they might or might not get in the way of things not yet designed.

Now consider that my weekly obligations tend to be sporadic, and that it’s the end of the semester, so everything is coming down to the wire. Throw in a few distractions like finishing the scooter and you get that the plastic milk crates holding up LOLriokart are slowly indenting where they meet the tube frame because it’s been sitting so long. Creep under continuous stress, or something…

But there has been progress made. I’m steadily pushing forward on linking the front wheels together and connecting them to the steering wheel. After that, the intent is to get the front brakes working. Then I’ll have a rolling chassis which can accept whatever powertrain I feel like building.

Here’s the progress over the past few weeks.

Robot Jesus blesses me.

Robot Jesus blesses me.

So the steering column will have two segments and three supports. The lower half is vertical, and is supported at its bottom and top by bearing blocks. The bottom support was easy – a bushing mounted to a segment of thick-wall aluminum rectangular tube, which is in turn split-clamp mounted to the original crossbeam which the shopping casters mounted to. This placed the axis of the shaft in a good position close to the front of the basket cutout, which is what I wanted.

The top support for this vertical section was harder. I didn’t have any billet or barstock of the proper size, and I didn’t think the aluminum tubing was going to be sturdy enough given that the wire frame provides no flat mounting surface.

Then Robot Jesus descended from MechE heaven to present me with another near perfect scrounge part. Now I really want to know what this crazy machine that was parted out by MITERS many years ago was – it has saved my ass a few times already.

And so it was that strange trunion-like thing with a 1/2″ Bronze Bushing of Convenience (+1?) only needed about a quarter inch trimmed off the flat side to align the steering column perfectly vertical. It even had large threaded holes on this side to mount on the basket!

....but stopped short of saving my soul.

....but stopped short of saving my soul.

Unfortunately, the front of the basket has an even number of vertical wires. The trunion-thing has an even number of holes. Since the bolts need to pass through the empty spaces between the wires, I wasn’t going to get a centered steering column with the stock trunion. No problem – while set up to take 1/4″ off the top, I drilled new mounting holes right between the existing ones, for three in total.

I also threaded the holes in the same setup, using the mill in low gear (and my big keyless chuck) to powertap the 3/8-16 threads. Apparently, I scare the crap out of everyone who sees me powertap on the mill. But it’s no problem if

  • The tap is huge, so it won’t just snap from the torque
  • The holder has some sort of clutch effect (clutched tap holders for mills are made especially for this purpose), or is huge enough that anything holding it is going to slip anyway if it is clogged or bottoms out. For more insurance, loosen the belt tension on the motor.
  • The quill is allowed to move freely, so the tap can pull itself in and out
  • You’re really fast on the motor switch.

With this setup and some mechiprudence, I have powerthreaded holes as small as 1/4″ and as large as 1/2″-13. The latter was pretty ballsy.

So what, I’m not beefy enough to crank large taps…gotta find a solution!

Next was the other side of the assembly.

To secure the shaft support from the other side, I quickly cooked up this clamp bar with 3/8″ through-holes.

Whole assembly fitted. The little rounded cutout on one edge is to hang onto one of the horizontal wires to keep everything more steady. It was made using a radius-tip (not ball-tip, just radiused) cutter that I got in a toolpile from Ebay.

This is yet another extremely overbuilt impromptu part that should never fail. If it does, I’m probably spread into a very fine film already, and thus would not have to worry about repairing it.  Other parts of this nature on LOLriokart include the steering brackets, rear bearing blocks, wheel hub and mounts, and… oh, wait.

I love billet aluminum.

Let’s move onto the steering linkages. While the big aluminum tube acting as the steering tie link was nice, it’s just too bulky. I came up with a new plan involving ball joints (or “rod end bearings”) and threaded rods.

A large threaded rod connects ball joints at both ends of the tie link assembly to a rigid center piece which mates with the eventual Pitman link (that in turn is rigidly mounted to the steering column). The ball joints go to the steering links on the front wheels.

It was Thursday night, and if I ordered ball joints from McMaster, they’d ship Friday morning, get here Friday evening, and be stuck at Shipping & Receiving all weekend. I don’t have the patience for that kind of stuff, so I just rolled (milled?) my own rod ends. Fortunately, the Media Lab guys bought a basketfull of ball joint bearings for the car project a while back, and ended up using a handful. I nipped two and carved some aluminum holders.

Installed. There’s a 1/2″-13 hole on the back side to accept the threaded rod, which I manually tapped this time. It was a great exercise in how to start a giant tap straight, and just great exercise in general.

I ended up using two steel pipe sections, 18″ long, as handle extensions on the tap wrench to give me more leverage.

Assembled. The ball joint gives about 10 degrees of tilt and roll in the steering linkage. I won’t need this, but it was nice to have, since I’m sure nothing is actually aligned properly.

Normal people would have welded all of this up and have been done in a day. I, however, have a disdain for welding because it’s just too permanent. Yeah, no such thing, right? I like having the ability to take my stuff apart. This will all come apart into neat aluminum billet pieces when I’m done – not ugly, globby things I have to hack off with a grinder.

Let’s hope it doesn’t come apart into neat aluminum billet pieces while I’m driving it. Maybe that’s why we weld…

I suddenly had a change of plans while building the linkage ends. Instead of using a single threaded rod, I could use two separate ones, and have the ability to tune the toe angle by moving the rods in and out and locking them in different places. Even better, why not use some big Grade-8 cap screws?

So the plan then became using the rigid center piece as a bridge between the two cap screws, using a clamp-mount type assembly at both ends to hold the cap screws right behind their heads. Thus, if I needed to tune the toe angle, I could loosen the clamp, turn the screws a bit, then lock them back down.

This meant I had to make the aforementioned rigid center piece longer than if I had just ran a threaded rod all the way through, which means I have to go buy more bar stock since it exceeded what I had available by a factor of 2. Effective use of available materials, people!

Or maybe I should go find cap screws 1.5 times as long?

Alright, enough fun for now. This is LOLriokart as of two days ago. It has clearly evolved from a shopping cart full of shit to a slightly more fancy shopping cart full of even more shit.

Ah, but useful shit.

This is probably all that will happen until I return from Atlanta for the remainder of IAP. However, I can still work on modules and parts while I’m in Atlanta – just won’t have the opportunity of knowing I’m doing something totally wrong mid-process.

One assembly that I intend to make in Atlanta which will make the project that much better, is a rear differential. As long as I’m not building a wheelmotor for each wheel (Hey, maybe that will happen in the future) and it only has rear wheel drive, then I should probably add a differential to the axle. Otherwise, the two rear wheels will try to rotate the same speed through a turn, which as anyone familiar with vehicle dynamics would know, is…bad.

I’ve sort of held off on this for a while, since getting the solid axle version running was a priority, but as long as I’m taking this long on everything, I might as well. Here’s the plan.

Wait… What the balls is that? That doesn’t look like a differential! There’s not a single bevel gear! This looks like a differential.

Correct. This is a design known as a planetary differential or a spur differential. It uses the principle that two mating spur gears rotate in opposite directions to produce the same action that a set of bevel gears does.

The two output gears are separated by a gap. Each output gear mates with a small, long pinion (longer than its diameter, such that it bridges the gap, but stops just short of touching the opposite output gear). The long pinion of each output gear in turn mate with eachother in the gap.

Result? Let’s rotate one output gear clockwise. Its long pinions will rotate counterclockwise. It mates with the long pinion of the other output gear, rotating it clockwise. In turn, the other output gear rotates counterclockwise, the opposite of the first. This is the same action that bevel gears provide when put into the classic box arrangement seen in open differentials. You may multiply the number of long pinion sets around the perimeter at will to get more strength.

Using helical gears and a slightly different arrangement gets you the Torsen type limited-slip differential. That’s an exercise for another day.

Benefits: No weird (expensive) gears. No weird geometries to machine. Can be made using standoffs, bolts, and stock gears. That’s what I intend to do. Except just big, so it can take the torque of the Etek.

Speaking of what the balls, I also drew up plans for a giant ball differential (as seen in R/C models) a while back, but at this scale, the design will be very inefficient.


Ãœberclocker rebuild is postponed

Dec 12, 2008 in Project Build Reports, Überclocker Remix

Ãœberclocker will not be appearing at Motorama.

A combination of several factors has led me to postpone the build for next semester or next summer. First off, it’s mid-December already, and Moto is in about two months. I have not started on anything, nor even gotten materials and parts. Finals are next week. Immediately afterwards, I’m flying back to Atlanta for about 3 weeks, during which time I will not have 24/7 access to machine tools, especially not a waterjet cutter (Dale, I sincerely recommend you buy yourself an abrasive waterjet cutter for Christmas! It’s a good investment, I promise).

After returning, I have all of 2 weeks left of IAP before the next term starts. So, half of the month I do have after that will be in conflict with classes.

This project is far too complex and involved to just wing it. I don’t want to go through the TB episode again, because that made it not fun any more.

Instead, I might concurrently build the bot alongside 2.007, the robot class (yes, I FINALLY GET TO BUILD ROBOTS FOR CLASS),or possibly extending it into summer. Summer plans as of this point are indeterminate, but chances are strong that I will be in Atlanta, exploring the engineering scene and annoying the neighborhood watch. This will of course all culminate in a hopefully kickass bot for Dragon*Con 2009.

So, I’m officially dropping the effort for now. But this means I’ll have to come up with something else to build while in Atlanta. Since going to a bot event far away just to bum around is lame, and could get me drafted to do event logistics, I’ll probably pound the dents out of either TB SP1 or the insect bots.

Test Bot isn’t all that banged up, just needing a new battery and lifter arrangement (in metal, please). NK needs detailing and a quick redo of the disc motor, which threw a magnet. And Pop Quiz, if I go for it, will actually be rebuilt completely. Why? 3/8″ thick bots are nice, but not very practical. Returning to a 1/2″ chassis allows me to use real drive motors and receivers. I have all the parts, so it’s just  a matter of carving a block of UHMW.

Time will tell. First, finals.

One week report!

Dec 09, 2008 in Project Build Reports, Project RazEr

So this project has pretty much moved past the “it works” stage onto the “it’s useful” stage. I think that might make it my first ever constructive build project… as opposed to, you know, destructive. Or just absurd.

Over the past week, I’ve been riding the scooter around to get to class and run errands. It attracts a fair share of stares and questions, since nobody really expects to see a Razor scooter cruising at 10mph. Yes, I mounted the real e-bike throttle onto the handlebars, so it’s actually controllable. No, I have not went back and fixed that terrifically beautiful hack of a signal interface.

I have also been pretty good at “path look-ahead” to avoid potentially lethal potholes, but have yet to attempt a flying leap over the railroad tracks.

The range from a full change has been confirmed to be over 3.5 miles. This is quite in line with the ~4 mile calculation, and seems to include all of the inefficiencies that I did not include in the approximation (rolling resistance, terrain variation, wind resistance, etc). This is perfect. Why? Because it lets me make several cross-campus trips per day – e.g. to get to class and back, to get food, and to get to MITERS and bumble for hours on end. Here’s an example.

Yeah, all the blue lines are on top of eachother. This consists of the following trips from today:

  1. To class in the morning (Ames St. @ Amherst to Mass Ave. @ Memorial Drive)
  2. From class to the Media Lab (Mass Ave @ Mem Drive to Ames St.)
  3. From the Media Lab to more class (Ames St to Mass Ave @ Mem Drive
  4. Class back to dorm, through the north building cluster (Mass Ave. to Ames St. via Stata Center)
  5. Dorm to Student Center for some dinner (Ames St. to Mass Ave)
  6. And back.
  7. Dorm to 7-Eleven (Ames St. to Main @ Portland)
  8. And back again.

That about covers my needs, really. 3.5 miles and back is also enough to make it to area hardware stores, which is of course a priority.

However, I have also been constantly maintaining the motor to keep it running. There’s nothing electrically wrong with it, but the design of the motor and wheel interface is that

1) it tends to force the motor apart, since the inside of the wheel has a chamfer on both sides and I have a matching outward chamfer on the motor endcaps, and

2) the wheel itself tries to torque the endcap tie screws out. How the hell does that happen? I had to cut indents into the tire’s plastic rim to pass those tie screws. They are circular in profile, and my guess is that with every compression cycle of the tire as it rolls on the ground, the indent sort of cams up against the screw and torques out a bit.What then happens is the screw sticks out far enough to bang against the aluminum frame, making a very audible click that tells me to stop immediately and ride unpowered the rest of the way.

This phenomenon has even defeated red Loctite.

I’m working on a design revision of the outer can that doesn’t rely on those screws to keep everything together. Either they will be routed internally, or the can itself will have some other method of fastening things while still allowing the tire to be changed. For now, the stator won’t change, since I can’t find another giant copier to rip another stator out of.

Oh, and I have also let all the Li cells go horribly out of balance. I haven’t made a charge plug for the DB9  charge/balance port – rather I have just been attaching alligator clips from my charger to wire leads shoved into the main + and – pins. Safe, eh?But now I have discovered that the cells have up to a quarter volt disparity, which is too much to let the whole pack charge at once. I’ll get around to making the balancer plug and cycling the cells appropriately.

The bottom line is that IT WORKS and IT’S AWESOME.