Archive for July, 2008

 

Ãœberclocker Update 13: You Mean Solid Objects Can’t Intersect IRL like they do in Inventor? Edition

Jul 29, 2008 in Bots, Project Build Reports, Überclocker

It’s slowly coming together.

Actually, there is exactly one more part in the mechanical assembly of the robot to make… in theory. After that, it’s all electronics work (and maybe some minor mods here and there). However, that one part is going to be a ton of trouble.

Anyways, pics.

Both ÃœGFBs in place. I’m missing 4-40 cap screws to assemble them, so I (go figure) need to run to the hardware store again, or just order a box from McMaster and hold my peace forever (until I need a longer one).

Each ÃœGFB contains 3 drill gearbox stages for a total of 216:1 per gearbox. The 3:1 chain final reduction takes it to 648:1 from motor to output.

This assembly will not take its own stall torque – I established that long ago and decided to just go with it.

Alright, so, here’s the last major fabbed part on Ãœberclocker – the gearbox main shaft. This was one of the parts that I designed at 5AM while hopped up on Jolt and never quite looked back until I had to make it – which is when I realized it was a HORRIBLE idea.

But it was an excuse to explore the threadcutting functions of the lathe. In order to mechanically decouple the fr0k spr0cket from the gearboxes, I decided to use a giant nut on the shaft along with washers and disc springs to set the “clutch force” needed. This required the cutting of a 3/4-16 thread into the steel rod. I had no 3/4″ die and probably couldn’t crank one even if I had it.

So, in a leap of faith, I read a quick webpage or two about threadcutting (“So that’s what the little dial thing is for…”) and went for it. Here’s the thread in the process of being carved into the steel shaft.

Well, the finish is horrible, the threadform isn’t exactly triangular, and there was a bit of “accidental overshoot”, but the nut fits. Not bad for a first shot.

The main shaft slipped in place…

…with a fr0kspr0cket and retaining nut.

About now is where I realized that the system had to be re-engineered. I didn’t “design” any sort of power transmission mechanism from the the gearbox shafts to the main shaft. So in a moment of brilliance, I tried to wing it with set screws.

However, the 3/4″ OD of the shaft along with the .472″ bore meant that there was a hair over 1/8″ of thread in the set screw holes, which is bad even if I flatted the drill gearbox shafts. Additionally, I had no space for washers or disc springs next to the nut, since it would cover the set screw hole if it moved any further towards the sides.

So I pretty much designed myself into a corner here. Fortunately, 2.5 weeks remain to re-engineer this section of the bot.

A quick check to see if everything lines up.. indeed it does. So maybe I was actually awake for this part of the modeling?

After my disc springs and random hardware arrived from McMaster, I needed to put together the clamp arm pivot block. This required pre-loading the disc spring stack a bit in order to cram the retaining ring onto the leadscrew nut assembly.

Unfortunately, I can’t do this AND wield a retaining ring plier at the same time. So this Somewhat Innovative Solutionâ„¢ was devised – push the springs down by clamping the pivot block’s edges and tightening the clamps.

The complete clamp arm actuator, with the motor mounted and previously interfering shoulder screw counterbored.

The corner I designed myself into has, as any good corner should, three sides – two ÃœGFB gearbox shafts and the leadscrew from the clamp actuator.

I’m not exactly sure what I was thinking at 5AM when I whipped this together, but it was probably not very much nor very coherent. At the clamp’s maximum travel, the leadscrew interferes with the diameter of the gearbox shafts almost to the 3/8″ thread.

This is bad – 3/8″ isn’t exactly very beefy, especially not when it’s hollow and loaded with stress-rising threads. I could remove the little retaining screw from the actuator, but that gains barely an eighth inch of diameter (and also risks running the clamp arm right off the end of the leadscrew, which is bad)

The best solution would be to just shift the motor mounting holes back an eighth inch or two… or move the fr0k pivot shaft a bit forward. However, that’s a nutty amount of re-engineering and rebuilding either way – I might have to dig up more 1/2″ aluminum and recut these pieces on the waterjet if tricks on the mill don’t work.

This is a very cool-looking corner. I think I’ll stay and stare at it some more. Preferably during the day when I have a clear head and no caffeine in the system.

So with little else to do until I had a solid design, it was time for Pretend-O-Bot!

This is essentially what the final bot will look like. Yes, this is why I love engineering.

Folded down in the convenient Stow-And-Go position. Speaking of that, I should figure out how on earth I’m going to get this enormous bot down to Atlanta before it comes to the night before departure!

Ãœberclocker Update 12: I Need to Go to the Hardware Store AGAIN?! Edition

Jul 27, 2008 in Bots, Project Build Reports, Überclocker

Did you know that I only have 2.5 weeks to finish Ãœberclocker? I technically do have a whole month until Dragon*Con, but only 2.5 weeks remaining around 24 hours of machine shop access and piles of potential parts. Back in the dayâ„¢, I’d have to plan for weeks to take advantage of a few hours of machine work at most.

I’ve been totally spoiled by machinery. Hey Dale, I think I’ll just camp out in your shop space when I get back.

Anyways, work is shifting towards completing the fr0k assembly. The frame and running gear, at this point, is complete, since I recut the bottom plates a few days ago. One of the fr0k gearboxes (the so-called “überghettofrakenb0xen”) is complete, and the other is coming. The fr0k actuator awaits springs from McMaster. After that, it’s all electrical work.

And Pop Quiz needs a blade. Like, now. But, first, Ãœberclocker.

I lied when I said the frame was complete. I got these Mabuchi 700-frame motors from Banebots and plan to swap them into the drive gearboxes when it comes time. Although this is not a priority, I would like stronger drive motors for Ãœberclocker if I can manage it.

Two 500-size drill motors, while punchy for a 12lber, is cutting it close for a 30lb ‘bot which needs its drivetrain as part of its weapon. Especially carrying another opponent – a gross weight of 60 pounds – the drive will be heavily stressed.

The motor swap should be relatively simple. Some drill gearboxes (not sure about these yet) already have the mounting pattern for the larger motors built in, as some 18 volt drills do have these larger-sized motors. The 15 tooth pinion on the 900RPM gearboxes have plenty of beef to bore to 5mm.

If not, I’ll just modify the motor mounting plate. I would get a few of those drills from Northern Tool to play with, but the two-stage gearboxes require a redesign of the back half of the bot. Not smart when the back half of the bot is already made.

Top and bottom plates. The top is from several weeks ago (the only part out of like 10 that the jet didn’t eat), and the bottom two are new. I made the rear bevel armor out of some .075″ steel for added protection and mass (though it’s still under a pound).

The steel is also dichromate plated, which gives it the weird rainbow effect. It doesn’t speak much to Ãœberclocker’s masculinity to run around with a giant rainbow-colored ass, but makes it all the more funny when it defeats opponents.

The garolite holes have minimal delamination. I found that the best way to avoid it is to just punch the hole and cut as fast as possible, essentially completing the hole before the high pressure water has time to force the laminations apart except in a very localized area.

And here they are mounted. Not all the holes on the beveled end line up (due to sketchy geometric projection and even sketchier edgefinding), but UHMW is pliable enough and the error small enough that I just let the countersunk screws align themselves.

Freshly gutted drill gearboxes lined up like freshly butchered meats.

It takes three drills with metal gearsets to make two überghettofrakenb0xen, since each one uses three planetary stages. I needed to swap the 15 tooth pinion on the 18v motors for the 9 tooth pinion on the not-18v motors.

Making the aluminum gear housing for the ÃœGFBs. To do this, I made some 2″ diameter, 1.325″ long cored rounds on the lathe, then finished on the mill (since there isn’t a proper boring setup for the lathe). The boring head is my new favorite tool.

I machined a light flat onto the casing before starting – this is so a plain milling vise with no V-channels can grip the round part effectively. Also it was to ensure I can remount it in the correct orientation.

Sectioning off a drill ring gear to use as the first stage. Being made of sintered steel, it absorbed alot of oil from the grease in the gearboxes. Result: dense grey smoke cloud, but the self-lubricating is nice.

And the ring gears installed. I (not purposely) used a massive, massive press fit for these parts. The drill ring gears are 1.495″ in diameter. The hole I bored was supposedly 1.490, but turned out to be 1.485… It took a very, VERY large torque bar in the vise and the grace of the Robot Gods to squeeze these rings in.

There’s a tiny bit of offset in the teeth between the half-ring and the intact ring, but that doesn’t affect anything, since the first stage fits fully in the confines of the half-ring.

I anticipated having to use set screws or something to keep the ring gears from rotating under high-torque loads, but if they let loose now, something has gone horribly wrong. Although this giant press fit was accidental, I’ll probably reproduce it for the other ÃœGFB

In a moment of genius (or perhaps insanity), I discovered that the chain breaker tool acted as a nice gear puller for the drill motors. It worked on the 15 tooth pinions fine, but did not on the 9 tooth ones.

Two gigantic flathead screwdrivers came to the rescue for that.

In another moment of insanity, I discovered that sticking a magnet on the end of a ratcheting 1/4″ box wrench made a very low profile right-angle driver that takes any 1/4″ hex screwdriver bit.

I’ll probably permanently epoxy the magnet onto said box wrench later on and press this into service as my “I-suck-at-designing-serviceable-machines” tool.

The basic form of the ÃœGFBs. These are not independent assemblies – they are designed to mate permanently to the fr0k main support towers.

A serendipitous side effect of cramming the drill ring gears into a space that was .6% too small for them was actually increasing the precision of the gearboxes somewhat. I noticed that I had to align the gears more before they’d drop into their pins, and the whole thing has less backlash and “wiggle room”. Presumably the slightly too-small ring gears cause more tooth contact – this is a good thing.

One small issue arose from the fact that I never fully measured the inside of a drill gearbox – only speculated on the length of parts based on the ring gear dimension. This was fine when I was making housings for stock two-stage gearboxes, but adding the third stage threw off my width calculations.

Result? Pilot ring on the motor mounting plate doesn’t go all the way into the ring gear housing. In fact, it barely goes in at all. The gearset sticks out so much that it’s better to leave the mounting plate flat and let the remaining space be “wiggle room”.

….so I turned the mounting plate around. I actually did make a new one, since the motor mounting holes had been counterbored halfway through. Now the ÃœGFB is a nice fit.

And I need more 4-40 cap screws, but the hardware store isn’t open this time of day.

Ãœberclocker Update 11: Photocentennial Edition

Jul 23, 2008 in Bots, Project Build Reports, Überclocker

I have taken over 100 pictures of the build. I take an excessive amount of buildpics.

This is actually not a bad thing. I have grown to like documenting my work as much as I can manage. You know, so in case I lose my memory for some reason, I can find out how to do it again.

Yes. Anyways…

More drills! I got some more 18 volt drills off mysterious, sketchy Yahoo Stores. Oddly enough, now that I’m actually looking for a 18v drill with the standard 36:1 gearbox, I can’t get one. Yes, these are the 900RPM type. Handy if I want to have a spare gearset for the drivetrain (which I do).

So, for the curious, the Great Neck “brand” of imported 18 volt drills also have 24:1, 900 RPM gearboxes. I’ve seen these at multiple retailers, like this.

Down to business. I couldn’t find any 4-40 cap screws, so I had to assemble the clamp actuator temporarily with little computer standoffs. Here it is mounted to its pivot by a shoulder screw and spacer (one of my new favorite building methods). The actuator is free to swing on the pivot point.

In yet another episode of “How the hell am I going to put this together?” I discover that screws do indeed have heads. This one contacts the sprocket, and will need to be counterbored a bit into the clamp arm in order to pass the chain later on.

Onto the actuator again. Here’s the beginning stages of the leadscrew assembly.

The 3/8-12 Acme screw has one end slightly turned down and bored to fit the 4mm shaft of the B62 motor. A 6-32 set screw drilled down from the screw surface holds onto the shaft flat and transmits torque (You can see it barely sticking up by the bushing).

This turned down section also runs in a short 5/16″ bushing. So, between the B62′s output bearings, the close 4mm shaft fit, and the outer bushing, the screw itself is pretty stiff.

The other end of the screw is also slightly turned down for… Well, I don’t know. It must have been for something.

With the leadscrew firmly stuck on the motor, it was time to work on the other end of things. The clamp hinge-pivot-clevis-trunnion-whatever is a multipart assembly consisting of the leadscrew nut, a nut holder (LOL U SED NUT), and the aluminum cutout of the pivot block. The nut holder is made of a 1″ round of steel.

Yes, I chucked an endmill and was using it to drill holes. How else can I get smooth, clean, flat-bottomed holes in a single shot?

So things got a little too hot during the turning process (by which I mean I was jumping back every few seconds because another smokingly hot oil-covered steel curly sliver would land on my arm). Using mad engineering skills, I make a convenient chip guard.

The lathe stepped up the game by firing chips under the guard. I bit it and just put on some welding gloves.

To keep the assembly together, I turn to my perennial nemesis, the retaining ring. I hate retaining rings with a passion – that’s why I use them, of course. To ease my suffering, I actually went and bought a set of retaining ring pliers with interchangeable heads.

Conveniently, a giant hacksaw blade measures in at the exact right width to cut a slot for this retaining ring. And so the Amputee’s Cutoff Tool was used to slot the nut holder (with the spindle running, duh).

A quick trip to the mill to add a hole and a slot and I have a nut holder (so I don’t have to hold my own nuts, of course). The hole is an on-the-fly part design change, since I figured a press fit was nice but not very serviceable.

Add one leadscrew nut. A set screw keeps the nut in place. Since this nut should only ever see a compressive load (lifting robots with the clamp arm is a bad idea), I don’t count on this being too much trouble. However, it’s equally less trouble to drill the set screw hole into the nut itself, so that might happen some time too.

The 100th Ãœberclocker build pic is of the nut assembly. Add to the nut holder assembly the pivot block and two Belleville discsprings, and it makes compliant clamp arm. When the arm clamps down onto an opponent, the motor will continue to drive the nut assembly (since it floats in the pivot block, held in by the retaining ring), compressing the springs and adding a bit of “preload” to the clamp arm. This means the entire system doesn’t have to be wound up in order to clamp firmly.

It adds a bit of compliance to the system, but is not designed to save the assembly if the opponent decides to force its way out of Ãœberclocker’s grip. The next failure point down the line is the end of the clamp arm itself.

Unfortunately, I couldn’t actually test this part, since I redesigned the nut holder on the fly. Originally, a bronze bushing separated the pivot block from the sliding nut assembly, and the neck of the nut holder was 1/2″ – which is what I ordered springs for.

However, in a fit of laziness and after discovering I didn’t order these bushings, I just decided to make the neck a bit wider and run it directly in the aluminum. It’s moving all of 0.1″ maximum – does it really need bushings? There are things in this world that don’t need bearings since they won’t last long enough, move fast enough, or need enough precision to warrant any (Ãœberclocker’s clamp arm is all 3).

Of course I forget the springs no longer fit – time to get different springs.

Shoved onto the leadscrew assembly, the (almost) complete clamp actuator. I don’t have the little metric screws to mount the B62, nor 4-40 socket head cap screws to attach the actuator body.

However, it’s fun to just putz the thing up and down the leadscrew. I didn’t get the “precision” Acme nut and screws, but it’s still very smooth, and can push with an absurd amount of force – theoretically over 200 pounds (translating to about 30 at the end of the clamp). Dunking it in some EP grease should make it even better.

While I was waiting on some people to finish welding in the machine area, I finished out the drivetrain on both sides by mounting the motors and adding belt tensioners. To dismantle the drivetrain easily, I only have to release the large tensioning roller, freeing up enough teeth to slide the wheels off.

There’s enough tension in the belts to not slip on the motor pulley under constant (hold-wheel-down-on-table) load, but only a drive test will reveal the true performance.

It’s ALMOST THERE!

Page updates and whatnot.

Jul 22, 2008 in Stuff

So I realized that the Projects page was getting horrifically out of date. I’ve updated the page with new pics and additional links to the newly created Ãœberclocker and LOLriokart project pages.

I’ll round out the updates with Pop Quiz 2.0 later.

Ãœberclocker Update 10: How the Hell am I going to put this together Edition

Jul 21, 2008 in Bots, Project Build Reports, Überclocker

There’s a very good reason why most real engineering groups and companies have multiple designers and some times even dedicated review committees – so one dude doesn’t make half of whatever is being engineered at 4am and absentmindedly forget to account for how it will attach to the other half.

I’m starting to run into little episodes of “Wait, how am I supposed to assemble this?” when putting finished parts together. Barely-accessible screws , questionable attachment methods, and mismatched hardware and holes to name a few…

Good CAD programs (like CATIA) actually have tool clearance detection and machining simulation (helping also to avoid “Wait, how the hell am I supposed to make this?” syndrome). Having messed with CATIA some, I will only say that its user interface designer needs to be brutally beaten with a monocrystalline turbine blade.

Recapping the weekend of work…

The main pivot shaft for the fork assembly. Made from a 3/4″ diameter round of aluminum that was 800-grit-sandpapered to just under .750″ to fit through the bushings with some wiggle room, then milled appropriately.

I chose to go with a live shaft over my usual preference for dead shafts since I wanted both fork arms to take the stress of lifting an opponent. Whether or not this decision will haunt me later – like the bot sagging in the middle just enough to jam the live shaft in bushings – I will have to see.

Taking .015″ off the 0.515″ thick raw waterjet-cut pieces to make them (essentially) .5″ thick. I could live with a thousandth or two like most half inch plate stock, but not what amounts to 1/16″ of extra width over the entire fork assembly.

Before starting, I gave the mill head a quick tramming (squaring), since I had used it to cut angles. Having a mill head “out of tram” or not square with the table will cause all your parts to become out of square. Successive, offset passes will have a sawtooth-like texture and won’t be smooth.

I did a rather hasty job, since I didn’t really care that much , so there is still a very, very light unevenness in the surface as the picture shows.

Planing down the fork arms was a little more interesting. These things are about 16 inches long and rather difficult to grab with a vise. I could machine one long section at a time, but there was an Awkward Middle Zoneâ„¢ where the two straight lengths met which would always hang off the edge of the vise. And thus there was chatter.

The second one was better, since I figured out a better way to stuff it in the vise, but a small Awkward Middle Zoneâ„¢ still remained.

Yes, proper procedure is to mount it straight to the table with clamps, but that’s too much effort. It’s really shiny, however!

Putting the Giant Set Screw hole into the top of the fork arms. I’m not going to show you how I actually did it, of course, and instead show a pre-process picture involving an edge finder.

So being satisfied with the arm work for the night, I turned my attention to assembling the drivetrain in its final configuration. It does seem like I will need a low-side tensioner, since with some running-in the belts began to stretch and started skipping on the motor pulley.

This is bad, so I’ve orderd some more little random rollers and bushings from McMaster to mount on the front. Testing showed that this “low-side” tensioner adds just enough tension to stop the belt from skipping.

Test-assembling the fr0k base structure. This was the first episode of “How the hell am I going to put this together?” – I won’t be able to reach the countersunk screws on the inside front without some sort of right angle ratchet or ball-ended wrench.

Mounting the assembled fr0k base to the frame. Hey, the interior is now totally enclosed!

Final assembly of the fr0k arms. It’s shiny – absurdly shiny, with both machining marks and the polished aluminum surface.

And so, Friday night’s Pretend-O-Bot is a second clampbot salute, only slightly less rigged. It’s looking almost finished…

What’s the extra hole on the end of the main shaft for? A potentiometer, of course, to keep track of the arm position. I’m not running any more open-loop appendages on the bots if I can help it. Eventually, I will settle one of several hundred trillion potentiometers that MITERS has around, and design a bracket for it to mount on.

Alright, back to it. Each arm motor has a single output bearing, so I broke out the boring head to make a pocket for it.

I love the boring head. Here’s a clean “Loctite Finish” on the bearing. A Loctite Finish is a very close-toleranced seat or hole which lets the object to be mounted slide in with only a light push. A drop of green Loctite will then retain the object forever.

In another episode of “How the Hell am I going to put this together?”, I realized that several holes on Ãœberclocker were spec’d for different screws than I had purchased, like the rotating hinge for the clamp arms. I needed a half inch long shoulder screw with a 1/4″-20 end thread for this one… but alas, none were to be found in the pile of random screws and bushings.

I have a feeling that Ãœberclocker is going to be a very difficult bot to service once assembled. Here’s hoping it NEVER breaks.

And the last piece for Sunday night is the beginnings of the clamp actuator. The waterjet-cut raw pieces were drilled, milled, and threaded to final spec. Go figure, I’m missing the proper spacer to actually mount it.

No Pretend-O-Bot for Sunday night, since… well, it looks the same. This week is the LAST THREE WEEKS(!) before I leave for Atlanta, and I only have three more full weekends to make this thing work. Where did all the time go? Why does class start in 6 weeks?

Bot on!!?