Totally Reproducible: The Quest for a second National Championship
by Robert Woodhead (Metal Munching Maniacs Producer)
After our magnificent (and somewhat surprising) success with Totally Offensive
in the 2004 Robot Fighting League season, James, Alex and I set our sights on repeating our National Championship.
Given that the frame of Totally Offensive was Totally Mangled in the championship fight, we knew we'd have to
build a new frame from scratch.
Since winning in 2004 gave us an automatic qualification for the 2005 Nationals, we decided that we could
spend the 2005 season trying out various configurations, in the hopes of dealing with some of the 2004 TO's
flaws, in particular the tendency to twist the front spar.
Also, I decided that this year, Alex (9) would do most of the driving, while James (12) would actually
do the building of at least one of the new frame(s) (based on my nefarious designs). This would give James an opportunity to learn
basic machining skills. I also arranged that the first build would be part of his big school science/technology
project, thus turning his hobby into homework. Am I evil or what?
Iteration 1: Totally Flexible
In this design, I attempted to deal with
the problem of the front spar twisting by moving to aluminum box section instead of channel, as well as making the spar shorter. I also
decided to connect the front spar to the back spar through a custom hinge, so that the two spars could
swing a bit relative to each other. The enlarged motor backplate also mounts some of the hinge elements,
and rubber standoffs (not yet installed) will keep the two components in proper alignment except during
big hits. Hopefully this will attenuate some of the impact loads and make the robot a little tougher.
The old batteries move from the top of the front spar into one side of the back spar, the
speed controllers go in the other side of the back spar, the radio and gyro go inside the box section,
and the contactor sits on top of the box section under a UHMW spring cover.
Except for some hand-grinding and chop-sawing, James did 100% of the work constructing this frame.
He measured everything, drilled and tapped all the holes, milled everything to exact proportions, and so on.
All Wrapped Up!
After building the frame and using as part of his school project (James got an 100, by the way!), we worked
together to install the electronics. As with the previous incarnation of Totally Offensive, we used curved sections
of UHMW plastic to protect the chewy goodness inside the robot -- our philosophy is that "air is the best armor", and our
opponents will have to get through the UHMW and the air-gap underneath it in order to damage the electronics.
Inside the rear spar are the drive motors, the two speed controllers, and the battery. The radio,
radio battery and gyro are tucked inside the front spar, and the weapon motor contactor sits atop it, protected by
a UHMW bonnet and the big self-righting ring. After powering up the robot, the rear flap is buttoned up with
some zip-ties, and we're ready to rumble.
We re-used the passive self-righting strategy from the previous incarnation of the robot. The big semi-circular
ring and the small UMHW finger at the front of the robot ensure that the only stable positions for the robot when on
a flat surface are either on its feet, or on its side with one drive wheel pressed against the floor. In the latter
case, which is unlikely because the rolling momentum of the robot will almost always put the robot upright, a small
pulse of throttle will right the robot.
The last photo shows the shock-absorption system. Thick slabs of rubber on both sides of the motor backplate
keep the backplate and the rear spar aligned, but have some give during impacts to stretch out the shock impulse
during the hit. Will it work? We'll soon see.
Next stop: Motorama in Harrisburg, PA, 2/19-2/20 2005.
Come see us destroy the opposition, or go down in a suitably spectacular manner!
Motorama Musings
At Motorama, Totally Offensive started out with two
wins, then lost two in a row to go down in what was generally agreed to be a suitably spectacular manner. One particularly
nasty hit by the famous drumbot
Helios bent a t-box shaft and shattered the bearing! The robot suffered from the same front spar twisting
problems that have always plagued it, leading us to the (somewhat expected) conclusion that bolting the frame to the top of the gearbox
is not going to be a viable design path. The box section worked a little better that C-channel, but not enough to make it
worthwhile pursuing. The hinge and shock-absorber idea did work fairly well, but given we are having to abandon the
front-spar-on-top design, I may not be able to reuse it.
To add to the fun, on the drive back, a dump-truck lost its brakes and rolled back into us, and we had to
use the mini-van as a chock to prevent it from rolling down a very big hill. A word of advice: if you're going to
get hit by a dump-truck, a 2mph collision is a good choice. Surprisingly, except for some scuffs on the bumper and
a couple of scratches on the hood, the car came through in good shape. I'm sure glad I brought an extra change
of underwear though...
In preparation for the next event, I'm actually building two completely different frames, one using traditional aluminum (see below)
and the other a radical carbon-fiber composite frame. As I'm sort of alternating back and forth between the two projects as the whim
strikes me, I'll separate the build report into two parts...
Iteration 2a: Totally Composite
Carbon fiber is incredibly stiff and lightweight,
so hopefully I'll be able to build a tough structure that will weigh less and, because it can be thin, can mount
to the bottom of the gearbox and still give us enough clearance to swing the blade.
After some research, I settled on DragonPlate carbon-fiber
sheets. DragonPlate is a very interesting material, a carbon-fiber composite sandwich with an aircraft-grade birch
plywood. This plywood is an extremely high-grade material, often used to construct aircraft props, and since (a) most of the
stress on a flat plate that is being bent is on the top and bottom skin, and (b) the plywood weighs less than carbon-fiber composite,
DragonPlate is stronger for the same weight than pure carbon-fiber. Anything using carbon-fiber is a bit pricy, but I've always
wanted to play with it, so this is a good excuse.
The plan is to use the flat plate as the main structure, and then bond carbon-fiber angle to it to reinforce the structure
and act as additional armor. Then I'll add some sort of top cover and one of TO's signature self-righting hoops.
DragonPlate turns out to be pretty easy to work with, it cuts and drills much like a softwood. The gotcha is that you
end up with a lot of carbon-fiber dust, and that dust and lungs don't mix well. So you need to take precautions, similar to
those you'd take when working with fiberglass; long sleeved shirt and pants, respirator mask, and eye protection.
I've found it really helps to wet down the area you are working on with a spray bottle and keep spraying it while drilling,
cutting or dremelling. Another tip: lay down some masking tape, and draw/scribe your measuring marks onto the tape, then work
right through it; carbon-fiber is hard to mark accurately and you don't want to scribe it as this could weaken it. You can also
use masking tape to mark out areas you want to sand before bonding (with special epoxy glue) two parts together. You need to sand
the top surface of DragonPlate because it is covered with a glossy topcoat.
Another tip: you can't tap carbon fiber, but if you're drilling a hole that will take a bolt that threads into something
else (like one of my gearboxes), and that hole doesn't quite end up where it should be, just run the right-sized tap through the
hole into the threads. It will chew out just enough carbon fiber to permit the bolt to be attached.
After spending
a few days adding some rails to
the perimeter of the robot (for stiffness and also to make it easy to mount the UHMW bonnets, it was time for final fit and
assembly. The radio and gyro went into two little rectangular areas I made out of scraps of carbon-fiber angle, with some velcro
for extra security and padding. I also added some angle at points around the motor and gearbox to help stop it from moving during
impacts. Underneath the robot are two long spars of angle that form a motor compartment and a blade guard. If need be I can add
another plate under the T-boxes to enclose them more completely. With all the electronics stuffed in, it looks pretty cool and all
the components are inset from the rails, so once the bonnets go on, they'll be well protected. The electronics wiring is a little
messy right now, but I'll clean it up once testing is complete.
A couple
of hours messing around with UHMW produced the bonnet and self-righting ring for the robot. The cute part was making the curved
sections that fit together and mount onto the front of the robot to stop it from having stable positions other than upright.
Due to the shape of the baseplate, this was a little harder to achieve than with Totally Offensive, and I am a little concerned that
a perfect "bad bounce" could put it in a stable position on its side that it would normally roll through. However, in this case,
I'll probably be able to rock it using the weapon motor.
Part of the fun of machining is coming up with ways of making the parts you need. I was able to take some scraps of UHMW
I had in my junk box and use the rotary table on the Sherline mill to make the curved sections. The challenge was in finding
a simple way to align and firmly mount the UHMW to the rotary table. I find the thought-process I go through when doing this
stuff is very similar to computer programming, something that surprised me greatly when I was just starting to mess around
in the hobby.
With that final bit of machining, Utterly Offensive was Totally Ready for
Battle Beach! See you in Florida April 2-3, 2005!
Iteration 2b: Totally Metallic
Having been forced to abandon the front-spar-on-top design, I had to come up with
a different design. After a lot of fooling around, I realized that if I made the front spar 4" channel instead of 3", then
the weapon motor could sit in it with the shaft projecting down through the base of the channel (similar to the composite
design).
The problem I faced was weight; 4" channel obviously weighs more than 3" channel, and Totally Offensive has always
been on the ragged edge of being overweight. So to make it work, I had to make the layout even more compact than ever!
I fooled around with many possible configurations, and even considered stacking the Victor speed controllers on top of each
other to save space.
Then I remembered that James and Alex had won a Vantec RDFR23
speed controller at Nationals. The Vantec was small, could easily power the drive motors, and best of all, was free!
After a little fooling around, a supremely compact layout emerged.
Nestled between the drive motors in a tiny 16" back spar is the battery. Since the back spar is 3" channel, the
custom batteries fit it perfectly. The front spar is only 13" long, and the speed controller, radio and weapon contactor
fit into a tiny 5" long bay.
The first step was connecting the front and back spars. While I'll eventually weld them together, connecting them
with screws helps get the fit right. To properly attach the two spars, I used a cute self-aligning trick. In the base of
the front spar, I drilled three holes. The outer two were the size used for tapping for 10-32 screws (.1590, #21), while
the central one was a .1960 (#9, close-fit for a 10-32 screw). Then I drilled and tapped a 10-32 hole in the exact center
of the back spar, at the same distance inset as the holes in the front spar.
Five minutes with the angle-grinder and a cutting disc converted the back 1.5" of the
front spar into a tab that slides under the back spar. Next, a single bolt through that larger central hole in the front
spar firmly attaches it to the back spar, and those two .1590 holes in the front become guides for drilling corresponding holes in
the back spar that precisely match. Drill and tap them, enlarge the two front spar holes to .1960, and the two spars are
firmly attached by 3 precisely aligned screws! Finding these cute and simple ways of doing things is one of the pleasures of the sport.
The weapon gearbox and motor mount was pretty simple. The only complicated part was that I had to cut a window
in the channel to accomodate the MagMotor endbell, which in this "upside down" configuration projects slightly further
down than the bottom of the gearbox. Some drilling, some sawing, and a little filing to fit was all that was required.
Then I drilled the motor mount holes in the back spar, did a little cutting with the angle-grinder so that the Whyachi
t-boxes would fit, drilled and tapped some 10-32 holes for later use in mounting the blessed UHMW armor, and dremelled
some slots for passing the pipe-clamps I use to clamp down on the motors through the frame.
Next, I took on the task of mounting the electronics into the tiny space
allotted. A complicating factor was that I needed to attach the contactor to one of the vertical rails of the channel,
and those rails have a 1:6 slope from the vertical. This meant that I needed to build a spacer with one side having
an identical 1:6 slope (about 9.5°). Fortunately, the tilting table that came with my Sherline mill made this
very easy. I milled the spacer to just the right width so that it causes the contactor to touch the case of the
Vantec (which mounts to the floor of the spar). Then I milled another spacer that fits between the Vantec and the
other vertical rail, so that the components support each other to resist sideways shocks, which will definitely
occur when TO hits something.
I then noticed that when the weapon gearbox was firmly mounted in place,
the main bearing was touching the frame. Since this was probably not a good thing, I popped the front spar into
the mill, used the hole saw outer casing to position the spindle dead center on the hole, and used the boring bar
tool to inset a 2" diameter ring 0.05" deep, giving the bearing plenty of clearance. Curiously, this didn't seem
to be a problem on the carbon-fiber frame (perhaps because it's slippery), but I'll have to check into that carefully.
Then I carefully wired up everything but the weapon motor (the
How Not to Blow Up Your Vantec
thread in the BattleBots forum turned out to be very useful!), crossed my fingers, applied the juice, and it all worked
the first time. Which of course leaves me deeply suspicious!
After getting the frame welded up, I made a few
tweaks. The first involved making the weapon gearbox and motor mount a bit more secure by inserting some wedges. I cut the rails off a chunk of
leftover alumnimum channel, trimmed them to size, spread on some epoxy, and dropped them down the sides of the front spar (with the
gearbox and motor in position).
Because I inserted them inverted, the slope of the rails of the front spar and the slope of the wedges cancelled out, and
the exposed sides of the wedges became nice and vertical. Once the epoxy set, I drilled and tapped some holes through the
rails of the spar and the wedges, and inserted some bolts for extra support. Now the gearbox fits very snugly between the
wedges, and any sideways forces are transmitted right into the rails. I also had a little backstop welded in to prevent
the motor from moving backwards and reduce the shear forces that might be put onto the gearbox mount bolts. I also drilled
and tapped some holes to mount the motor clamping strap securely to the underside of the front spar, ensuring that it
can't possibly touch the blade. Also, if you look closely at the shot of the underside of the robot, you'll see a modification
that Terry of Team Whyachi made for me; a 5/16" tapped hole in the gearbox shaft that lets me use a buttonhead cap screw as
the contact surface of the shaft. The idea here is to reduce the torque that causes the robot to tend to turn when the
blade is spun up. Interestingly, in a test spinup, instead of causing the robot to turn as it once did, now the robot
tends to want to move slowly backwards. Go figure!
My initial thought was to mount the radio battery on top of the main battery, but this turned out to mean I would need
a rear UHMW bonnet that was bigger than the one I'd made, so I went to Plan B and installed it next to the contactor. The
electronics compartment got a little crowded but with a little work, everything fit snugly. From front to back, there's
a Futaba GYA-360 gyro, the Team Delta switch that controls the contactor, the radio battery, the contactor, and the radio.
Behind them is the Vantec speed controller. Many zip-ties died to make this work.
I'm a little concerned about the terminal strip at the top of the Vantec; I'm worried that terminals will pop out
under big impact loads. I plan to build a little attachment that will mechanically hold the terminals in the right position.
The small square hole in the front bonnet is actually a finger-hole; it lets me support the big powerpole while I'm
plugging it into its mate to turn on the power. The small UHMW triangle on the front top of the gearbox is there to eliminate a stable
position and force the robot to self-right. As with previous versions of Totally Offensive, the robot will either self-right automatically
or get to a position where it can be self-righted using the drive motors.
The final result is the most compact and IMHO nicest-looking version of Totally Offensive yet. And it weight a mere 28.6lbs,
which means that if the design proves worthy, I'll be able to up the power to 30V and increase the weapon KE by 56%.
Battle Beach: Totally Self-Destructive
Well, we learned a lot at Battle Beach. Unfortunately, what we learned was
mostly “nice idea, but it doesn't work!”
Utterly Offensive, the carbon-fiber framed bot, worked great in its spinup tests in the bot. Then, as we were driving
with the blade spun up, it hit on of the floor seams and, as the saying goes, explosively disassembled itself. As far as we can
tell (and as some wiser heads on the forums predicted), despite the reinforcement, the frame flexed enough to permit the blade to scrape
it, and within a fraction of a second (several blade rotations), the robot tore itself into two large bits. As luck would have it,
we weren't videotaping the test, but it was fairly spectacular. No real damage to any of the electronics or components, but the
frame was shredded.
Meanwhile, Totally Offensive was up to its old tricks -- hitting hard and bending its own frame. Despite attempts to
untwist it between fights, as the tournament progressed the blade got more and more tilted, though still functional. In addition,
we found a new failure mode! In our second fight, which went the whole distances, we pushed the battery pack so hard that one
of the cells shorted out, and didn't recognize the problem. This had the result of reducing the pack's capacity, which cost us
the next fight when we ran out of juice halfway through. But it had a nastier side-effect; it caused the pack voltage to
increase, which stressed the power transistors in the Vantec controller. The Vantec held up during the next fight (with the
backup battery pack, which survived the disassembly of Utterly Offensive), but a few seconds after powering the robot for its
final fight, a series of loud bangs and huge clouds of toxic smoke announced the demise of our speed controller. So this version
of Totally Offensive got a Viking Funeral.
Iteration 3: Totally Titanic
With the demise of both versions of the robot, and the realization that we probably were not going to be able
to fix the twist problems of an aluminum frame, the search was on for a new approach. It occurred to me that the only material
I hadn't tried yet was Titanium.
Ah, Titanium! The strength of steel with only half the weight. Twice as dense as Aluminum, but over 3 times as strong,
and with one important property -- it likes to elastically deform and spring back into shape (as opposed to aluminum, which
tends to plastically deform). All that, and pretty white sparks when something hits it.
The downsides of Ti are (1) it's expensive, and (2) it can be a bit of a pain to machine. I knew I could ease the pain
in my pocketbook by getting some surplus Titanium from Titanium Joe, the bot-builder's
friend. And I knew, from a previous project, that the ever-dependable (and long-suffering, given the number of questions I
ask him per order!) Terry at Team Whyachi could shear and bend the Ti into the shapes I'd need.
After a week or so of kicking around designs with fellow Tarheel transplant Peter Smith
of Team Rolling Thunder, I came up with
a basic design for a Titanic Totally Offensive. It's similar to the previous versions, and incorporates a
a C-channel rear spar made of 0.100 Titanium plate. (A PDF of the design sketch is also available)
With the design finalized, I sent the titanium off to Terry.
For a mere $189, he was willing to shear all bits, drill about 80 holes that I could later tap, bend some long strips into
channel, and make a couple of custom motor-support studs. Before I sent it off, I measured and drilled the weapon gearbox mount holes
and shaft hole in the Ti that was destined to
become one of the mount plates. I did this because it was much easier to measure things while the plate of Ti was still rectangular,
and because it would permit me to create the other mount plates by using the first one as a template. While I was waiting for the
Ti to come back, I played around with repairing the
wayward magnets on my MagMotors.
A few days later, I got an email saying he was having some problems; the Titanium cracked when he tried to bend the first
length of channel. Apparently, some Ti plates can be a bit brittle in this way, it's the luck of the draw.
After a bit of back and forth, we decided he should try to bend the second strip after really heating it up, and also try creating
some channel out of some thicker 0.125 Ti, which Terry thought would bend more successfully.
A few days after that, everything arrived. The second strip had been turned into a perfect length of channel (except for some
minor bowing, which will not affect the design). The original cracked strip doesn't actually look too bad, and the cracks could probably
be TIG-welded to repair them. And also included was the thicker 0.125 channel, which, while very slightly oversized, looks perfectly usable.
Not only that, but Terry didn't charge me for the Ti stock he used to make the extra channel, which means he's either a total
mensch, or totally absent-minded -- but either way, I'm totally happy!
The first step was to use the template weapon mount as -- well, a template!
Clamping it on top of one of the blanks, I drilled out the 4 weapon mounting holes. There's a cute trick for getting
the drill bit in exactly the right place. Just plonk the drill bit down into the template hole and rotate the drill bit backwards
a few times to get everything lined up, then clamp down the whole kit-and-kaboodle and drill away!
When drilling Ti, there are a couple of things to keep in mind. The first is to use low speeds and lots of pressure (if you
are doing it right, you get these lovely long spirals of material). The second is you have to avoid getting the Ti too hot, or it will
work-harden and become almost impossible to machine. This means cutting fluid, and lots of it. A bit messy but the results are worth it.
My original plan was to cut out the shaft access hole with a hole saw. I'd done this on the template but the honest truth
is, a bimetal hole saw and Titanium aren't really a good combination. It sort of works, but it takes patience and the hole saw
will get blunted pretty quickly. By the time I got halfway through the second plate, I knew it was time for plan B.
After a little thought, I realized that with a little reconfiguration, I could just barely fit the weapon mount plates on the
Sherline mill's rotary table. First, I put the original template on, and cleaned it up a bit. I'd originally hand-ground a slight
indent around the shaft hole so that when it was tightly clamped against the gearbox, there was no chance that the main bearing
would rub against the plate. To make this look nicer and test that I could work with a weapon plate, I plonked it on the rotary
table, ran the conveniently-sized end mill down the central hole in the rotary table to center it, and then milled the indent.
Then I repeated the process with the second mount plate, using a drill bit this time to center the plate on the table and the table
on the mill. I also used the original mount plate as a spacer underneath the one I was working on. Then I milled out the shaft hole
0.01" at a time and added the indent.
The next step was to cut a rectangular hole in the bottom weapon mount plate.
This is needed because the endbell of the MagMotor projects below the bottom of the gearbox by about 1/8". This minor asymmetry
is one of my few quibbles with the TWM-3RS gearbox design.
I knew I could mill out the hole using the Sherline; the problem was how to mount the big plate to the Sherline securely so
that I could mill it. The solution became clear when I noticed that the width of the slot was identical to the distance between
the two mounting slots on the Sherline's table - 1.5 inches. So I marked out the slot and drilled holes in it in the corners,
such that the center-center distance of pairs of holes was 1.5 inches, then used buttonhead cap screws to bolt the big plate to
the little t-nuts that slide into the Sherline mounting slots. With the plate aligned and locked down, I then milled out the
perimeter of the hole (except for the bolts of course). Finally, I removed each bolt in turn, milled out that corner, and
replace the bolt, adding a washer, to keep the plate in position.
That done, I mounted the two weapon gearbox plates (a little filing required
for a snug fit), and added the 4 long studs that will help support the motor. I then started work on the first length of channel.
This would require 5 holes in the central part, one for the MagMotor's rear shaft, and 4 to allow the motor support studs to pass
through the channel. I first drilled out the central hole to a convenient size, and used that to clamp on one of my old Whyachi
MagMotor backplates, which I used as a... wait for it... template for drilling out the other holes. Due to the size of the channel,
I had to use a hole saw when expanding the central hole to make it big enough for the shaft, but it worked reasonably well, because
the diameter was considerably smaller than the weapon shaft hole, and thus the saw teeth were not moving as fast as they cut into
the Ti. Finally, I took a small plate of Ti and, using the Whyachi backplate, created a new backplate that bolts onto the studs and
locks the channel to the back of the MagMotor. It actually bends slightly and forms a spring.
The final bit of work in this weekend build session was to bolt the upper and
lower gearbox plates to the channel. You will recall that I had Terry drill some holes in the channel for this and other purposes.
However, the channel ended up having taller legs than I expected, so those holes couldn't be used; I'd have to drill some new ones.
I decided the best approach was to drill a line of holes in one of the mount plates, and use that as a template (have you noticed
the theme of this build report yet?) to drill holes in the other mount plate, and then use them as templates to drill holes in the
channel. One of the reasons I decided to do things this way was that the mount plates were not quite identical in size. Terry had gotten
them to within about 0.10 inch, which is pretty good, but if I tried to measure and drill the plates and channel separate from each other,
it'd be a nightmare. So instead, after drilling a line of holes in one of the plates, I could use the gearbox mount holes to clamp the
two plates together and drill holes in the second plate that were in exactly the same places relative to the gearbox
mount holes!
Then I could clamp the two plates onto the frame and use them as templates to drill holes in exactly the right plates. Finally,
at a later date, I could take the third motor mount plate blank that Terry had made for me, template it off the bottom motor mount
plate, and have a plate that could be used to replace either the top or bottom plate if the need arose.
Sounds like a plan. I drilled the line of holes, clamped the plates together at the weapon mount holes, and drilled
the matching holes in the second
plate. After I drilled a hole, I used a screw and nut to clamp the plates together even better, and drilled the next. Due to the
configuration of my drill-press, I had to move the clamping points around a bit from time to time so the plates would lie flat, but
this wasn't a big deal.
Then I remounted the plates onto the frame, and used them as templates to drill holes in the channel. Here the problem was that
I needed to drill holes of a smaller diameter (centered on the template hole) so that I could tap them to take a buttonhead cap screw.
The trick to do that was simple. First, I used the centering trick mentioned above with a drill bit that filled the template
hole, and drilled a little dimple that would serve as a starter hole. Then I swapped over to the smaller drillbit, and centered it
on the dimple! Every couple of holes, I stopped, tapped the holes I'd just completed, and bolted some buttonheads in, to make the
alignment even tighter.
With that complete, I turned to mounting the drive motors.
This simply required drilling a few holes, and milling a couple of slots for motor clamps. The only tricky part was figuring out
how to mount the channel to the Sherline mill properly; I ended up using the existing weapon motor stud passthrough holes as part
of the solution. Then a few minutes with the angle grinder and a cutoff wheel chopped the ends of the channel so the gearbox could
project through it.
Then I started playing around with component layout. My initial plan was to mount the speed controllers and weapon contactor
in the channel, and the battery and radio above the weapon, shieled by a UHMW bonnet. I knew I was going to have to play with
the battery mounting, so I started with the channel components, mounting and wiring them -- a job I've done so many times on
the various versions of TO that I can do it in my sleep!
It was at this point that Nick Martin of Team Overkill in Australia, who also runs a MagMotor-powered spinner named
Jolt,
came up with an absolutely brilliant idea. Instead of making a rectangular cutout in the bottom plate to permit the motor
endbell to properly fit (as I'd done to one plate already), he suggested that I simply mill a flat in the MagMotor endbell
so that the cutout was not necessary.
Ace, cobber! A quick check indicated that
Nick's idea was bottler (okay, no more aussie slang, I promise). Not only is is simpler and easier, but it also mechanically
prevents the motor endbell from rotating. I quickly endmilled a flat into a trashed enbell I had as a proof of concept, and
once I was sure it worked, I made better flats in my good endbells with a fly cutter.
Next, I templated my third frontplate and milled out the hole for the weapon shaft and clearance inset for the shaft
bearing. Now I have two frontplates with no slot for the motor, and a third that I can use in an emergency or if I need to
use a borrowed motor without the flat.
Assembling everything, it's starting to look like a robot. I made some mounts for the battery on top of the robot,
and clamped it down, mounting the radio (for now) just in front of it on a slip of UHMW. My initial plan is to build a UHMW
bonnet for it later. For now, I made the bonnets for the rear spar.
Down to the garage, a quick powerup, and... it drives like a dream. Nice and straight, no pulling to one side, turns
on the weapon axis. Lovely! I haven't run the weapon yet, because I really hate cranking it up outside of an arena, but I'll
do that next week (robot in center of garage, me outside of garage, inside minivan, wife at phone ready to call (a) auto mechanic
and (b) 911, hopefully in the right order).
Then it's off to the office and the UPS scale, which gives me a weight of... 27.2 lbs! Yahooo! Even with the weight of
the UHMW self-righting loop, I've got at least 2 lbs to play with.
On the drive back home, I figure out what I'm going to do. I've got some extra 3"x4" plates, the same size as the backplate
that the weapon studs bolt into. The distance between the top and bottom plates is 2.875", so i can easily mill down an edge of
some of the extra plates so they fit snugly inbetween; this would form a nice connection between the two plates, "boxing" them
in, as well as providing some extra armor for the MagMotor and, quite possibly, a nice place to tuck the radio.
The only problem is
, how do I attach plates together? Then I remembered this cute stuff that Peter of
Team Rolling Thunder sells called
nut strip. It's
rectangular aluminum rod with alternating horizontal and vertical threaded holes. Peter's nut strip isn't quite what
I want, but it's a great idea and I'm happy to steal it and make some of my own. 10 minutes later, I've placed an
order with Online Metals for the raw stock I'll need.
While that was en-route, I milled down the plates to the correct size, and made the self-righting hoop that
has become a constant element in the TO design. In order to get the robot to properly self-right, I had to make a little finger extension
for the front to provide a pivot-point. As with previous versions, there's another finger on the back to prevent the
robot from having a balance point when pointing straight up.
Then I started making a jig to streamline the manufacture of the nut-strip. The idea is that it lets me drill a hole,
then the previous hole acts as a positioner for the next one. There's also some 90-degree holes that let me keep the
two lines of holes properly aligned.
With the jig made, I then drilled the mounting holes in the top and bottom plates. The bar stock arrived on
schedule, and I started making nut-strip. Once installed into the frame, I used the nut-strip to template the locations
for the holes in the side-plates, and drilled them. It turned out that my plates didn't quite fit right; something in
the several assembly/disassembly cycles had shifted the plates a tiny bit. But they were close enough and everything
fit together.
Then I used some thin aluminum plate to make a little radio-box for all the electronics. This should shield it
from any electrical noise from the weapon motor.
The final bit of work involved the battery. One of my "brick" batteries was dead, and I was concerned there was
some internal damage on the other one. So I ordered up a new battery from Steve and Sarah Hill, the
BattlePacks people,
using some new NiMH cells that had 50% more capacity than my old ones. Because the form-factor was different and the
pack wasn't sealed like the old packs, I had to be concerned about properly supporting it, so I built a battery box/clamp
out of some aluminum tread-plate. This holds the battery securely on the top of the bot and gives it some armor.
Now all is in readiness for the New Orleans event!
Totally Titanic is Totally Triumphant
New Orleans was a great success! We've finally licked the
frame twist problem! TO's frame came through the event with no twisting and only a couple of minor scratches, and
dealt out some serious damage along the way. Our only loss en-route to a first place finish was to Killabyte,
a nasty Magmotor-powered full-body spinner, and we beat them in the losers-bracket final to earn the right to
face the infamous Morpheus. The frame of this tough brick could take the punishment, but his motor magnets
could not; TO cracked several of them, making Morpheus almost undriveable.
Almost all the driving was done by Alex, in preparation for defending our title at Nationals, where we are
sure to meet Morpheus again for a grudge match.
The only remaining problem with the robot is the MagMotor magnets; the repair I tried failed after several
fights (fortunately not costing us a victory!) and I had to swap it with my spare. I also need to support the
battery pack a bit better. So there's plenty of work to do over the rest of the summer.
While not helping his brother fight the robot, James spent most of his time running an in-box camera,
helping out Trey Roski of BattleBots fame, who was there supervising the event videography.
Preparing for Nationals
Despite
our success in New Orleans, there are several areas that could stand improvement. First of all, it was
clear that the robot needed a better battery holder to protect the batteries against shock impacts; in our
second-to-last fight, some of the cells of our battery were actually crushed, and lasted just long enough
to win. Only the loan of a battery from the Robotic Death Company
got us ready for the finals.
Since most of the weapon impact shock loads in TO are sideways -- recoil from the blade -- I decided that mounting
the batteries so that the long axis of the cells was oriented the same way made sense. After consulting with Steve Hill,
I ordered two 12volt packs, configured as a 4-3-3 hill-valley brick. When they arrived, I started building a sturdy battery
box to fit on top of the frame. The sides are aluminum, milled from 3" channel, and the front, top and back (not yet built)
will probably be carbon-fiber. The box holds the batteries snugly, with just enough room for foam padding (get it at Wal-Mart
in the crafts section). I'll complete this when the carbon-fiber arrives.
One neat trick I did when making the box was drill some alignment holes in the actual table of my mill, so I could use
dowel pins to easily align the parts for milling.
My other major project was toughening up the weapon motor. As you will recall, I've been having big problems with
the magnets popping loose and sliding around the inside of the motor can. After a lot of thought, I decided that the only
solution was to totally cage the magnets so they simply could not move.
The MagMotor is held together by two long bolts that slide between the magnets, and the exact orientation of the magnets
and the bolts determines the timing of the motor; if the bolts are up against a magnet, then the motor will be either forward-
or reverse-timed, and run faster in one direction than the other. If the bolt is right in the center of the gap, the motor will be neutrally
timed. For drive motors, you want neutral timing, but for a weapon motor like TO's that only runs in one direction, you can get
more performance by timing the motor in the direction it will be spinning (in TO's case, this is reverse timing, because this spins the
blade clockwise (viewed from the top), which in turn ensures the cap screw the blade shaft glides on won't unscrew).
What I noticed was that with the mount bolt up against the magnet, there's room for a second mount bolt in each gap that
touches the other magnet. So by adding 6 extra bolts, I could completely cage the magnets in one direction, and make the motor
sturdier as well!
Doing this involved using the Sherline rotary table. Mounting it on the mill, I centered the mill on the rotary table by
running a tight-fitting drill bit down the center hole. Then, mounting a motor can on the table and using a drill bit of the same
size as the bolt, I translated the mill out to the perimeter of the can, and was able to measure how much rotation was needed to
make the drill bit touch the other magnet (13°).
One cute bit of hackery I had to do in order to use the rotary table was mount my lathe chuck on it, so it could easily
hold circular objects in exact alignment. The lathe chuck screws right on, but there's a minor problem; during some milling operations,
it might start to unscrew. The solution was to use one of the tightening bars, a screw and a plastic spacer to keep the chuck
tightened down no matter what. This also required a thin washer on the shaft that screws into the rotary table in order to align
the chuck tightening hole with one of the slots in the rotary table.
Once this was done, it was a simple process to align my drill bit with one of the existing holes in the faceplate, and rotate
around the faceplate drilling the new mounting holes (I made them very slightly larger than the existing holes, about 0.2030, to give
a little wiggle room). I had to be careful not to drill into the chuck jaws in one case, but this was handled by drilling a starter hole,
then returning later, loosening the chuck's grip, and rotating everything a bit until a clear hole could be drilled.
Then on the other side of the faceplate, I had to mill inserts for the bolt heads.
Next, I had to drill and tap extra mounting holes in the endbell of the motor. This was done in exactly the same way as the
faceplate drilling. One catch is that one of the mounting holes goes right through the wire entry area of the endbell, so that hole
cannot be drilled as deep (and the hole right next to it just kisses the side of the wire entry hole).
The final step was to assemble the motor with 8 bolts. The easiest way to do this is partially screw in the 4 "main" bolts
that fit into the milled recesses in the faceplate, then get the other 4 bolts started, then tighten the main bolts, then tighten
the other bolts. I had intended to use a shorter bolt in the position that intersected with the wire entry area, but forgot; however,
this didn't seem to hurt, it appears that the bolt just pushed the wires aside. I'll fix this next time I assemble the motor.
With everything clamped down, I tested the motor -- success! It runs sweetly. Except that I got things reversed and drilled
the holes in such a way as to make the motor forward-timed! Oops! Well, it's not a big problem, I can drill some extra holes to
fix that, and thus have a motor that can be set for either timing. I'll also have to mill some inserts into the Whyachi gearbox
mount plate to accomodate the extra bolts that are sticking up, but that's an easy job.
Now that I've caged the magnets in one direction, the next step is to cage them front/back. To do this, I intend to drill and
tap some small holes in the actual can, so I can screw in bolts that will just touch the magnets at the top and bottom. But
that is a job for another day!
Update: It was a pretty easy job to drill the restraining bolt holes in the motor can.
Using the rotary table and the right-angle attachment, I mounted the can so I could position the drill bit properly. Then, using a
cute little centering tool, I centered the drill bit on the can and drilled away. Then I tapped the holes, and cleaned all the
steel filings out from inside the can (a bit of a pain given how powerful the magnets are. I laid down some epoxy fillets along the
top and bottom of the magnets, both to immobilize any remaining steel filings and to secure the bolts. Then, while the glue was
drying, I milled some recesses into the
Whyachi gearbox mount plate to accomodate the raised bolts on the motor. A little assembly and the completed motor and gearbox
was spinning away merrily!
Building a better Battery Box
Completing the battery box was a project that
took several days, mostly because of the time required to glue the carbon-fiber plates together; the adhesive used has a 24hr cure time. I
needed to bevel the edges of the aluminum side-walls (and one of the CF plates) to accomodate the curvature of the CF angles used to join
all the parts, so I used the tilting table on my mill; this required adding a few extra mount holes on the table itself.
It turned out that the fit was really tight but I had just enough clearance on the mill to do all the machining.
There are three separate CF components; the back plate that mounts to both the side-walls and the bottom Ti panel,
the top plate, and a front plate. In the gluing pictures, note the use of some blue masking tape to keep the glue away
from the aluminum -- a cute trick that I wish I'd figured out on the first part I glued...
The front plate isn't quite as long as I'd hoped, because I ran out of salvaged DragonPlate, so there's a small gap at
the bottom. My current story is that it's for cooling airflow... yeah, that's the ticket!
The various panels all interlink and, with the help of some Wal-Mart adhesive-backed foam, securely clamp the batteries
in place.
There's a new MagMotor in town, Bart!
Over the summer, rumors swirled that Satcon was going to make
some changes to the S28-150 “Shorty” MagMotor. While no Bothans died getting me the information, several of them got nasty infected cuts dumpster-diving,
and all they could tell me was that (a) RobotCombat was getting the new motors around Labor Day and (b) it had something to do with
set-screws.
Once the "new" mags were in, I ordered one from “Evil” Jim Smentowski, and promptly started taking it apart to
see what the changes were. The most obvious one was that the new shorty Mag has 4 bolts mounting the frontplate and endbell instead
of two. One of them is slightly shorter than the others; this one goes in the hole opposite the wire-entry area so as not to
press against the wires. Also, they are not magnetic (hmmm... how do you tell if they are stainless or something else?).
The other big change is two tiny set-screws (thank you, Bothan spies!), one hidden under the label. These fit into a groove
in the redesigned frontplate, and apparently their purpose is to make it harder for the frontplate to move. The question is -- will
this help with the moving magnet problem? Note: in the picture, the other frontplate is one I modified to handle 8 mounting bolts.
One other observation: the tolerances on the bearing that fits into the frontplate seem to be tighter than before; getting that bearing
on and off was much harder than on my other MagMotors.
Fear will keep the other robots in line... Fear of this battle robot!
Well, after a few more weeks of minor noodling and touchups,
TO is ready to go! The new battery box is totally built and installed, a spare set of batteries was obtained from BattlePack, and I managed
to install a real antenna and a Whyachi power switch that I had lying around. No more connecting powerpoles and zip-tying them in the box!
The robot ended up a couple of ounces overweight, which I fixed by putting some lightening holes in the self-righting hoop (a UniBit
step drill is just perfect for this; everyone should have one!) and sanding off all the paint on the blade (!).
TO still has its traditional weak spot -- the exposed wires from the battery box -- but quite frankly, the philosophy of the 'bot
has always been, "If you can get past the blade, you win!". Given their raised position, they aren't actually all that vulnerable, and I
may be able to slide on some PVC pipe as armor at the event; it all depends on the scale used to weigh the robots. TO is currently 29.95
lbs as measured on my UPS scale, and the traditional event scale is only accurate to 0.5 lbs (and rounds down!). This leads to the
classic builder question:
Q: What is the maximum weight of a 30lb robot?
A: 30.4999 lbs.
In any case, TO is packed and ready to defend its championship.
Good Things come in twos!
When all the dust had settled... when all the shrapnel had stopped ricocheting around the arena... Totally Offensive went
3-0 and repeated as RFL Featherweight National champion. Video of the fights, and pictures of most of the robots, can be
found here!
We are already hard at work improving the robot for 2006!
Current Build Cost Tally
The following table lists the component costs and sources for all new parts used in the construction of Totally Offensive 2005.
Needless to say, 95% of the robot's components are carried over from the previous incarnation.
| |
