Project LandBearShark

This is what Boston looks like 6 months out of the year.


Seriously! I swear! We don’t even get those cool growth lights like Norwegians do!

A climate like this is one in which none of my less-than-noble steeds – RazEr, melon-scooter, or even the venerable LOLrioKart, can operate. Trust me on this – I tried getting to MITERS with melon-scooter a few times when the snow wasn’t ridiculous and it always ended with me having to push the thing after a few hundred feet because the rear tire turns into a solid cylinder of snow. And diving through the salty slush-covered roads is just asking for it.

I’ve been spoiled by having a fleet of small, overpowered personal electric vehicles for a long time now. I’ve become so reliant on their convenience that convincing me to walk some place is often an exercise that takes more time than if I had just gotten off my ass and walked. So having my entire collection be… hallwayed (since grounded doesn’t really apply here) because of such an obstacle as snow is just plain unacceptable. After all, other people seem to deal with it day-to-day just fine. But since I’m both conceited and spoiled at the same time (right?), it’s an excuse to open yet another project build; what, after Segfault and Make-a-Bot being fired off in rapid succession from their idle states for the past few months, I either have to go back to working on RazEr rEVolution (eww, another scooter), or finish Melontroller (AAAHHH SOFTWARE). At least this would give me something mechanical to crank on, and might actually prove useful in the end.

So it’s now time for me to explain this picture. If you pay close attention to my site, you might have noticed it at the end of the last Melontroller post. If not, go back and look. It’s definitely there – it’s kind of like the guy in a monkey suit running into your lecture while you take a surprise mini-quiz.


The bogie frame looks kind of like a banana, I guess.

So… I’m building a tracked vehicle, right? Nothing that exciting. Knowing me, it’ll probably be powered by an Etek and be based off some large, borderline absurd vehicle frame. It’ll probably have garish lighting and a 4 channel sound system. Hey, why not just put LOLrioKart on tracks? At least make it a half-track, at which point it should power through the snow just fine. I could probably rent it out to MIT to clear the inner roads around here too.

The tracks are from a junked Craftsman industrial snowblower, but similar ones seem to be found on all sorts of large snowblowers. These are the ones generally sold to people with long driveways or large yards, not the kind you attach to a front-end loader or small Bobcat style loader. They’re far from being even snowmobile sized. So really, LOLrioKart would be too large for these tracks anyway unless I were to actually build it as a halftrack.

But that’s kind of lame. I like LOLrioKart the way it is, and this is intended to be a new build after all. Which is why I now have to explain this picture:

Alright, so there really is no explanation. I’ll be honest: this got started almost exactly like how LOLrioKart got started two years ago – by having heaps of parts close to eachother, I catalyze their reaction into vehicles which should not exist. The tracks from the parted-out snowblower were sitting on top of the MITERS Public Skateboard temporarily when I happened to look in their direction while holding a melon.

So that makes me some kind of engineering enzyme?

landbearshark

The name “LandBearShark” came about after a long period of wanking about what you could possibly name such an abomination. Many ideas were pitched, mostly bad puns stemming from the DTV Shredder. Finally, someone suggested the pithily-simple name “land shark”. Nice and functional, but a bit overused and outdated. Plus, everyone knows that the hot topic in today’s marine biology scene is bearsharks.


What a bearshark might look like

The only thing more fearsome than the combination of a landshark and a bearshark is, of course, a landbearshark. Possibly made worse if ridden by a manbearpig.

technical

LBS is designed as a stand-on vehicle, unlike the DTV, which has handlebar controls. The skateboard is essentially a regular length board – not a longboard or mountainboard, so the ground contact length is actually not that large to react against you pulling on a handlebar. The idea is for it to be more freeform and rideable more like a real powered skateboard, such as the Exkate. But unlike the Exkate, I intend to not have a discrete hand-held controller at all. Instead, expanding on technology I first explored in Skatroller, I’ll have a double-handed control built into wrist guards that enable you to use your wrist force as tank steering inputs.

That’s right – hands-free operation obtainable with application of wrist bending force in either direction. On the skates, upwards force applies forward throttle and downward applies the motor braking, with center (no pressure, free hanging hands) as just “coast”. I anticipate doing much the same for this vehicle, except downpressure would instead reverse the drive it is linked to. With both hands, I have a full tank steering control arrangement.

A short summary of specifications:

  • 30″ overall length set by the skateboard, 26″ overall width set by the skateboard + two track widths
  • 20 to 25 miles per hour top speed
  • …provided by dual Turnigy 8085 “melon” motors. Since Melonscooter runs at about 25 miles per hour with a single motor, this will be more than enough horsepower to pound through any terrain.
  • 38 volt electrical system composed of 12S lithium nanophosphate cells.
  • Controlled by dual Melontrollers. This is good pressure on me to finish it and get it reliable so I can build another one.
    • A good reason to get the current-control loop on Melontroller running so I can avoid “Melonscooter launches” where surprised novice riders are instantly ejected by Melonscooter’s difficult to control starting torque.
  • Wireless fingers-free (because it technically DOES use your hand) throttle similar to the RazErBlades.
  • Tilt-and-leanable skateboard mounting
    • This will not control the vehicle in any way, but I foresee it being a “Operator is remaining on vehicle and has not been thrown off on launch” switch to prevent runaway flying tanks.

Now, let’s move onto the pretty CAD pictures.

Here’s attempt 1 at the design. Notice that the stock steel sides are already gone – I’m going to integrate the motor mount into aluminum waterjet-cut side plates instead, which will also form part of the frame. By the way, if you’re interested in your own melon-death vehicle, I have a .STEP file of the motor.

The shocks are some cheap mountain bike shocks I found one of in the Media Lab to measure up. In all actuality they’re probably too stiff to contribute any shock travel whatsoever, but they’re nice ornaments. I might also consider replacing the springs with something less stiff. They are found on Electricscooterparts.com under part SHK-634R.

This was about as far as attempt 1 got before I realized that there would be no way to connect the tracks to the chassis unless I turned this into a scientific exercise in suspension design. At the minimum, I was going to need something resembling A-arms, which would add alot of width to the vehicle. There’s also not that much space inside the tracks proper to put a proper bogie-wheel suspension, or even enough length for it. So I settled on a quasi rocker-bogie type configuration, which led to version 2:

In this version, the sides are split into two halves pivoting around a solid center axle (which is rigidly attached to the vehicle. The shocks mount to the ends of those halves, and therefore allow for essentially up and down travel of the track sprockets.  It’s still an arc motion, but the longitudinal displacements are small, and so the travel is linear enough. I’ll see small variations in belt tension for sure, but the massive gnarly track teeth will make it inconsequential.

I’ve also made a rough sketch of the body here. It will, once again, be T-nutted aluminum plate made on the waterjet.

Come on, I designed it. Do you expect it to be something more?

I’ve added more details in the region of the skateboard mount. The rings are high-deflection rubber shock mounts from McFaster-Carr. In this case, they are used as springs to keep the skateboard centered unless you lean. It is likely that the board will still be too “soft” even with those kinds of standoffs, so I might find myself paralleling them. Conventional rubber “sandwich” shock mounts are actually too stiff and do not allow for much travel. Optimally, the board should lean almost as much as the average longboard or skateboard.


How to tell what kind of monitor Charles uses: Inspect his CAD screenshots.

More details have now been added to the frame, including most of the slotting (but no T-nuts yet). I’ve made a front bush-bar-like structure that’s mostly decorative but serves to break the monotonous square corners that the frame otherwise has.

I added internal pattern cutouts to the large drive sprocket since it has to be custom-cut anyway. I’m not really concerned about weight on this vehicle – instead, the heavier IT is, the better. Maybe it’s a pathological thing I do since my standard technique of robot building involves beautifully (haphazardly) trussed (hollowed) waterjet-cut frame members.

The overall reduction from motor to drive is 4.5:1. If I elect to use a 9 tooth instead of 10 tooth sprocket on the motor, it’s 5:1. As on Melonscooter, I’ll take care of chain tension using one of the motor mounting standoffs.

The design as it stands right now. I finally manned up and figured out the “pattern along curve” option in Inventor, so I was able to model real tread teeth. Not that they are in any way representative of the actual snowblower treads…

I’ve modeled the two Melontrollers on the “control deck” as Eagle screenshots pasted onto flat squares. That’s essentially what a PCB is, right? I’ve also added the battery. It will be two packs of 6S, 6P A123 lithium nanophosphate cells. Essentially two 150% \M/etalpaxXx in series, for a grand total of 530 watt hours. What’s not shown is the third board that will contain the XBee radio with another Arduino that will manage signals coming from the two hand controllers and drive the motors accordingly. The rear cavity is to make space for supporting electrical components – contactors, power bus distribution, main battery switch (actually sticking out the back – it’s just an el-cheapo key switch), and possibly an onboard battery management device.

Now that Make-A-Bot is active and ready, I’ll be able to make slick cases for all the electronics and possibly even a real hard case for the batteries. There’s not a soda bottle in the world big enough to contain these packs for sure. No guarantees that any of it will be waterproof, but the big box frame will at least make it easy to seal around the bottom.

timeline

ASAP.

One important things that contributes to the rushed timeline for this thing include the fact that winter doesn’t last forever and fresh snow only falls every once in a while. The massive piles of snow already built up by the roadside and around campus are nice, but they get disgusting and hard after a few days. The best time for LBS to shine is out during a blizzard, or shortly after one.

Second of all, I won’t have all January to bash around MITERS. From the 11th of January to the 20th, I’ll be in SINGAPORE!!!!! at the new Singapore University of Technology and Design, which is a MIT collaboration to build a better technology & design university. They’re so new they don’t even have a campus yet. Those things are made to order and take a while. Anyway, I’ll be manning the deck helping with a student recruitment & general interest event. Details are yet to be settled, but I’ll probably bring over RazEr (rEVolution?) and a robot or something.

I’ve already ordered just about everything this project needs. Still, with the fresh snowstorm that just pimp slapped everyone in the New England region, parts might take a few more days to get here. After that, it’s going to be waterjet hell. In the mean time, I’ll work out how the DOUBLE SKATROLLER will be implemented.

Oh, one final thing. This is totally not being built to show Shane that I have a bigger deck.

How Many Make-A-Bot Puns Could A Make-A-Bot Make If A Make-A-Bot Could Make Make-A-Bot Puns?

About this many:

Above is the product of one night’s printing. Tons and tons of calibration cubes, then a few reruns of shapes I tried already to see if the quality improved on anything but a cube.

Let’s start with the first one, which I have nicknamed the Abortion Cube:

I tried the advice on the Makerbot Blog about one way to calculate a workable feed rate for the machine. From pulling a test extrusion, I found that the extruder actually worked at around 26mm/s. Unless I really missed something, that did not work out at all. It was clear from watching the epic blobular formation that the machine was in fact moving far too slowly. So I went for the logical next step: turn it up.

So here’s a test print of a 20mm cube at 40mm/s.

Muuuuch better. Now it’s on par with the stuff I see coming out of other peoples’ Makerbots. This was still on 0.4mm layers, though. I began playing with the “aspect ratio” of the filaments by adjusting the Width Over Thickness values and making more cubes. A wider thread adheres better to the one below it and can fill areas quicker, but appears to make the machine more prone to globbing.

After Cube #2, I decided that a stiffer platform was in order. Further lowering of the layer height was probably a futile effort if the platform itself has more “rock height” than the layers. With Advanced Circuits still working on my trippy heater PCB, I elected to make a stopover solution in the form of a 7 inch square chunk of 6mm acrylic. This represents the final outline of the PCB (yes, it’s massive), but the heated area is a 6×6 inch  patch in the middle.

Here’s another cube that turned out great. This represents the numbers I ended up settling on for the time being. The layer thickness is now 0.3mm, and the feed rate is 55mm per second.

I began logging all the settings used for each cube as I made them. From left to right, the lineup represents a gradual reduction in the layer height and increase in the feed rate while trying to maintain a constant thread aspect ratio.

While making the solid test cubes, I noticed the internal fill cooling unevenly with respect to the outside edge, causing “cube obesity”. To mitigate this, I adjusted the aspect ratio of the infill threads such that Skeinforge takes them to be wider (the 1.75 in “1.5 / 1.75”), and thus puts less infill paths down. This solved the bulging issue exhibited by the two middle short cubes.

The cubes were some times stopped after 20 layers or so because I figure by that time I got the point already and could move onto the next.

I sent MaB off on a mission to make another funky quarter sphere thing. The effects of the increased feed rate were primarily seen in the outside shells, where there was less “globbing” from the machine pausing briefly. I also went back to using the ‘straight criss-crossed lines’ method of infill, over squares or circles.

The first useful part MaB makes, however, is this spacing piece for a…. well, you’ll just have to wait and see. The part was designed to be 3/16″ (0.1875″) thick, and this print managed to hit 0.180″ consistently. Accounting for plastic shrinkage, that’s pretty close. The hole internal diameters came out slightly small, but there’s a setting to compensate for that in Skeinforge.

Here’s the family photo again. I remade the Companion Cube using the 0.3mm layer height, and oddly enough, it came out a little shorter than the first 0.4mm print. I also had a second go at making the funky triangular prism (for lack of more descriptive terminology…)

The finish at 0.3mm per layer on convex (external) surfaces and straight lines is pretty impressive. Concave surfaces, though, like the inside surface of the quarter spheres, are still a bit rough.

Overall, I’m content with how dialed in the machine is right now. The current setting is now my baseline for experimenting with new ones.

In the coming days, I hope to push MaB to the limits of sanity in terms of axis speed. I’m currently aiming to hit 70mm/s and 0.2mm per layer.