Archive for February, 2011

 

Land-Bear-Shark: Beasting Everything Ever

Feb 27, 2011 in Land-Bear-Shark, Project Build Reports

At the start of today, LBS was a small pile on the ground whose satellite components were slowly being diffused into the MITERscape.

At the end of today, I’m proud to say that LBS has reached criticality (in this case defined as when a project can support its own weight and leftover parts). It’s not rideable yet, and it’s still missing a control system, but mechanically speaking it’s pretty much there. I would like to thank… uhhh, the waterjet, t-nuts, my parents, and… wait, what?

Anyways.

First order of business was to get the missing parts cut. Or, in the case of the front and rear skid plate, recut. I really have no clue where they ended up. But, I found a piece of plate aluminum that had their exact profiles missing, so I must have made them once before.

I’ll probably discover them in a few weeks and use it as an excuse to commission a second LBS.

The biscuit plates are the axle mounts that turned up short last time because I forgot what symmetry was.

Assembly now begins for real. I bolted in the shafts (which are stationary spindle pins) into their respective side plates, then mounted the resprung shocks. The large center 20mm axle was trimmed to length on an abrasive cutoff saw (the temperature decoloration is visible at the bottom there). The shocks themselves are secured using shoulder screws.

Slipping the tracks on was a matter of pulling things the right way. I had to loosen the front tensioner axle in order to slide the whole assembly on. Axial alignment of the plates was done by tightening down all the axle screws, which forced the plates to be 6 inches exactly apart and planarly aligned with eachother, then applying the shaft collars to the center pivot axle to keep them in place.

It turned out that “Bottomed out” was a good tension setting for the axles, so I left them there and secured the pins with 262 flavor Loctite.

Here’s a view from the other side, showing the sprocket spacers and the shaft collar arrangement. Most of the spacerwork was Dynamically Generated on the lathe using 1″ delrin stock.

After rinsing and repeating for the other side, accounting for chirality (oooops… let’s try that again):

Hey, this could totally be a kickass 60lb robot.

At this point, I took it for a leg-powered joyride down the hallway, almost running down some MIT Museum tourists in the process. For having tank treads, the thing glides pretty far. The treads were, of course, “set” into a favorite position from being parked in someone’s back yard for half a decade.

The rubber was dry and cracked in many places, so I began an emergency WD-40 marination treatment. The solvents in WD-40 penetrate the cracks and the rubber surface, softening it and infusing it with the oil component. A more legit method is making a tank tread and kerosene soup, but I’m not that hardcore (and had a can of WD).

While the treads were exfoliating, I started stripping down the skateboard. The very rusted truck attachment hardware took alot of effort and careful force application to remove. I also decided to try and repair the splintering front end of the board by filling the gaping holes with wood glue and epoxy. That’s what those clamps are holding together.

The new deck hangers (planters?)  are now attached. There’s a 0.400″ thick UHMW plastic block separating the bottom of the board from the hangers, since I needed to slip a bolted joint in that region.

Here’s what the whole system looks like from underneath. The rubber blocks let the board tilt a few degrees to emulate the action of a normal skateboard. It’s not perfect, but I didn’t particularly feel like engineering a linkage to use the existing trucks.

Note that the pivot pin attachment point is slotted. This is to enable the board to travel up and both about 1/8″ or so when the pivot bolt is installed. I decided that this interface would be a good one to put a “rider is still on vehicle” switch  on, since if this thing is going to be wireless, I absolutely do not want runaway conditions. A set of microswitches (which were not thought out well enough to have mounting holes designed) will detect the depression of the pivot in the slot, and interlock actions will be handled in the controller software. It’s as simple as cutting the throttle command or logic power to the motor drivers while opening up the main battery contactor.

As simple.

Watch that come back to haunt me.

Alright, with the board mounted and still drying, I turned my attention back to the motors. Previous rewound but never finished, these things needed wire pigtails and Hall sensors added. Both were easy enough affairs. The hall sensors are current attached using hot glue, which admittedly is not the best of materials for holding things in a motor.

But it was right there.

After more cutting, soldering, gluing, and some hammering too, the genetically modified melon motors get reassembled. I’m not sure how two motors in the same order came with totally different screws, but one was also way harder to put back together than the other. Quarity!

I Dynamically Generated these motor mounting standoffs using 1/2″ aluminum rod stock. The outermost holes in the cruciform motor mounting flange were threaded for 1/4-20 threads. This assembly is less meaty than the enormous billet mount that’s on melon-scooter, but the shorter overhang distance makes up for it.

One of the legs of the cruciform had to be cut off because it interfered with the rotation of the track’s drive sprocket. The nubby things protrude further, it seems, than what I designed for. That and the fact that the holes in the cruciform mount seemed to be hand-punched and were all over the place. It was fine in the CAD model.

And here it is. No sprocket on the motors yet, because I ran out of Beast It points for the day.

Overall, the rocker suspension worked great. The re-sprung shocks are just at the “right” spring rate; perhaps just a little soft, but definitely in the range where jumping up and down causes deformation and springing action, but just standing doesn’t. I wouldn’t mind taking a curbhop with this thing.

The skateboard hangers worked about as well as I expected, which was in fact not at all. Those rubber shock mounts are just too soft, and they deform almost immediately to flat when someone gets on. Maybe if I get ambitious I’ll engineer a torsion spring or leaf spring system, but for now, leaning action (roll axis) is poor but pitching and heaving (forward and backward tilt; up and down motion) are acceptable.

Left to do:

  • Sprockets?
  • Design the double wireless hand controller!
  • Make the double wireless hand controller!
  • Salvage the deprecated Kelly KBS motor drivers from the defunct Citycar model
  • Melontank?

Land-Bear-Shark Episode 4: Now that There’s No Snow on the Ground…

Feb 21, 2011 in Land-Bear-Shark, Project Build Reports

Poor Melontank.

Another victim of my “randomly open another bracket” style of ADHD-engineering, LBS has been sitting in a pile for a few weeks now thanks to me becoming acutely interested in failboats. With Make-A-Bot finally down for the count after baking its extruder controller, I now have the chance to pry myself away from staring at it and revisit LBS. Part of the reason for the slowdown is the lack of motivation due to the nearing end of winter. The past few days of warm weather has pretty much melted all the accumulated snow away, and what’s left is just hard solid frozen chunks of ice – not very fun. But, LBS can work on (ideally) any terrain, so I’m going to finish it off because I’ve committed too much funds to just ditch the project and repurpose the parts.

As I came to discover, I missed cutting a few parts during the last major waterjet run. Additionally, some of my hardware has gotten lost in the recent rerganization at MITERS, so that’s now going to lead to more delays as I either find it or give up and order more.

Here’s the pile of LBS, which was formerly under the workbench that is now halfway across the room. There’s been alot of reshuffling lately, to put it lightly.

I laid out all the flat parts that will form the new track side plates only to discover that I was missing four of those round 7-holed biscuit plates (for lack of a more descriptive scientific term). It turns out I may have only cut enough for one track module, judging from reviewing my tool files for the machine. Oops.

First order of business was assembling the “biscuit plates” onto the side plates, which involved tapping something like 56 holes.

What? Actually tapping a thread again? That’s definitely something I haven’t done in a while. But old is new, and spiral point taps in a cordless drill  make for easy beast-threading of 1/4″ aluminum. Add in the fragrance of aluminum Tap Magic (that Shane and Amy don’t like because they’re crazy) and this process went rather quickly.

Next, I decided to knock out the four large standoff-axles that bridge the side plates. I have no clue why I chose to go with 5/8″ steel for this. It’s so ridiculously oversized, and hard-anodized aluminum would have done just fine as a bearing surface for the large nylon track sprockets. Even worse, I’m not sure why I went with the case hardened shafts. I think it’s just because they were the cheapest on McMaster for whatever reason.

Either way, a normal drill bit wouldn’t touch this stuff until I broke through the casehard layer with a carbide endmill shoved into the tailstock.

The Old Mercedes looked great in the winter afternoon sunlight today, so I took the opportunity to snap a picture. The 1956 South Bend 10L has been with the club for almost 2 decades. Even though the tailstock is now a good .01″ out of level, all the headstock bearing shims have been bottomed out, and the cross slide has 2/3 of a turn of backlash, it stills hauls and kicks ass (for work close to the chuck – it also turns a natural .01″ per foot taper because of the almost .005″ of bed wear at the front). #dontmakethemliketheyuseto

A while back, I scrounged some interesting looking springs from a pile of crufted hardware. Turns out they were a great fit over the cheap mountain bike shocks I picked up for the bogie suspension. I actually think they might be too soft – since I can actually manage to compress them slightly with my hand – but I think they’re better than the 750 pound per inch “springs” that came with the stock units. For 750lb/in, I might as well just bolt the tracks to the body with rubber blocks. All four shocks were “re-sprung”.

Alright,  that’s pretty much all I can do for now. Both motors have been rewound, but haven’t been terminated yet. I also need to locate some 4.5″ long bolts to reassemble the drive sprocket with, and get the supplemental biscuit plates cut. It’s even more disorganized than last time!

micro-chuckranoplan: I’m Getting Closer.

Feb 19, 2011 in Chuckranoplan, Project Build Reports

Ever closer. I think the design of μ-Chuckranoplan 0002 is sound; of course, I say this with absolutely no qualifications or numbers to back it up. It’s visually engineered, as I now like to call it.

It will fly, but it’s still obese. Maybe if I pump it with helium…

Here’s the head of a Chuckranoplan after being stuffed with electronics. I elected to temporarily hot glue the two micro ducted fans onto a single carbon fiber stick pylon. The 3 cell lithium polymer battery ended up wedging very snugly into the nose cavity. The receiver and two controllers form the entrails that get shoved into the main body.

And here it is, all aseembled. The tail struts are all carbon fiber rods… well, except for one of them. Turns out that one was aluminum.  Whoops.

The wire access holes in the sides were dynamically generated (read: shoving a hot soldering tip through the wall).

The final weight of everything? 14 ounces as measured by a chintzy digital hook scale. I’ve successfully built a decent brushless-powered antweight. Maybe I should have entered it into the Motorama Robot Conflict competition or something.


Yup, still looks like some kind of stoned manta ray with a birth defect.

Here’s a front view, while it’s charging. Note the tiny 2-56 screws holding the head on…

So how did it perform? Better than attempt #1 for sure, but still a ways to ago, almost all on the weight reduction side. The “flight” was sort of a very strained and awkward hover, and it actually couldn’t budge itself from a standstill. However, with full throttle on the fans and a push, the model did flare down the hallway. At least, until it ran into the side of the wall from unbalanced reaction torques from the fans. They should really be spinning in opposite directions.

I’m definitely approaching the problem like a mechanical engineer, and the laws of flight don’t really scale down to things this small (something about area vs. volume or some gibberish). Overall, it’s too well-built.

The double-walled PLA prints, while “light” in the hand, all add up when the pieces are put together. I’ve been trying to single-wall some test pieces to not much success, as the PLA just sort of sags and falls over. Before I managed to successfully execute a single-wall ABS print, I accidentally half-baked the extruder controller on MaB, so that needs to be repaired first (The test piece’s first few layers looked good, however). I’ll  give single-wall printing a fair shot before manning up and just learning how to do it The Real Way with covered frame members, or hot-wired foam. The project has caught the attention of some Course XVI majors I know, who know how to do it The Real Way and may be able to get me some time on a CNC hot wire machine.

Micro-Chuckranoplan: I Guess It’s More Modeling Practice

Feb 15, 2011 in Chuckranoplan, Make-a-Bot, Project Build Reports

What happens when something I build doesn’t work? Generally, I go back and rebuild it. It might take a while, part of the time being spent recovering from the emotional scarring and psychological trauma that comes with things not working. In the case of the major vehicle projects LOLrioKart and Segfault, it took several tries and the better part of a year. But what’s nice about 3d printing is that you can just whip out another iteration pretty much on demand – at least when you’re not running desperately low on material. Come on, Makerbot, restock your big ABS and PLA reels!!

This whole Chuckranoplan thing only started like… last week. As soon as I realized that μ-Chuckranoplan 0001 wasn’t going to work out, I started designing the next version, so to speak. By “designing”, I really mean staring at more pictures of delta-wing craft on Google Images and wondering (besides why the hell they look so derpy) what they have in common. Tangentially related, Chuckranoplan now has its own page, and several other parts of the site have been updated also.

Here’s the designing and (now in process) build of mCRP 0002.

And we begin. This version started out much like the last, except now I had a much better grasp of what the Loft feature does. I was also directed to a database of downloadable airfoil content, so I could even use a REAL AIRFOIL! this time. For the new main wing, I selected the Clark YH section for its flat bottom with mild reflex, and for the tailplane (still unmodeled above), I stuck with a cambered NACA 6412. The issue is that the airfoils there are still not parametric, so I was stuck with hitting the Scale button to get the chords I needed. That’s fine – once I have a need to actually tune the airfoil to my liking, I don’t think I’ll be using a premade database of them.

I decided to ditch the 3d printed vertical stabilizer tail and instead go for a triangular loop of sorts with the verticals made of carbon fiber rods to save weight, since this is probably going to weigh 8,000 pounds as well. The CF rods will be wrapped in something light and filmy to replace the former vertical stabilizer section.

Alright, we’re getting somewhere now. It’s starting to look like some kind of strange alien bug with four eyes.

The nose was more “scientifically designed” this time, and by that I mean I actually had a fixed order of operations that resulted in something vaguely hydrodymanic, instead of “Loft, loft, fillet, cut, fillet, loft….” until it looked like what I would have drawn in 3rd grade.

The wings actually wrap around the body completely this time, in accordance with Youtube videos and pages with low-resolution pictures from the late 1990s. The fuselage blends into the wing on the bottom, and the attachment is just done with a flat feature in the cross section of the wing root.

Now I’ve added the canopy bubble and mounts for the eventual More Pylons that I Must Build. Notice the little ears next to the bubble – they are actually screw attachment points, and there is one more on the bottom. The body is designed already as a thin shell, and the nose portion can detach cleanly so I can actually stuff in electronics. The tailplane has also been modeled here.

If you’re wondering why the 30mm fans don’t suddenly look derpy any more, that’s because this model is bigger. Way bigger. The total wing area will be effectively 5 times or so of m-CRP 0001.

I had to remember alot of my coordinate transformations to model the tailplane, because it involved pivoting a reference plane at 60 degrees off of a line drawn on the XY plane at 45 degrees to the X axis, then drawing a line offset by 33.01 degrees from that 60 degree rotated plane. Or something – maybe not in that order. This is so the CF rod holes in the tail are directly in line with the ones in the body. The fat extrusions at the end provide the anchoring point for the tail rods.

I’ve also added the model CF pylons, but haven’t designed the little adaptors that will mount the fans, since they still haven’t arrive yet.

m-CRP 0002 in profile. While the fans haven’t been rotated to match, notice that the pylons are actually canted downwards towards the body by about 5 degrees to act as PAR.

oh hey, fans bus motors!!!

I finally got the shipment in. Here’s the AEORC EDF-30 system that Hobbyking sells. They’re SO CUTE. Like little ducted kittens.

And to my utter surprise, my guesstimated dimensions were spot on. At least in the important part – the outer diameter of the case.

Along with the stock fans, I got a sampling of other small Turnigy inrunners (10 and 12mm) and some GWS 30mm mini-EDF parts. They will be used if two bus motors prove inadequate.

While the product has a ton of negative reviews on HK, mine seemed to be fine and balanced running on 7.4 volts. On 11.1 volts, I estimated that the thrust was greater than 3 ounces, but one started to smell slightly funny after 5 or 6 seconds. So it looks like m-CRP 0002 might be running 7.4 volts to save weight and to prevent me from Beasting It.

printing

I made a key modification to MaB before starting this build. Previously, I had a 125mm case fan just kind of chilling on top, pulling air upwards past the part. The problem is this really only worked when the part was very flat, so the fan was forced to draw air past the platform surface. Once the part got tall, the fan could get air from anywhere.

I popped into MaB’s design model, positioned the geometry of a 80mm case fan, and using my newly acquired lifting skills, designed this ventilation duct that parks directly under the extruder motor. It holds itself in place by virtue of enveloping some of the cap screw heads on the platform surface. A dab of glue to secure the fan, and I was in business. Now the fan blows down, impinging upon the top layers of the part and spreading along the platform. The 80mm LED fan is much weaker than the 125mm server fan I used before, so I just instruct the machine to always keep it on.

This resulted in a much more predictable dimensional variation for the builds.

Here’s the three body sections after the print run. The fan helped greatly on the small profiles of the nose and tip of the tail. I split the main body into two pieces – one of which is my artificial shell hull, and the other one is nominally “solid” but is just a 2-walled shell with zero infill. The single piece body would have been too tall to safely print, despite MaB having the Z-axis travel needed.

The addition of the cooling fan brings out an interesting feature of the print – namely, the thermal variations of the extruder. Check out the horizontal striations on the parts. Initially, I thought it was an artifact of the Z axis leadscrew because of my irregular machining. Further investigation proved that it was in fact the temperature of the extruder fluctuating in the hysteresis band around my nominal extrusion temperature.

I’m not too concerned, but I do wonder what can be done to minimize variations – I can think of a few things, such as less thermal mass of the extruder, higher power heater for the overhead, better coupling between thermistor and heater core, and maybe changing up some control gains or making the control nonlinear. This is something to explore for the next iteration.

And now for the big guy This is one wing (I forget which…) in progress. Notice that both ends have curled up slightly – that just because this thing is so big that printing in anything but a closed and heated cabinet is essentially asking for massive thermal stresses with the associated strain.

I actually resorted to supergluing the ends of the first layer down using ultra-thin Loctite 420, which wicked into the gaps left by the plastic threads. Even this lost out a few times to the power of curling plastic. Eventually, I found the amount of superglue which would hold the wing flat as it finished. It’s an absolutely disturbing amount.

Number 2 on feature request lis for next 3d printert: Heated cabinet.

That’s all for now. With both wings and the main body printed, it’s time to get started on the even-more-ginormous tailplane (as well as wait for my carbon fiber rods…)

micro-chuckranoplan: how i know i’m not an aerospace engineer

Feb 12, 2011 in Chuckranoplan, Project Build Reports

The verdict is in: μ-Chuckranoplan sucks. Well, I think design-wise it isn’t too bad, but it just weighs a ton. Specifically, the rear end is just too heavy for its airfoil size, despite being printed as a shell. The airfoils turned out to be too thin in cross section and printing them even as a shell made them almost solid.

It might do better as a 3 or 4 foot model, but certainly on the 9 inch scale, it’s too brick-like to glide. Maybe I should just take up crafting these things from foam.

I elected to print the body and nose sections vertically as hollow shells. I discovered that printing them “flat side down” made the prints prone to corner lifting with their broad surface. Additionally, the extreme overhang angles that form the bottom of the hull led to very poor prints on a fully hollow shell, or caused the g-code generator to insert several fullly filled layers for support, which caused them to be too heavy.

Afterwards, I cut a solid end layer out of each half such that they could form a continuous cavity inside.

Despite my requests to print a fully hollow shell, I still received some halfassedly filled layers and lots of internal webbing. To get those out of the way as an anticipatory move for installing little lipos and motor controllers, I blasted the interior for a few seconds with a hot air soldering rework station set to 400 degrees C. This melted away the thin webbing but kept the outer shell intact. Detailing included scraping some of the melted webbing out with a craft knife. The two halves were then joined using some thin CA glue – Loctite 420, the kind that wicks into crevices, especially those formed by your finger contacting the part in question.

Continuing the assembly of the whole thing. The ass-end of the craft was printed horizontally because some of the sections were too small to print vertically (at least not without massively slowing down or waiting for the last layer to cool). I continued using the cut aluminum rivets method to align the parts.

And here it is. I even included the bubble!

So how well does it fly? The answer is really not at all. As mentioned before, the tail is so heavy for one reason or another that it dominates the entire thing – the center of gravity is something like 80% of the way from the front of the wings (in airplane words, I guess that’s “80% chord”?) so the only thing it does when given a flick from ground level is stall out and then faceplant on the ground.

I taped little machine nuts to the front as a “counterweight”, but then it just faceplanted, being too heavy to do anything at all.  So for now, μ-Chuckranoplan 0001 will just be a display model. I’m not going to bother fitting it out with electronics.

How to fix this issue? Well, I could either build a model the way you’re supposed to build small aircraft models, which is with stiff foam, balsa wood, and plastic film, or make a better printable design. The real way is hard, 3d printing is easy.

I’ll probably redesign this thing to more closely follow the actual German delta wing designs, which, as this video demonstrates work excellently if they don’t weigh like 8 pounds.