Archive for March, 2011

 

Chuckranoplan 0003B and this Monaco thing

Mar 29, 2011 in Chuckranoplan, MIT, Bostoncaster, Cambridgeshire, Project Build Reports

I’m going to Monaco.

Yes, that Monaco.

forever.

Actually, for EVER. See what I did there? The story of this one is that a few months ago, the MIT Scooter Party (all… three? of us) wrote a technical paper on how amazingly badass our collective scooterness is, and submitted it to the Electric Vehicles and Renewable Energies conference that is held in Monaco every two years. By some seemingly stochastic alignment of the universe, we were admitted to the conference.

So now we’re bringing the party to Europe. I swear, I’m going to run that F1 track-equivalent at least once on RazEr rEVolution. This trip would make RazEr my most well-traveled vehicle, since it also went to Singapore. As a microscooter, I guess it has an affinity for microstates.

I wonder if I can sell a few to Sealand.

Anyways, that’s tomorrow through Sunday. For now, Chuckranoplan!

Here’s MaB executing the best one-wall print I’ve seen it do yet. It’s printing the front end of 0003B as detailed last time. The walls are 0.5-0.6mm thick, and the internal spine is 1mm thick to allow for two passes. I was surprised at how well the Dual Integrated Bus Motor Mounts came out here.

Next I dropped the two middle segments on. I placed these two parts next to eachother such that their bottoms (keels?) were almost touching. That “encouraged” Skeinforge to make the layer changes and jumps at that point, and I must say it worked very well.

After all four segments were printed, I Dynamically Averaged (read: belt-sanded) smooth the interfaces and put them together with a CA glue initial bond, then Goop seal over that. The weight of this assembly is about 1.5 ounces.

A look down the tunnel of the hull before I closed it off. Having the ribs there prevented this piece from distorting and buckling all over the place quite well.

I split the tailplane into 3 parts and arranged them with the trailing edges close together. This view also shows the extra material inside to prevent warping.

The next effort was to print off the new wing design. This, unfortunately, turned out to be quite an exercise in futility for reasons which will be detailed. I’m not sure if it was the ambient temperature, the weather that day, or if I was sitting too close to it, but the prints failed repeatedly after two or three inches. The plastic would buckle, and generally with very little warning just completely delaminate with a very loud crack. It’s just a consequence of thermal cycling for a part this large.

Heated cabinet… heated cabinet… heated cabinet. Need to get on that.

Oh, ignore that McDonalds box. That’s a very warm spot, okay?! Once I have said heated cabinet, it’ll get even worse.

What happens when you try to print skyscrapers on a machine designed to print flats.

I have several ideas I want to try regarding how to prevent the severe thermal warping effects, including splitting the wing print in half laterially (so there’s a front and back half). I already tried splitting it longitudinally (to left and right halves), and it did work, but I also discovered…

z-axis drift

…that for some reason, the Z axis motor is not locked in one position during a layer. I’m not sure when this changed, or if it was always this way, but my Z-axis is now so incredibly smooth that it wants to fall back down the leadscrew.

The problem is that my layers are so fine (0.25mm) that even a sixteenth turn of vibration-induced sliding causes the smearing of an entire layer. I’d notice horizontal stripes forming on parts which were very large in cross section (noticeable on the tailplane prints). The machine loses enough layers such that the final part may be more than 1/4″ shorter than normal.

At first, I thought it was due to excessive friction on the Z axis causing the steps to fail. However, after lubing it up and noticing the problem getting worse, I concluded it had to be the opposite problem. So now I’m trying various things to increase the static friction on the leadscrew. The best solution would really be to make sure the Z motor is held with current during a layer.

The most severe vibrations happen on 45 degree passes when both axes are running at roughly 3/4 speed, so I may try positioning the wings horizontally or vertically after splitting them in order to minimize time spent moving both axes near the machine’s resonance peak.

How the hell did I manage to design a machine to so perfectly resonate at the exact speed I’m going to run it at?

 

Chuckranoplan 0003B: What’s probably going to take up the entire rest of spring break…

Mar 25, 2011 in Chuckranoplan, Project Build Reports

My initial goals for Spring Break:

  • Get the melontanktroller designed for LBS
  • Finish some summer project proposals
  • Finish an experiment and associated paper
  • Finish a paper for another class
  • Prepare for EVER 2011 (more on that later)
  • Whatever else minor project maintenance needs to happen.

What has actually happened:

Ah, productivity.

Anyways, with the relative success of 0003, I became more or less entranced over designing the next iteration. It was going to look much the same, but feature the internal printing trickery I explored for 0003, so I’m going to not designate it 0004 for now. It also doesn’t help that I bought yet another book on these things.  Maybe by the end of break I’ll actually know how they work so I can find employment building them after completely destroying this semester grade-wise!

The goal for 0003B is to keep the same overall layout as 0003, but make it powered. To that end, I’ve redesigned the nose and body to hold the two 30mm ducted fan units from 0002 and provide real electronics bays.

The nose is no longer a fully hollow shell, but now features a network of internal webbing that I guess is kind of like bulkheads and stringers. The Dual Integrated Bus Motor Mounts replace the dumb carbon fiber rod and hot glue mounts of 0002. The weight is still minimal, however. In ABS plastic, this part should weigh half an ounce.

After a few more hours of not paying attention to classwork…

I’ve added more internal spars to the wing shell, mostly to discourage the buckling and warpage seen in 0003′s wings. Note also the “electronics bucket” where I normally put a bubble canopy model. I’ll probably load this up with batteries and then cover the top with a strip of duct tape. Headless but flying Chuckranoplan is better than visually appealing but dysfunctional.

In addition, there’s one more design experiment being applied to 0003B, and that’s the attachment of the tail rods. In previous iterations, I designed the rods to be inserted into printed holes in the body. However, I’ve found that MaB really can’t build a perfect hole at all, and especially not at such a steep angle from the horizontal (printing the holes with the fuselage vertical means 60+ degree overhangs, which without support material, is very difficult). To fix this, I’ve located the rods just outside the body, sitting in angled and rounded channels. Now, the channels route as part of the outer perimeter, which means it ought to be much cleaner.

The same operation was applied to the tailplane. The center part, formerly with four through-holes, now has four edge slots. The idea this time is that I can actually mount the center piece first and check the angle before closing the rods off with the end pieces. So I don’t end up with another drag chute tail.

Also note the reduced amount of span members in the wings. I decided five was excessive and added weight which I needed elsewhere.

You would not believe how many tries it took to get this picture, since the simulator in ReplicatorG runs at 0.8 Mach. There exist Linux hacker ways of slowing it down, but I’m not in the mood to recompile a kernel, so I just kept recapturing the screen hoping to get a complete profile. And I did, finally. The above is a shot from a test route of the nose portion, showing the outer perimeter (just past one of the fan mounts), the internal cavity, and two underside reinforcement members being routed all as one single-wall profile. Ideally, if MaB just does this kind of thing for an hour, I won’t end up with gaps or thin spots.

Stay tuned…

Chuckranoplan: It Does the Thing It Does the Thing It Does the Thing!!!

Mar 24, 2011 in Chuckranoplan, Project Build Reports

IT DOES THE THING

What’s the thing? Well, I don’t know, but IT DOES SOMETHING.

As a reminder, here’s a picture of Chuckranoplan 0003.

It’s impossible to see in this picture, but I’ve very carefully designed it as a hollow shell with 0.5mm walls and 1mm thick internal ribs. I played around with the hollowness in order to try and trick MaB into making each cross section a single path, so there’s no jumps or hops for it to leave ugly gaps everywhere.

I used a variation on the incomplete hole such that the routing script Skeinforge was forced to follow a 0.5mm wide path around the perimeter, then forced to go back and forth across a 1mm wide internal rib that actually dead ended 0.1mm short of completing an internal compartment. If this wasn’t done, MaB would try to jump across the part and make closed loops, but the extruder doesn’t start and stop that quickly, so earlier test prints got nasty gaps and chunks missing. The 0.1mm gap means physically the ribs still join the outer surfaces, but the software thinks they do not join.

Using this technique, I was able to produce this one-wall-thick Chuckranoplan 0003 hull. It weighs under 1 ounce, and in fact does float.

Okay, so it’s actually four chunks. Single wall parts like this deform and split much quicker than multiwall parts, since their structural integrity is so poor. Now that I’ve learned how to dupe the printer, I’ll probably begin adding internal ribbing to prevent splitting and the buckling that is clearly visible.

A few hours later and I’ve managed to pop off the wings too. These things are the second largest objects MaB has ever printed – the first largest being the wings of 0002. These wings are shorter, but the chord is the same. There was some base warpage, but this time, they’re at the outside pontoons, which means I really have no fucks to give.

Notice the splits and cracks at the trailing edge, as well as… well, all over the place! The heat-cool cycle is really rough on the single wall plastic. A heated cabinet would have prevented alot of these stress cracks from forming.

The gaps were patched up with Goop cement, which was also used to join the wings to the body.

Next, a propeller.

Alright, just kidding. These are actually the three parts of the tail airfoil, arranged in a pretty flower shape so they could be printed off all at once.

The tail suffered some splittage issues too, but they were repaired with superglue and… more Goop. By the time I’m done, this thing is going to be half Goop.

To assemble the tail, I cleaned up the 1/8″ through-holes using a hot stick of steel to melt-form the ABS plastic. The print process doesn’t leave perfect holes, so this was a way to make sure the carbon fiber rods didn’t get stuck or were … denied entry or something.

Alright, here’s take number 1 of Chuckranoplan 0003. This is probably the biggest thing MaB has ever spit out… and now that I’ve figured out how to make these things for realsies, they’re gonna get even bigger.

Okay, here’s take two.

omg what happened to the tail?????

Well, if you observe the previous picture, you’d notice that the tailplane isn’t in a lifting configuration at all. In fact, it’s at about -3 degrees from horizontal or so. This was due to my inattention when supergluing the tail rods together. So what do I do now that I need to change the angle of some superglued CF rods? Well, I have to break them apart and add in spacers.

The crunching sound of the tin shears was rather sickening, like re-breaking an improperly set bone.

Unfortunately, I couldn’t quite control how much spacer I added, and the tail ended up this time at like 30 degrees. I think it’s legitimately making more drag than lift.

The method of joining was gobs of superglue with a heatshrink exterior, then wrapping the entire thing in blue masking tape because blue masking tape.

I added some cute trailing edge flaps with… more blue masking tape. Hey, if you look at it from afar, it looks like I actually meant for them to  be blue.

The machine nuts up front are center of gravity adjusters to bring the CG away from the back of the wing. Unfortunately they also added another ounce.

It turns out that those flaps were totally worthless as-taped due to the tape separating underneath and just adding more drag.

Another view. I must have reglued that tailplane like 6 times, since chunks kept falling off.

Alright, here it is after “flight testing” for the day.

That looks so disgusting. I don’t think I’m using that picture as the “press photo”.

next

Chuckranoplan 0003 explored the “DACWIG” type layout where the wings are a substantial percentage of the body and also have a relatively large chamber underneath. This particular topology is a compromise between legit stubby-winged airplanes (standard WIG, Russian ekranoplani) and dynamic air cushion craft (skirtless hovercraft). This model was more of a CAD and printing exercise than anything. Now that I have a better idea how to manipulate my 3D printer to bend to my wishes, I’m going to rehash the design using better internal ribbing for more strength and less buckling. And perhaps I can include actual electronics and battery mounting facilities this time too!

There are pics, so it definitely did happen. But what’s better is video… so here’s some throw-testing of 0003. You have to promise to not laugh, however. The thing barely skates along the ground, and is certainly not as successful as this dude’s models. However, I believe it is supporting most of its weight in air cushion at this point.

To do more testing, I really need to move to a larger and more open space. Those collisions in the video weren’t the only ones, and certainly were not the most destructive ones (I had to legitimately glue the right wing back together after one run-in with a door frame). There’s too many things sticking out in the hallways for me to feel comfortable giving it a real toss.

WHAT’S NEXT?!

LandBearShark: Okay, so I Sold Out.

Mar 20, 2011 in Land-Bear-Shark, Project Build Reports

No, not to someone who wants to mass produce it for the world (for better or worse), or to mass media (yet), but to brushed DC motors, just like I said I wouldn’t do last time. I should consider a career in politics or something. Anyways, out of practicality reasons and an attempt to not to try and address 5 design flaws at once, a change from my usual build tactic, I’ve elected to switch LBS to a POD (Plain Old DC) system. DC motors don’t have blank sensor states, and controlling them doesn’t involve a state machine (or nested loops with rotor position estimators and current sensors and matrix transforms). In the space available for drive motors I won’t be able to mount anything as powerful as the rewound 80/85s, but judging by how well I was able to stay on the thing during the preliminary trial, I don’t really have a problem with this. My main goal is to get the fingerless-control designed and built so I can test if it’s worthy of further investigation.

By the way, here’s an interesting build from a few years ago that accomplishes everything LBS was supposed to do, and in a better form factor. See, I should have just done this straight away.

The concept is pretty close of the Boolean union of Snow Scooter and LBS.

Oh, also, here’s a sneak peek of Chuckranoplan 0003:

No more Delta wing?! I’ve been eyeing different wing planforms since I got this book (which is awesome but clearly shows the imperfections of being a first print edition). The authors break down the cladistics of GEVs more finely than I do, separating the field into five categories based from my understanding on operating altitude and the ability to statically hover. Of particular interest to me was everything that wasn’t the Lippisch delta wing. The real reason I don’t like them as much despite their clear advantages in ground effect is that I currently have no clue how to build a tapered wing For Real. Unless, perhaps, I build a garage-sized 3D printer, which of course is not out of the question.

And they look derpy.

I’ve been looking into some of the Chinese GEV designs recently, since they have all featured square wings with small vessel-length-to-wing-chord ratios (the wings are very long compared to the vehicle length) and aspect ratios (almost square or even rectangular the wrong way). A square wing is much more in my comfort zone of things I could conceivably build without getting too deep into specialized materials, tools, and techniques. This is important, since a 20 foot Megachuckranoplan is still lurking in the back of my mind.  One example is the Tianyi 1. This vessel (and its forerunners) are the subject of some pretty intensive analysis in the book. They represent a middle ground between the more ship and hovercraft-like types which can only skim over calm water (such as the Aquaglide) and the Legit Ekranoplans of yore which could operate at higher altitudes…. and by higher I mean like 10 feet.

To prevent this from turning into a Chuckranoplan post, I’ll leave it at that. This model will still be 3D printed hollow to the best of my ability, but the square airfoils mean that I could easily make a more conventional model aircraft wing out of some laser-cut balsa wood. Weigh estimates put the above at 220 grams dead empty, and only if MaB gets it right.

oh yeah, melontank.

This thing here is going to save my day.

What IS that?

It’s a weird little integrated 4:1 planetary gearset from some kind of Currie electric bicycle. Around 2 years ago, they started showing up on the surplus channel, but it looks like they’ve been around before that, just more expensive. The construction is fairly “discount EV” standard – sintered steel gears, cast aluminum carriers, and rivets. What’s cool about them is that they have inbuilt support rollers coaxial with the planet gears that turn the whole thing into a really big and shitty roller bearing, so they can take some moment load. Presumably in the bicycle, there was no other bearing for the chain sprocket.

I have a little bit of history with these, as they were first discovered and used for the Greenwheel project. But more recently, they’ve been ingeniously used to great effect on the ExkateCD project as a “preduction” stage to replace an existing drive motor with a faster one but retaining roughly the same output speed after the belt drive.

Hey, sounds like something I need to do. The motor they used was a stock CIM motor, the same kind used on FIRST robots, but the motor I’m going to use is a…

… Okay, it’s a CIM motor. Whatever. Stop judging me.

Incidentally, it’s the same motor that powers Segfault. They’re made in the tens of thousands each year for FIRST robots, are rated for a solid half horsepower each (which really means you can punch 2 or 3 into them for a very short amount of time), and aren’t too obnoxiously sized. Oh, yeah, they’re cheap, about $30 new, which makes the price to performance comparison very good for a stock janky DC ferrite magnet brush motor.

So the plan for LBS is to down-convert the 5000 RPM @ 12v by 4 first, using the Shady CurrieBoxen, and then run the stock 5:1 chain drive that’s in place now. Essentially a direct swap-in of the two brushless outrunners that are in there now. The combined 20:1 reduction might actually be too much, so I also have the option of going to 4:1 chain drive with the alternate sprocket set I have. Doing this would mean that I’d have to drop the 36v power system for LBS and go to at most 24v, which will be plenty anyway.

This will be accomplished using the unique mounting configuration for the Currieboxen that ExkateCD explored. The ring gear is bolted to the motor mount, the output is taken through the carrier plate, but the output completely floats on the motor shaft. The CIM motor shaft is just the right length to stick out past the gearbox enough to put a ball bearing on. The output coupler can ride on this bearing while being attached to the carrier plate at the same time, which is firmly held in place axially by the built-in rollers. The assembly is made stiffer than the stock (very loose) output because of the addition of the coupler-side bearing. And of course the motor mount has the same bolt pattern as the 80/85 motors.

Here’s how the assembly goes together. The ring gear attaches to the motor mount through the perimeter bolt holes, which appear to clear #8 machine screws.

And the assembly from the back, showing the coupler with 8mm ball bearings.

Here’s the assembly dropped into the existing motor mount on LBS for a fit test. The new motor setup is longer, extending all the way to the other side of the track pod, but it’s smaller in diameter. Overall, it’s not a painful conversion. I just have to sit on my ass now and wait for the motors, supplementary Currieboxen, and random hardware to ship from all over the place.

In the mean time, my focus will be on implementing the fingers-free controller. The spring demo season (where I spam exhibitions, fairs, and events of all kinds with everything I’ve built ever) is under way, so I also need to get some of the other projects running again. The poor RazErBlades, for instance, are out one bettery pack (and the replacement I bought doesn’t fit because fuck you Hobbyking for changing the dimensions of products without updating the website and picture!).  Ideally, Melontank will be ready….

by Commencement.

[dun DUN DUUUUNNNNNN]

FanKartTwo: Full Metal Something-or-other

Mar 17, 2011 in Fankart!, Project Build Reports

That sucking sound you hear is either my motor bearings about to disintegrate or the fact that the whole thing has no inlet lip whatsoever.