Category: interesting objects

idk i guess its not a robot, van, or dumb scooter

Kitmotter

chorl3 Comments

Hey Charles, where do I  buy one of those motors you use on your scooters?

Hey Charles, where can I get parts for my own hub motors?

Hey Charles, how do you build a hub motor?

Hey Charles, can you make some motors for me?

Hey Charles, have you ever watched Air Gear?

Hey Charles, what does the third wire do?

OKAY ENOUGH ALREADY

Ever since the first RazEr motor was completed and validated, I’ve been getting various permutations of those questions. The operating mechanics of a hub motor vehicle are mysterious to many people not directly familiar with brushless DC motors. What I have some times found is that people (here, anyway) are willing to learn to understand said mechanics, and possibly build their own vehicles around them. Another thing I (and certain other scooter-fancying parties) have found is that building your own little EV, or even a motor, can be a decently well-rounded engineering learning experience. After all, even a simple EV has a chassis, a motor, a drivetrain, control electronics, and some kind of user interface. It’s an all-in-one package for learning the basics of mechatronic engineering, and yields a practical and usable thing at the end.

Unfortunately, the motors I’ve built are all handcrafted pieces of metal billetwork. RazEr rEVolution’s motor is crafted from a 4 inch aluminum pipe with 1/2″ thick walls and some 3.5″ diameter aluminum solid billet.

Such processes aren’t very “democratizable”, for lack of a better term. Few people will have regular access to well-equipped machine tools that they know how to use. I could also just mail one to a shady backwater motor shop in China and potentially make negative millions of dollars, but that’s not very fun and engaging now, is it? The material costs for such a motor like the DNIR alone are pretty hefty. Certainly the first batches for the early adopters (like all 3 of them?) wouldn’t fall under $1,000 USD each. And again, a shiny aluminum cylinder is very “black box” and doesn’t tell you much about what’s inside, and if the point is to make you see what’s inside, then it’s not very… err… productive. As the “decentralization” of manufacturing continues, so the accessibility of previously proprietary and industry-specialized technologies spreads. In general, that’s good for the likes of Makerdom. It also means people can stop bugging me about how to build these damn things.

Soon, anyway.

That is why for the past short while, Shane and I have been investigating how to manufacture these motors with now ubiquitous personal fabrication methods and short-run online fabrication services. Such methods include 3d printing parts from places like Shapeways or your very own machine, laser cutting components using services like Ponoko or Pololu, and abrasive waterjet machining by places like Big Blue Saw (HI SIMON). Other typical motor components like magnets and wire can be purchased readily from mail-order dealers right now. There’s still some kinks to work out, and some machine work is sort of unavoidable for most motors, but we think the majority of a motor can now be cobbled together using nothing but the willpower of the Internet.

Without further ado, I present Project Kitmotter.

Alright, so what’s going on here?

First off, the motor architecture is similar to the motors of BWD, Pneu Scooter, and the Jedboard. Shane & Company have piloted the “stacked plates” method of motor design, where…. well, stacked plates of material are aligned by pins and secured by through-bolts. The rotor is made of several stacked rings of waterjet-cut steel that have magnet-securing indents as part of the structure. The endcaps are formed by a spacing ring and flat plate with a center bearing bore. And out of my design book, the stator is secured on a (possibly machined) center shaft with a channel cut out on one side to route wires out.

The three motors above are identical topologically, but there are minor variations resulting from playing with the design.

The first motor uses a 63mm (or 2.5 inch) stator with a can that is 82mm (or 3.25″) in diameter. This is particularly interesting because 82mm is generally the largest you can core out a 125mm scooter wheel to. The 150mm variants such as this or the first three or four on this page are also corable to the same range.

The shaft, as-designed, is a single machined piece of aluminum that is double-stepped, 1″ in the center and 3/4″ on either side, with a milled channel and two end-tapped holes.. This motor was originally designed a little while ago as a prospective machining exercise for the introductory machine tools course here, and so contains a sampling of machined features.

Motor #2 is a 68mm (2.6″) stator’d design with a 3.75″ outer diameter. This thing has massive magnets – 0.25″ thick, for maximum airgap flux density. Given that these motors are designed with very wide tolerances (airgaps on the order of a whole millimeter or more, unlike my ultra-tight 0.5 to 0.3mm airgaps!), I think the fatter magnets are beneficial as the ratio of airgap to magnet thickness becomes smaller. The center shaft on this one is much the same as the first: single piece and machined.

I don’t know of a wheel that fits around this motor at the moment, but a 68mm stator was chosen for reasons I will detail shortly.

Here’s something a little different. This motor is essentially motor #2’s design – quarter inch magnets with the same rotor plates, but 3d-printable endcaps and center shaft. The shaft has been up-armored to 1 inch diameter (with accompanying absurd bearing) for  extra strength, since it would be 3d printed from plastic.

A plastic motor shaft might seem like a horrible idea, but it’s important to remember that in these motors, the shaft is stationary. Therefore, it can actually be considered part of vehicle structure. Furthermore, the anchoring points of the shaft on the vehicle are usually close enough to the bearing that the plastic is loaded for the most part in shear, not bending, which is one type of loading where the diameter increase helps substantially.. And if all else fails, just drive a 1/4″ bolt right through that center hole there and all the problems are solved. The vehicle loads will then be transmitted and taken up by the steel hardware, and the plastic shaft around it becomes more or less just a shaft diameter spacer for the bearing.

Printing a shaft that’s 3-4″ tall and only 1″ across at the base, though, is kind of problematic. Especially with the moving-bed designs typical of smaller and DIY fab-scale 3d printers, including mine. So I dropped in a quick way to print the shaft as two seprate, shorter prints.

The two halves are joined by a dovetail interface for alignment purposes. The idea is then to finish-drill the pilot holes to 1/4″ diameter after fusing the two halves together with epoxy or CA glue, then putting a bolt all the way through. While the shaft could be used by itself without metallic reinforcement, adding a bolt in the through-hole would make the whole thing more stable. I came up with a few other dovetail joint designs, including one that doesn’t have the mating faces handling bending loads (i.e one that is like the above but turned 90 degrees), but the idea is the same.

I printed some design iterations out on MaB to test for tolerances and fits. For the truly budget- and sanity-constrained, I think the 3DP shaft is perfectly fine as a solution.

ok, so where do i get one of those stator things?

As mentioned before in my electric hub motors writeup on Instructables, the stator is the most difficult part to get for a motor, usually. That’s because it’s a component typically produced in the thousands or tens of thousands at a time by automated machinery, so you can’t get just one. The stator is always a series of stacked, very thin, silicon-alloyed iron sheets that are insulated from eachother by an oxide, phosphate, lacquer, or epoxy layer.

No, cutting up sheets of Home Depot galvanized roof patching does not work. Nor does machining it from a solid piece of steel.

Citizen-obtainable stators include the stock factory-made ones found on places like Gobrushless to custom laser-cut laminations that you can get from places like Protolam (who supplied the rotor plates for BWD and its kin). However, the latter option tends to be rather expensive. A custom stator is, therefore, the trivial solution if you have money to throw at the problem. For Kitmotter, I’m taking the approach that I have historically used: by harvesting stators from existing sources. Generally the source has been large laser printers and copiers, which tend to have outrunner-style stepper motors to crank on the paper feed path or spin the transfer drum.Copiers and printers are copious here at MIT (and I would guess the same for any large institution with a paper-pusher army). However, what I never did was think about logging the part numbers or models of machines that my stators for RazEr, the skates, and the like came from.

A while back, I sleepily ordered a pile of different copier motors from Ebay, and therefore finally had the opportunity to do some cataloging.

I’ve saved these numbers in a spreadsheet here. Yeah, I know, there’s all of 3 different motors on there right now. If you come upon a reliable model, you should help me append the list.

Bottom line is, the 3.25″ kitmotter uses a stator from the Docuprint 2125, which seems to be rather rare, even on eBay. I only got 3 of those motors, and now can’t seem to find more. The stator was chosen for its good fit to a 3.25″ motor. Further “production” of the motor, such as for the machine tools course, would probably necessitate a custom-cut stator. The 3.75″ kitmotter uses the 68mm Phaser 4400 main drive motor, Xerox part number 127K35701 (That sounds so much like a McMaster part number. I wish…)

In fact, I use three of these things back-to-back-to-back in RazEr rEVolution’s motor. The Kitmotter only uses one stator for simplicity. They appear to be fairly common on Ebay, so I’ve temporarily designated the 3.75″ kitmotter as the “safe design”. What’s even more convenient is that these 68mm stators are the same type used in the Turnigy “Melon” motors. So, conceivably, you can just get a Turnigy motor for the sole purpose of harvesting the stator out, and making an ultra-wide motor. Or splitting the stator into several smaller segments. Trust me, you cannot find a custom stator that big for only $100.

where does the wheel go?

As it stands, the kitmotter designs have no wheel mounting provisions. For instance, I have settled on a threaded locking ring design, whereas Pneu Scooter from above actually secures the motor into the wheel using machine screws. The method used on BWD, which is wrapping polyurethane strips around the tire and attaching them with high-strength urethane glue, is currently the method of choice for Kitmotter. McMaster has urethane strip in many different hardnesses and thicknesses, so it’s not too difficult to wrap a custom tire. The tire could be sanded to fit a curved profile, or like BWD, just run straight and flat.

Alternatively, one rotor plate can be expanded into a flange shape and have circumferential radial holes waterjetted into the flange so a tire of arbitrary size can be bolted on. I have yet to find a tire that is amenable to this treatment, and that is of course a branch of investigation for the project as it progresses.

tl;dr

I’ve already ordered some of the parts needed to finish one kitmotter, the “#2″ 3.75″ diameter, 0.25” thick magnet design. The rotor plates will be cut on-site here, along with the endcaps. I’m not going to release the designs quite yet, since they’re still unproven and may need some more dimensional adjustments. I’ll also look into tire-mounting provisions.

Regardless, there’s already more than enough information lurking in this post and others for you to boot up your favorite CAD program and fire off a design to Big Blue Saw.

Make-a-RazEr Round II

chorl2 Comments

Remember the trippy PCB?

During the last few days, I finally got around to installing it on MaB. I must say that the hot build platform made a tangible difference for large surface area parts. However, I also found out that I need to study PCB heater trace design more, since it turns out I designed the trace length without remember that resistance increases with temperature. That means this trace actually reaches thermal equilibrium when the temperature is about 95 C. It literally can’t get any hotter on 12 volts. I’d need to rig 15 volts or 18 volts to the thing somehow if I wanted the temperature to reach the 110 to 120 degree temperature that ABS is usually deposited on.

Oops. But, even 90 degrees has enabled me to start printing large structural parts. We’ll get to that later. First, the trippy PCB.

The aluminum plate will be bonded to the PCB with some kickass silver-filled epoxy. I imagine it’s kind of like Arctic Silver heatsink paste with a dose of resin and hardener in it. This little bubble cost $23…. so maybe next time I will just mix Arctic Silver with a bit of epoxy and call it a day.

For reference, the McMaster P/N is 7661A11.

The mounting process is just like applying said heatsink paste to your expensive multi-core unlocked CPU. I was surprised the stuff could be spread so thin, but I guess thin is better for heat conductivity.

The hole in the middle is to jam a little thermistor in for temperature readings.

The piece of Uniform Thermal Mass goes on…

…and is held to set with about 80 pounds of steel. This picture was also taken before I applied ~18 volts to the thing and let it bake under pressure while setting.

Next, I attached the thermistor to the underside. I later came back and edited this mounting with more Kapton tape and a little fuzzy ceramic blanket (the same stuff that is wrapped around MaB’s extruder, and comes in the kit).

And on it goes.

I ran a few bonding experiments with ABS to see what kinds of build surfaces performed the best. First off, bare aluminum is just not happening. Nothing stuck, at all.

I used the 6 inch wide Giant Kapton once, but also got very poor sticking performance. I’m led to believe that the temperature deficiency is to blame, since the greater 3d printing community reports good performance with ABS at 120 C.

Overlaying a piece of acrylic (i.e. the original Interim Build Surface) and letting it heat up worked great, but what the hell is the point if I have to keep laser cutting and replacing the acrylic slab?

Finally, I went and bought some blue painter’s tape, which is the other favorite of the Internet.

BUNNY… well, half a bunny.

Go figure, it works great. This kind of tape is also known to work when cold, which could come in handy for small parts.

I found that the tape’s adhesive was less than full strength 90 C, which caused some parts to lift if they were placed right next to a seam. The solution was to drop a perpendicular layer of tape over the strips to anchor them down better. That, or make sure the part is bigger than the seam.

With the heated build plate, I was able to embark on some large surface-area prints that would have failed before due to warping or the next layer not being able to stick.

I made this cute mecanum wheel as a test of a LSA part. I’m not currently working on anything that has a mecanum wheel, but it was a good test of 45-degree overhung holes too.

Finally, I went all-out and decided to make a part so large it literally takes up my entire build surface. As in, the Z axis was about 1/16″ away from colliding with the table as it passed by on the furthest lines.

Something remotely this huge wouldn’t have made it past the base layer stage on the old acrylic platform for sure. I also noted better perimeter adhesion than previous unheated prints, though I would also wager that was due to me setting the “first layer head height” lower, telling the machine to mash the first layer down harder.

So what can it be?

Okay, so it didn’t quite work.

But it’s a Pop Quiz frame. I selected PQ as a good candidate to try and 3d print the entire frame at once for, since it’s small, flat, and exhibits no strange shapes or overhanging layers.

It started out well, but it seems even large surface area parts have a limit as to how large they can be  on my current setup. The corners are all lifted about 1/8″ or so, which means the frame is pretty much unusable. I suspect at a certain point, just the surrounding ambient air is enough to upset the thermal equilibrium of the part – especially when there are alot of people walking by or working near, causing breezes.That’s why commercial 3d printers have their entire cabinets closed and heated to 90-100 C.

I don’t think I can pull that off yet, but when the second iteration rolls around, it might be worth a shot.

razEr rEVolution

Make-A-Bot has proven itself  now as my first maniacal creation which can make parts for my other maniacal creations. I see a thousand ways this can end badly, but in the mean time, Make-A-Bot is now fully able to print structural mechanical parts. The first real task I assigned it is to help finish RazEr rEVolution.

You know, the next open bracket that I paused for no reason in the middle of the term.  When I last left this thing, it had a motor with 33% functional Hall sensors, no controller, and a front wheel that looked like it belonged on a… well… push scooter.

Those first two problems are electrical, and so I will never solve them. But the last is mechanical. Let’s see what I can do about it.

This is a meaty, manly caster wheel from Surplus Center. I bought it with a shipment of melonshark parts as an investigation, and overall I like it. Most importantly,, it’s the same color as RazEr’s back wheel.

It came with the standard 3/4″ Enormous Roller Bearing center hub, which, needless to say, I couldn’t easily use. Those bearings don’t even start turning until you have a thousand pounds on the thing (but when you do, you’ll be really glad you have them). No ball bearings I had on hand were large enough to not fit into the center bore after the removal of the roller bearing. In other words, I was intending to just bore the center out, but didn’t have bearings big enough. A 6201 type bearing would have been optimal, but I only had 6001 bearings.

The first chance that Make-A-Bot got to be useful was making a bore adaptor for the bearings. This was a simple thin-wall tubular part.

I found out that the diameter difference between the bearing and wheel bore was too small for the software to route more than 2 perimeter layers. There was also a discontinuity in diameters between the bearing pocket and the center portion of the tube. Overall, I tried fixing it for 2 hours before giving up and just carving one out of Delrin on the lathe. Imagine that. Me, machining something again. It was a theraputic experience.

With the wheel bearing problem resolved, it was time to figure out how to mount this new monster truck tire. The stock Razor scooter fork that I had on the front is thinner than the webbing in the plastic hub of this wheel.

Instead of doing it Shane Style, I elected to see what Make-a-Bot could do.

The answer: Massive brick of ABS that happens to share the same mounting dimensions as the stock steel fork. This piece is a good 4 inches tall and 2 inches deep (in the above coordinate system).

Here it is imported into Skeinforge. It took a while to route just because it was massive.

I set the interior fill ratio at 50, so it ought to come out sort of hol…. wait, 50? Yeah, that’s actually 5000% fill. So it means I should be able to enter the 4th dimension and pull 49 additional copies out of this thing, right?

Anyways, by the time I figured out why it was doing every layer at 100% solid, it was already far enough along that aborting would have been foolish. I left, and returned several hours later to…


It’s replicating! AAAHHHHHH

…a hornet’s nest or something growing on my 99% done part.

A fellow MITERer told me that it seems to have just crashed at 99% and kept the extruder going full throttle for several minutes before it was discovered and shut down. Wow. Well, at least I can mill it off. The SD card I had been using was exhibiting file system errors, so that might have been the cause.

Unfortunately, after milling the tumor off, I discovered that some of the major dimensions were off. It didn’t actually fit on the fork since I made the two slits to fit the stock one too narrow. Even worse, the wheel rubbed against the underside of the thing.

Lame. That was like $5 of ABS right there. I should figure out how to recycle MaB’s own printer droppings.

So what’s next? Nothing except apply the changes and try again. Taller fork, wider slotting, and making damn sure it’s on 50% fill instead of 50 * 100% fill.

Here’s the result.

I must say, from afar, it looks very professional and almost like some kind of bizarre aftermarket upgrade for a push scooter. I would still need to make the 12mm wheel axle and figure out some kind of pin or bolt solution for the interface between the fork and the steering neck. The rubber block sits in the same place it did before and provides a minute amount of suspension-like behavior.

So is this part structural? It should be. I guess real testing will tell me for certain.

With the aforementioned pins and axles installed, the whole thing looks badass. The black ABS matches well with the black deck and contrasts nicely with the silver metallic aluminum frame. I should take up interior design.

Said pins and axles were popped out on the lathe after I spent about an hour trying to figure out how to print a bolt. There are some times where I should just learn to suck it up and beast it.

Also pictured is the piece I was making on Make-A-Bot as a test. It’s actually the front “window” for the frame. I might remake this piece from PLA plastic so it’s reasonably translucent, or embed LEDs into it.

A closeup of the front end.

I’m absolutely determined to bring this thing to Singapore in a few days. All that’s left now are fixing the DNIR motor’s sensors and making sure some variant of the Turnigy controller or Melontroller works by Wednesday. Then I need to figure out how to not evacuate both Boston Logan and London Heathrow with a scooter, two robots, and a set of electric ninja skates. I’ve never had issue flying with equipment before, but the sheer volume of it this time, combined with the international trip, might make things a little more interesting.