Archive for the 'Project RazEr' Category

 

Random project updates

Jan 09, 2008 in Bots, Project Build Reports, Project RazEr, Test Bot 4.5 SP1

I remember two weeks ago when I said that I would build nonstop through the break; anything I could think of, and come up with more things if I ran out. Anyways, that totally went to hell. But progress is slowly coming.

I’m in the process of completely redoing the Snuffles Reloaded wheelmotor. The previous one was okay for getting the concept to materialize into something solid, but was done across 3 different machines and with little regard (or ability) to center the workpiece properly.

There are also some design changes this time around. Most notably, I’m redoing the magnet can out of steel to better fit the very strong magnets that will be used, as well as to simplify things.

The finished product. This was the single most torturous piece I have done yet. Steel of this diameter needs to be turned reeeeeeeally slowly. I mean like 70RPM slowly. It was very hard for me to keep my feedrate down (because I’m used to small aluminum parts), and if I went too fast, I would just be cutting shallow screw threads.

Not to mention that it was a hot-roll steel pipe that I carved this out of, so the steel was total shit and was gummy. The surface finish is horrible, but it’ll live. The perimeter holes were put in on a CNC machine (most awesome thing evar, only matched in awesomenss by a 5-axis abrasivejet. The two combined can carve the laws of physics and grammar such that they can both be most awesome.)

I’ll do the aluminum parts later. I might possibly redesign some of the internals to make the parts less weird.

In TB4.5MCESP1-related news, the arm controller came.

Ooh, shiny. This is a Syren25 single-channel controller from Dimension Engineering. It’s a nice piece of work, and supposedly features regenerative braking. That’s hardcore. So having opponent bots backdrive TB4.5′s lifter could actually be a good thing, but they should do it slowly and gently!

Since I was going to remount the ESC on my own heatsink plate anyway, I decided to take a look at the underside where the driver chips are.

O GOD WAT

NO! NO NO NO NO…. For the love of Robot Jesus Michael Fizgerald Jackson Christ! SO-8 surface mount FETS?!

I am SO GLAD I’m mounting this thing on a plate of aluminum that weighs more than it does.

I hate those tiny SMT FETs. Especially when the manufacturer (generally of brushless controllers) use like 20 in parallel to get a decent amp rating. What money and weight do you save then?

It’s even worse when you listen to the chip maker who says a 25A part is a 25A part. The SMT has absolutely no thermal mass and no capacity to withstand momentary overload. In fighting bots, you NEED THE OVERHEAD. I’m sorry, I will NOT run 25 amps through this thing for any extended length of time, and am very glad the arm is an intermittent duty cycle.

I’m making my own ESCs from now on, or I’m never building another robot of any kind.

Oh, and the heat spreader plate doesn’t contact the driver chips. That’s even less of a 25A ESC. DE might want to consider backing off the pressure a bit on whatever press that is stamping the aluminum sheet.

Anyways, time to actually start building the bot. The carbide teeth on the miter saw’s new blade are slowly subliming, so it’s best to use them while they’re still fresh.

Snuffles Reloaded: Update 6

Dec 10, 2007 in Project Build Reports, Project RazEr

It’s shiny. Really shiny.

After another day of furious metal shaving, the structure of the motor is mostly complete. Here are some in-process pics.

Motor side plates before length trimming. They were done step-by-step, one then the other, such that I had to move the asstastic tool holder as little as possible. In retrospect this was probably not the best option, since I did not have the luxury of a tailstock center, so the runout from rechucking a few times is probably obscene, but out of retrospect – oh well.

Test fitting a side plate before finishing the holes.

The internals.

Putting the stator together. No, that’s not the final winding. The wire coils are there to hold the individual laminations in-line while I pressed the whole thing onto the mount and while the epoxy hardened.

Threading the holes in the ring. Here’s Critical Design Flaw #1 (you knew it was coming somewhere!). The wall of the ring is too thin for anything other than, say, 2-56 screws. However, the waterjet made the holes bigger than 2-56 clearance-size. Oops. The next option was #3 screws, which are stupid and rarely used. Then #4, which was just on the edge of bursting through the walls of the ring. I risked it anyway, and things turned out okay.

Hey, the internal ferrous magnet ring. This was actually a rolled strip of 1/32″ steel. This is Critical Design Flaw #1.5, because in retrospect it would have been much better and more effective to make the whole ring from a piece of steel. I wouldn’t have to worry about the ring bursting out and implanting itself in my face, and the magnetic flux would have a better path to travel through. The bolt holes wouldn’t be so close to the edge, and the whole thing would be stiffer anyway.

The precarious seam that somehow holds everything together.

Test installation of the guts. Hey, it seems to be working out!

Actually, it doesn’t. The tolerances and machining imperfections disagree with eachother. Normally, this wouldn’t be that much of an issue, but inside the motor, the airgap between the stator and the magnets is tiny by design. Throw in a dubious lathe with no centers, repeated rechucking, transferring work between different machines, and my awesome skillz, and the magnets don’t fit in. A bit of sanding made things slide in, but it’s inconsistent over the course of one revolution.

There’s alot of improvements that could be made before I continue, such as redoing the can out of steel. Everything else seems to be fine – it rolls smoothly and there is indeed clearance for wiring and windings.

I could redo it, or I could hire it out to a machine shop to take my inconsistencies completely out of it. Although it’s fun to DIY, some times you just want things to work.

It’s finals season, so I have other priorities (like TB4.5-SP1!) to worry about before continuing on this. There’s nothing that prevents it from working conceptually, just manufacturing.

So that probably means SR is going to be on hold for a little while. But it’s still REALLY SHINY!

Snuffles Reloaded: Update 5

Dec 08, 2007 in Project Build Reports, Project RazEr

The calm before the storm has arrived. About a week from now, I’ll probably hole myself up in my room and prepare to attack four final exams in four consecutive days. If I survive, I’ll post.

Anyway, tonight was “chill night” after the rush by professors to get the last bits of work and final-exams-before-final-exams in. By chilling, I mean heading to MITERS and, after collecting as much balls as I could, hop on the lathe and work more on the wheelmotor.

The MITERS lathe is probably older than the cumulative ages of any randomly-selected 3 or 4 people in the room at any one time. It’s a medium-sized “South Bend” model, but is probably the single largest chunk of cast iron I have ever personally messed with.

 

After I got the hang of it, it’s actually a great machine. The tool selection is lacking, but I intend to raise the proposition of purchasing some tool bits and tool blanks.

 

There’s one thing I don’t like about it. It uses a “lantern” style toolpost that seems to have 8 degrees of freedom and can jiggle in any direction except the one you want. A forged tool holder sits on a weird-ass banana thing which sits in a shallow bowl. You have to hold down the bowl, the banana, and the tool holder, then tighten down the clamping screw. At once. Up hill, both ways, in the snow. It is evil. Then the end of the tool holder is a complete clusterfuck on its own, since that clamp screw likes to drag everything with it.

 

I ended up having to tighten everything down out of whack and then use a mallet to bash things into alignment.

 

Anyways, stuff turned out well.

 

The finished stator mount doohickey, with a bearing. Even with my eyeballing, newbie toolgrinding, and random bashing of the tool setup, it turned out well – most dimensions are within .005, which is just fine for starters.

Since I did not intend to disassemble this ever again, I press-fitted the shaft into the stator mount using an interference fit of about .01 inches. Three times what you are supposed to use. I stuck it in the biggest milling vise there was and cranked it with the biggest wrench I could find.

 

If this comes apart, I probably have also.

 

On a related note, BATTERIES! Here’s two lithium polymer packs I snagged off EBay. Each pack is 11.1v, 3900mAh. Downside? They were 10C. That’s what I get for going “ooh, cheap” – even though my application doesn’t call for military-grade batteries, I still need the newer 20-25C packs for the low-end acceleration.

 

Since 40 amps is unlikely to move even my ass around, I will probably save these packs for Test Bot 4.5 Media Center Edition Service Pack 1, under final planning for the 2008 competition season.

 

Snuffles Reloaded: Update 4

Nov 30, 2007 in Project Build Reports, Project RazEr, Stuff

Over the past two weeks, I managed to loiter enough around the Media Lab to get some work done on the wheelmotor. All I can say is that it’s “getting there”… those weird things called “classes” get in the way of building things. Who the hell needs ‘em?!

The first step in shoving a motor inside a wheel is to carve the center out of the wheel. This proved to be challenging – polyurethane wheels are squishy and would be nearly impossible to mount on a lathe chuck the conventional way. The spokes also prevent grabbing anything inside the wheel.

Naturally, I come up with a clever, dangerous, and potentially expensive plan to get around it. The method of attaching the wheel on Snuffles 1 was pressing them onto a circle of pins in a large aluminum pulley.

Large aluminum pulley? Perfect! I popped the other wheel off on a press and mounted the new wheel. Then I ran some high speed steel into the plastic spokes at the diameter of the rim, stopping just short of breaking the rim off so I could snap the core out and avoid possibly getting the rim trapped in between big steel spinning things.

The stator mount designed for the motor is a dished-in piece of aluminum with stubs on it to seat the bearings. However, I decided that it would be much easier to make the dish separate, then press the shaft through. It would involve less wrangling and maneuvering of the tool bit and simplify the process.

The shaft was the first piece I actually made on the lathe (to precision – ripping the spokes out of a plastic wheel isn’t particularly clean). This is a CNC-only lathe that has meta-handles to let you jog the axes. Maximum resolution is .0001 on fine stepping, which was really awesome.

Unfortunately, I can no longer bump the tool into the material to set a zero, because… well, the handles aren’t force-feedback. I had to eyeball tool contact, but could otherwise hold any dimension I wanted down to .0001 +/- the deformation of tool and piece. It turned out well enough for pure eyeballing of the reference zero – .590 was actually .587, but Loctite fills spaces.

After a night’s work, the pieces. I was halfway done with machining the dish part when I realized that I had probably outstayed my welcome – it was almost midnight!

Notice the ring inset into the wheel. This is the “can” part of the motor which will support the magnets.

In order to not have to deal with aligning the finished motor parts on a vise to drill out all 16 perimeter holes 3 separate times, thick aluminum round pieces that have the holes already piloted were cut on a waterjet. That way I could trim them to taste on the lathe. All dimensions on the ring were made .125″ oversize. The two thick blocks under the wheel will eventually become the side plates/bearing mounts, and will be chrome plated for pimpness.

Not.

Here’s one result of misjudging zeros by a bit. I thought I had lined up with the inner part of the dish, but actually ran the unforgiving Z-axis slightly into the material . Hence zero was a few hundredths off. Nothing life-threatening, but I had to stop and re-eyeball it again. This small part would be better done on a smaller manual machine since it requires smaller tools and more improper machining techniques than the CNC allows.

Merry wtfkthxgiving and a Snuffles Reloaded update!

Nov 22, 2007 in Project Build Reports, Project RazEr, Stuff

Mecha-Turkey says hello.

I need to modify one of those 'Easy' buttons to instead announce 'HEAD SHOT'

I’m taking a target shooting class (Yes, Charles with guns. Run away quickly.) , and this was just something fun done before the wtfkthxgiving holidays. Unfortunately, I missed the head shot, but hit the H in “shot” dead on. This was under a half inch tall from 50 feet away and I’m not particularly steady. Mecha-turkey would have mowed me down with its minigun and rocket launcher.

In somewhat related news, I’m going to take the opportunity of missing the opportunity to fly back to Atlanta and get some work done on the wheelmotor. These stupid “class” things really get in the way of building cool stuff. As I had some spare time but no materials for most of the past few weeks, I did some more design work to optimize and simplify some things.

16 gauge wire sculptures, anyone? A box of McMaster stuff. Some of the Media Lab guys are starting to take note of the work I’m doing, and so have extended some lifelines, so to speak. The 3.5″ aluminum round is for the motor body parts. The wheel will be supported by the 15mm bearings.

I just now realized how ungodly huge 16 gauge magnet wire really is. Not that it won’t work – there’s plenty of margin around the stator – but I’m going to have no hands left after winding 30 stator poles using it. I’m not sure how this guy does it with monster wire like this, but… good freakin’ robot Jesus. The ML has miles of 20 gauge I might just end up paralleling to wind the first motor. It’s a semi-prototype anyway, and those few efficiency points I might lose because of wire choice – oh well.

No, sorry, no multilink setup here.

Something else I fiddled around with was the idea of adding rear suspension to the A3. As much as the front suspensions on the scooters try to absorb the impacts of sidewalks and bumps, the rear is stiff and unforgiving. It didn’t help that all my electronics were located right on the rear wheel centerline before, and so every bump in the street was a major shock.

If all of this actually works, I'm going to be all giddy and Japanese schoolgirl-like.

This little module will slide and mount inside the aluminum T-tube that forms the chassis of the A3. I particularly like the design of the A3, because you can mount and number of things just by sliding them onto the channel and locking with screws from the side. I’ll have to trim the metal around the back wheel a bit, but that’s no issue.

By examination, the “shock towers” sort of interfere with operation of the rear fender brake, but I’ll actually have to build it to see those tolerances.

The sliding block will be brass, bronze, or some other metal better at being a bearing than aluminum, which tends to gall and gunk everything up. It will replace the entire axle spacer assembly.

Springs will be some small die springs from McMaster, rated to 250 pounds per inch of throw. I’ll only have about half an inch of travel because of the springs’ solid length, but that’s enough. At a total of 500 pounds per inch of travel for two springs in parallel, it should not be too floppy. 500 sounds like alot, but you can easily exert that much force by jumping, and a good bump will certainly exceed that instantaneously.

So it looks like if I’m building anything, I’m building the whole thing. The ML guys are interested in seeing the motor itself first, though, so that’s number one on the priority list. I’m also using the same “slide-in module” design for the electronics bay and battery pack. It’s quite handy.

Bot on!