More Testing - 3D Printed Gears

I finally needed to print a functional gear. Driving my RC car in the snow was a bit too much for the main reduction gear (My 3D printed A-arms seemed to hold up fine this time). Snow got into the gear (due to a very poor design that left the gear exposed) and was turned to ice by compression from the pinion gear. The built up ice seemed to push the motor out of the correct meshing distance. The motor then proceeded to grind away most of the teeth.

The snow was deeper than this in most spots. I took this picture after the car had already stripped the teeth from the main reduction gear.

I put the car in the bathroom to wait for the snow to melt. The electronics avoided the water for the most part, so the only damage  done to the car was the stripped gear.

This is the slot in the bottom of the chassis that leaves the gear exposed. The replacement gear is already in the car at this point. I managed to get a face full of gear teeth after a second or two of run time. I didn't have the car on the ground, so the forces on the gear teeth shouldn't have been too high. The gear teeth seem to have shattered. I think this is due to warpage in the gear after it originally printed. The portion of the gear teeth touching the support material had significant warpage. I found this was a problem in my early gear tests, but I decided to try running the gear anyways. I increased the spacing between the pinion and the reduction gear to compensate for the wapage and prevent binding. I also post-cured the gear, which increases the strength and hardness. The post-cure also made the gear more brittle which would explain the shattered teeth.

It's hard to get a good image of the gear without removing it from the car which takes a while. I didn't want to take apart the car until I had a suitable replacement ready to test.

I tried printing with my black resin instead of the clear resin. I had hoped the black resin would have less warpage than the clear resin.

The gear in the bottom left is the original gear. The bottom right is my new replacement gear printed in the black resin. The top two gears are spares that I printed, but didn't release from the support material.

I used a machinists square to visualize the warpage on the gear teeth. The majority of the gear has gear teeth that are square to the faces of the gear, however a small portion (~20%) has slanted teeth like these. This makes it impossible to have a proper gear spacing without binding. I will keep trying different orientations and possibly different gear geometry to avoid this waparge, but for now I may have to order a replacement set of gears for my RC Car.

Test - 3D Printed Gears

One of my goals for the 3D Printer was fast and cheap production of gears. Buying plastic gears can be expensive and they often require modifications. This can waste a ton of time. In addition, they might not be the exact number of teeth required for a project. I decided to test the 3D printer and determine what settings and orientations were best for high quality parts.

I started off by printing some 64 pitch gears. I was curious to see if the printer was capable of making these parts. The results were pleasantly surprising! All of these close up shots were taken with my phone. I hacked together a macro lens using a thumb sized magnifier. I taped the lens over my camera and got a 10x zoom! The only down side is there is a fair amount of distortion over the images. However, it reveals details that the camera or my own eyes couldn't see alone.

This is one of my first test prints of a 64 pitch gear. There is some surface roughness that can be seen on the gear, but all of the teeth seem to have a reasonable profile. This gear was printed in the coarsest layer height (0.1mm per layer). The layers don't seem to create "steps" in the teeth.The surface roughness is still pretty good even though the layers are visible

This is a picture that just looked cool. I have no idea what happened with the focus of the camera but it added some neat effects. This gear was printed on the highest layer resolution (0.025mm per layer). Here the layers can't be seen. It just looks slightly opaque.

Here is a side by side comparison between the course and fine layer resolutions (0.1mm and 0.025mm per layer respectively). Both gears seem to have reasonable tooth profiles. It is hard to tell whether there is warpage in the teeth as the lens created considerable distortion in the picture. The lens also created a very shallow depth of field so only portions of the gear are in focus at once.

I also worked on larger gears. The majority of these gears are 32 pitch - 40 tooth gears. I wanted to see how different orientations of the gears in the printer would cause different amounts of warpage on  the overall shape. I also wanted to see how various levels of post curing reduced wear on the gears. I also printed a set of 20 pitch - 25 tooth gears. All of the gears meshed fine, however some warpage in the gears led to noticeable wobble on the shafts. I found increasing the resolution and adding more supports during the printing helped to mostly eliminate these issues. I'll have some projects coming soon that will be built off of these gears. There is also a cluster gear in this photo that contains a timing belt pulley in addition to a 32 pitch gear. This part couldn't be made through traditional machining techniques, so I'm very excited to see what unique structures I can design and build with the 3D printer.

Servo Spline Adapters

My transmission design for the Mini Combat Robot uses large servo gears. Unfortunately the output from these gears is a spline shaft. I had to come up with a way to easily adapt to the spline without making a sketchy connection with the servo horns that are meant for those servo gears. My first technique for making a servo spline can be done with a milling machine. It's a fairly straight forward and doesn't require special tools.

I put a "blank" shaft into a collet block. The collet block isn't required, however holding a round part vertically in a vise can be difficult to align. It also has a good chance of slipping which could break tools and ruin the part. The first thing I did was drill a set of starter holes. I used a small carbide ball end mill because it was the only tool that could make a mark smaller than the drill bit I wanted to use. Each hole corresponds to one of the teeth in the spline. I designed the holes so the outer edge would meet up with the tip of each spline tooth.

Once I made all of the starter dents, I proceeded to use the final size drill bit. Each hole was very close, but none of the holes intersected. If the holes intersect the drill bit will likely drift and break. The final step was to mill out the center. Milling out the center creates the inner part of the spline. It is important to design the geometry such that the leftover wall between the drilled holes fits between the spline teeth. I simply plunged and endmill down to the desired depth and let it swirl around to the correct diameter. This opened all of the holes drilled to the center.

This is the final shaft. The part fit snugly on the spline and didn't seem to damage the spline even under loading conditions. I used this part on the Mini Combat Robot until a design iteration forced me to a 3D printed design (I no longer had access to a mill). Any mill with CNC or even a digital readout can produce splined holes.

This is a 3D Printed replacement for the metal adapter. I needed a different pulley, but I couldn't use the metal spline shaft for the upgraded design. I decided to 3D print an adapter instead. I used the same geometry as the metal spline and simply printed a new adapter. It also slipped right on and worked first try.

Here is the 3D printed servo adapter as well as a splined shaft that copied the original servo spline. I doubt an FDM type 3D printer could produce the details required for this spline to work, so I'm glad I went with the SLA type 3D printer (Form1+). Making giant servo splines led me to test miniature servo splines.

The first part to my process to make a servo spline is getting a picture of the spline itself. I use this picture and one reference dimension (the outside diameter of the spline) to trace the spline profile. This seems to be pretty reliable and is able to get details that my calipers can't measure.

Here is a screen shot from the CAD I used for the servo spline. I get the spline dimensions by tracing the profile that comes from the drawing. It is surprisingly fast to CAD this way. The camera image reveals a lot of details that my measuring instruments won't capture.

Here is a servo spline made for a micro size servo (9g servo). The printer was able to handle the small details required to make the spline.

The gear fit perfectly onto the servo first try. I attempted to strip the spline, however I only managed to cut myself with the 3D printed gear teeth. I was unable to get the spline to skip on the servo.

For future projects using large servos I will attempt to make more metal shafts and splines. I find them more durable than the 3D printed parts. I also know the servo gears will strip before the metal spline slips. The 3D printed splines for smaller servos are too awesome. I'm still amazed the printer can handle details like that. The cool part is these splines can be put into any 3D printed part. It doesn't matter whether the part is round, square, or even a hexapod leg!

New Machine - Form1+ 3D Printer

After thinking it over for a few months I decided the usual FDM type 3D printer I was building (the kind that squirts out plastic like a hot glue gun) wasn't going to meet my needs. I just moved and no longer have access to mills, lathes, or other metal working equipment aside from my trusty file and hand drill. I decided a 3D printer was the only way to continue my projects without making a mess in my apartment. After looking around at the different available 3D printers I came across the Formlabs Form1+ printer. This printer uses SLA technology. SLA printers use a photosensitive resin that solidifies when exposed to a light source. The resin in this printer cures with a UV laser.

I need a printer than can make parts with fine details, high accuracy, and decent strength. Although I'd love a printer that could make metal parts, however I'm a few orders of magnitude away from being able to afford one. I decided to settle with a printer that makes plastic parts. The FDM printers (like the one I am/was building) are nice because they have become very popular. The popularity has driven down the material costs which makes the printers very affordable. There is also a large number of users who have lots of great tips and tricks for getting high quality parts. Unfortunately the nozzle design in an FDM pritner limits the minimum features size. The fusion of layers through heat also tends to leave voids which causes very different properties depending on a part's orientation in the printer. This can make "engineering" grade parts difficult to make.

The SLA printers have only recently become available to hobbyists, so there is not a very large number of users. The material is somewhat proprietary, so it costs 3-5 times more than an FDM printer for a given volume of material. The quality from the SLA printers can't be beat by the FDM printers. The laser curing technique is able to make much finer details in the parts. I expect to be printing gears that are 32 and maybe even 64 pitch. The parts are also solid, so the material properties are going to be the same regardless of the part orientation. There should also be fewer voids in the parts, which will make "engineering" grade parts easier to produce.

Here's the printer in all its glory. I personally find the printer aesthetically pleasing. It also fits well on my bookshelf.

Here's the setup for my first prints. These parts are shock spring mount replacements for my rc car. The original ones no longer fit properly due to my machined replacement parts not being the exact same geometry as the original components.

The printer readout is pretty cool. It tells me the remaining print time and the number of completed layers. The remaining time is exact and can help me schedule my time around the printer so I can be as productive as possible.

This printer is upside down compared to FDM printers. The table rises out of the resin to grow the parts. It's a pretty neat concept and seems to work well.

Here are the final parts after removing the excess resin with isopropyl alcohol. They came out perfectly first try. I'm sure more complicated parts or parts with tighter tolerances will require some test parts, but parts like these take nearly zero effort to make.

Overall I'm pretty happy with the printer. I have more project ideas than ever now that I can make parts without labor. I'll be running test parts for things like gears and press-fits. For now I'm not going to finish the 3D printer I started making. The FDM printers can't make the parts I want to make. It is a good XYZ CNC platform, so I should be able to use it for another project some time in the future.

New Project - Mini Combat Robot

For a number of reasons we (the combat robot club at my school) have started work on a mini combat robot competition that will take place on campus. This summer we've worked on getting the competition ready for the coming school year. The first thing I started with was the drivetrain. I wanted to avoid making all of the gears as I have for many of my other motorized projects, so I looked for a cheap and easy to purchase solution. I found large RC vehicle (car, plane, boat, etc...) servos have very cheap replacement gear sets. I bought two sets of the largest common servo I could find. This gear set works with the HITEC 805 servo series.

All of my motorized projects now use brushless motors. They deliver the most power in the smallest package. They are also very cheap compared to other motors. These little transmissions use a set of 220 watt motors. Each motor is about 20mm in diameter and 40mm long. The only down side to these small motors is they spin very quickly which requires a large gear ratio to make them useful. Thankfully the servo gears make this gear ratio easy to obtain. I designed the maximum output torque of the transmission not to exceed the original servo's rated torque. This should prevent the gears from immediately stripping. Wear may become an issue because of the increased gear speed, but I'm not too worried about that because they're a cheap and easy to replace solution.

Since these transmissions are for a mini combat robot, the screws also have to be small. I had quite a few #2-56 holes to tap. As always I was afraid of breaking the tap, but in the end I tapped 24 holes without incident.

Dowel pins are used as a bearing for the intermediate gears. Everything except for the #2-56 screw is metric in this transmission (I didn't want to buy metric taps and screws when I already have a large stock of English taps and screws). The output gear takes ball bearings, which should make this transmission pretty tough and smooth.

Overall the transmissions look pretty beastly. I'm waiting on pinion stock for the motor gears to come in from an order. I'll test them once I can actually transfer motion from the motors to the rest of the gears.

The next step in the mini combat robot build process is to get the motor controllers designed (yes; I'm making the motor controllers from scratch) and to finish the chassis CAD.

New Project (finished too) - Heavy Weight Combat Robot

I guess there aren't going to be any project updates for this project considering its already done. This is the first amount of free time I have had all semester... and it's finals week. At the beginning of the first semester of this school year a small group of my friends and I decided it would be a good idea for the school to have a combat robot club, with the ultimate goal of competing in the 2013 RoboGames competition. With a little bit of work we were able to obtain adequate funding from the school to do pretty much whatever design we wanted.

Originally we wanted to enter into the 110 pound middleweight competition, however after watching videos we decided this weight class was a little lame (very wrong conclusion, even the 3lb robots are scary and exciting to watch). After some group brainstorming and preliminary weight estimates for our design, we
realized the 110 pounds wasn't going to happen, so we just switch to the 220 pound heavy weight class. Looking back, I kind of regret this decision because it made for twice the work, but I'm also glad we chose this weight class because there's a fairly small group of individuals that have attempted the 220 pound robots. It also really pushed me to put myself to my limit to get this project done while maintaining my grades!

One thing we wanted to do with the design was use components that teams don't currently use because either they're new and untested, or harder to use. These three sets of new components are as follows:

-Brushless motors
-Li-Poly Batteries
-Neodymium Magnets.

The first thing we decided to work on for testing was the brushless motors. These aren't just regular brushless motors, they're RC car brushless motors. The funny thing is that most people wouldn't consider these motors because they're designed to run RC cars that are only a few pounds. In addition, the motors are very small (around 1.5" in diameter, 2"-3" in length, and around 1 pound). Compared to the motors used on most combat robots in the 110 and 220 pound weight classes, these things are toys (most 220 pound combat robots use brushed motors that weigh upwards of 10 pounds and are 4" in diameter) The reason we thought these motors could work is that they had power ratings over 1800 watts, not far off from the big motors.

We decided to make a test platform to determine whether or not these motors were actually able to put out the power they advertised.

The idea for this platform was to measure the torque output over a range of different rotation rates while the motor was provided full power. Normally this is done by connecting the new motor to an existing one that has known parameters. We couldn't do this because we didn't have a motor that could spin 40,000 rpm and handle 2-3 HP. Instead I cooked up a platform of our own that tells the torque and rpm. There is an aluminum disk bolted to the motor. A neodymium magnet is spaced very closely to the aluminum plate. The neodymium magnet is held by an axle that is supported by ball bearings. The axle transmits torque between the magnet and a lever at the end. The lever is then placed on a scale.

The setup works by the same physics that cause a magnet to fall very slowly through a copper pipe. When the motor spins the aluminum wheel, the opposing neodymium magnet generates eddy currents in the aluminum (this is going to be a bad physics explanation, but bear with me). These eddy currents generate a magnetic field that opposes the magnet. The eddy currents turn all of the output power from the motor into heat because the aluminum has electrical resistance. The torque is transferred by the eddy current's magnetic field and the neodymium magnet to the lever arm. This torque causes the arm to press down on a scale. This allows us to measure the torque. The rotation rate is found by counting how many times a black stripe on the wheel passes by a light sensor. (Note that the setup uses some of the electronics from the "anti-gravity" robot that my roommate and I were constructing at the time.)

The system showed the motor was providing similar power outputs to the specifications. One mildly dangerous thing about the test setup was the temperature of the aluminum disk. Since the fixture is 0% efficient, all of the output power goes into heating the aluminum disk. Within a few seconds of testing, the disk was well above boiling (wet paper towels hissed as if they were touching a soldering iron)

The video shows a small piece of tape on the end of the motor shaft. The motor was surprising loud. Given that we verified the power output we decided to go with the brushless motors for their higher power to weight ratio. The only down side is that they need to be geared down a ton. ~40,000 RPM is not particularly useful on a combat robot drivetrain.

The weapon for our deign is a spinner. Spinners require massive amounts of horsepower to spin up within a reasonable amount of time. We bought the massive PERM motor. it can run up to 72 volts and have a peak power output of around 34 HP.

Even with only 12 volts I almost wasn't able to hold the motor down during start-up. We scored a good deal on the motor using ebay. It was sitting in someone's garage for a few years, but it was in perfect condition.

Naturally, the robot was drawn up before machining to ensure the parts all fit together and there wouldn't be any nasty surprises along the way.

I had trouble getting the colors to look good on the full render, so here's a contour render. It makes a great desktop background. (I've been making more of these contour renders because I think they look much better.)

I'll just include a bunch of pictures of the machining process. There were lots of cool parts. Since this was a school club project, we worked to get as many people involved in the production process. I was able to see many students go from having never seen or heard of a mill to being able to operate a CNC and make perfect parts.

This is the first time I ever used a water-jet. It made production of the large plates on the robot much smoother. This part was made in January, and was one of the first parts made for the robot.

Here you can see the motor on top of the belly pan. The belly pan greatly increases the shear strength of the chassis. It is also a great mount for the electronics. It is 34" x 22" x 1/8". This also shows how massive the spinner motor is compared to the rest of the robot. The limiting factor on the robot height was the motor.

After being back at school for less than 24 hours (came back from winter break), we already cranked out the transmission plates and some of the axles. In addition we picked up the gear stock for the transmission. One drive motor and motor controller are also in the picture.

We made a will call pickup from McMaster! (You can order on a Saturday and pickup the order an hour later!) Lots of screws. The metal came from Online Metals, but we put it in the McMaster box to keep the dorm room as clean as possible.

Here's our double vice setup for the chassis side rails. There's only one vice per mill in the school shop, so we had to jack the second vice from another mill. There are 4 of these parts on the robot. Lots of tool changes without an automatic tool changer is a horrible pain.

I used a machinable collet to hold the gears. The collet was bored out to the OD of the gear stock. This way the gear teeth were guaranteed to be concentric to the bore. This was much more pleasant than making custom jaws for the 3 jaw chuck.

Here is a finished gear stock. Buying 3 gears was more expensive than buying a whole gear stock, so for the 64 tooth 32dp gears we had lots of spare material.

The cutoff tool was used to remove most of the material, but the horizontal bandsaw was used to cut each blank off of the stock. This way stock could be supported by the live center at all times while inside the lathe. The rest of the facing and boring was done with the custom bore collet.

Gear stock makes great noises...

I designed the transmission to use as many of the same tooth number gear as possible. This way I could use the gear stock to its fullest. I made a full set of spares just in case some catastrophe happened during competition.

In just about 2 weeks we managed to make it this far. This is amazing considering we all have copious amounts of homework and there is only one CNC mill and lathe. The limiting factor has really been the single CNC mill. The manual machines are in such poor condition that even facing stock turns out poorly. All milled parts on the robot have to pass through one machine.

Thankfully I didn't cause this crash. That was 1.5" diameter stock. The tool holder was a little messed up afterwards, but thankfully there was no harm to the machine.

PRO TIP: Watch where you extend your stock relative to the zero of the machine. The program was written to face off the first .1" of the stock. If the stock is extended past that by lets say.... an inch or two, the machine rapids into the part.

The crash didn't really set anything back (aside from morale). The hubs for the wheels were made in a few hours. We used 3" diameter wheels. This reduced the amount of gearing required for the transmissions. The cool thing is that the chassis is so short that the wheels still stuck up past the top.

A quick rolling test verified the chassis could roll. We added some cardboard boxes to make it look cooler... not sure if we succeeded.

I flew back home to get the rest of the water-jet parts made. Lots of 3/8" 7075 aluminum plate and 1/2" 6061 aluminum plate.

 This is one of my favorite pictures from all of the machining.

Yeah... the spinner is pretty large.

A couple hours worth of water-jet time made the biggest jump in completeness for the robot.

I got to use my insert endmill again (GMT tools are really awesome in looks, performance, and price). It's pretty boss.

These bearing blocks are pretty large compared to the mill. The bearings are 70mm ID tapered roller bearings. Considering they handle all the damage potential, they need to be pretty strong. I also realized how much the student CNC mill needs flood coolant and an enclosure. removing copious amounts of material takes forever.

The finished wheel assembly looks really awesome. The sprockets were cut on an EDM machine. All of them were cut in one stack, so it wasn't a waste of machine time.

I made chain tensioners, but hoped the chassis wouldn't need them. I lucked out and the chain was perfect. #35 chain is pretty robust and can take some slop without any problems.

This is starting to look like a real battlebot.

A quick drive test greatly boosted morale (morale was good before the test, but it was super high after the test). The robot easily carried 3 people. The acceleration skids the wheels on the floor. The massive rpm of the motor makes the gears scream. It sounds pretty mean. This also proved we didn't screw ourselves by going with the brushless motors. The next priority was to build the rest of the spinner.

The stock was only wider than the 3 parts by .03" and thicker than 2 parts by .01". We had to get creative. This was very material wasteful, but considering the original 27 lb slab of 7075 was from a metal recycling place for only $2 a pound, I was cool with machining away half of the material.

Getting closer to the final assembly. Finishing the robot in time for competition is a little more stressful that we had originally thought. The CNC mill kept needing to be used for class projects, which killed the productivity.

The spinner looks pretty menacing on the robot. Now all it needs are the hammers and tool steel blades.

This shows how the parts interlock. The hammer is locked by geometry to the arm plates. This ensures the bolts don't need to take the shear force from the impacts. The 7075 of the arm corroded some which is why it is so much darker than the 7075 used in the tie rod blocks above the hammer.

I cut the protective lexan shields for the electronics using the CNC router. It was surprisingly easy considering some of the troubles I have had with the router in the past (my longboard).

Carbon Fiber tie rods. These parts connect each of the arms to improve the overall strength. We knew carbon fiber might be a mistake, but we used it anyways because it was light and we had it. It was a mistake, but we didn't find out until competition.

The spinner looks really awesome... mostly because its shiny. This weekend was a major push because testing was scheduled for that Sunday. We got access to a gravel pit across from the school to ensure we could be a safe distance from the robot.

Remember the idea of adding neodymium magnets that I talked about at the top of this post?

Yeah they work... We were able to jump on the chassis and not have it fall from the thin sheet metal door. (The concept of adding additional down force led us to design the "anti-gravity" robot)

We also tested it on top of 1" thick steel plate at the school construction site to make sure it could drive with the additional couple hundred pounds of down force. It didn't really seem to notice the extra force, but we had trouble removing it from the ground.

Greasing the spinner bearings... YUMMY

We finally made the S7 tool steel blades.

And heat treated them too... (had to pull 4 all nighters in a week to get to this point because the robot had to be finished for testing the following day)

We also have a robot name!

Considering this was the first combat robot any of us had ever built, we needed to test it. We needed a safe place to test it where we could control who was going to be near the robot. In addition we wanted no property near it just in case there was flying debris or a catastrophic spinner disintegration. We got permission to use the gravel pit next to the school. It's a giant gravel pit that takes up a few square blocks. No debris could escape.

The tool steel blades look pretty scary. The mac was broken, so it was cool to smash. You can imagine what shape the mac was in after the impact.

The mac was eaten up pretty quickly.

I like how the CD drive ejected. It isn't visible in the pictures, but there were a number of IC chips that flew off the board. The impact G's were so high the surface mount components flew off the boards! That's pretty cool.

Needless to say, the microwave didn't stand a chance.

The microwave got smashed in one solid hit.

The tool steel spikes seem to do their intended job pretty well.

Testing was a great way to prepare for competition, mostly because it was fun.

I'll summarize the competition in another post