In early 2012 I built the first version of Nyx. After 2.5 years and over 30 fights across 6 events it was due for a major upgrade. The Instructable for the original Nyx can be found at http://www.instructables.com/id/30lb-Fighting-Robo...
With a win percentage over 70% Nyx had shown that many of the core elements worked quite well, but it had a few design elements that caused issues during events.
In addition to addressing those issues, a major goal of the design was to develop weapons that would operate in a more exciting fashion than the lifting spike.
If you're thinking about building this yourself, make sure to read through all sections of this instructable. If you have any questions, feel free to ask in the comments or via private message. I'll answer them as best I can.
Step 1: Designing the new Nyx
Once the concept was settled on the design began on the base platform in parallel with the initial lifter and axe concepts. As the initial design approached completion the crusher module was added to the list of weapons and was added to the development process. Once the platform and all three modules were generally laid out a great deal of time was spent populating hardware and making finer adjustments to the systems to ensure that they would be within the 30lb weight limit.
The initial goal was a drive platform that weighs 18 pounds and a budget of 12 pounds per weapon module. By the completion of the design the drive platform weighed approximately 17 pounds and the heaviest module weighed approximately 12.5 pounds with gave a reasonable amount of margin for wires or calculation errors.
Design of the Platform
The core goals of the drive platform were to retain the speed and agility of the original Nyx, provide a compact, light platform for the weapons to mount to, and to at least meet the same level of durability of the original Nyx.
The design of the base heavily features Nutstrip and slot & tab design elements to create a rigid and easy to assemble body. The internal sides slot into the front and rear panels and are locked down with Nutstrip at each corner. The top and bottom armor are designed to bolt directly to more nutstrip running along the chassis side rails. The design utilizes wheel guards that strongly resemble those on the original Nyx. These wheel guards both protect the wheels and provide an outboard support for the axles to reduce the chances of bent shafts causing drive issues during an event.
The Dewalt Powerdrive Kits were brought directly over from the original Nyx, however for this design they are placed at opposing corners of the chassis and directly drive one wheel with the other wheel on the same side driven via roller chain. Compared to the original Nyx this setup will provide a higher top speed, lighter weight, and reduced complexity.
The initial wheel plan was to use a pair of BaneBots 2-7/8" x 0.8" wheels at each corner, however during the build process these wheels became difficult to find in the correct durometer and they were replaced with 3" Colson caster wheels and custom hubs.
No changes were made to the battery, as it had proven to be a good solution to powering Nyx for a full length match.
Over the course of the original Nyx competing it switched from Holmes Hobbies BR-XL ESCs to the Ragebridge ESC from e0designs.com. The new build integrates two Ragebridge ESCs, one for the drive system and the other to allow quick weapon integration with adjustable current limiting which allows the electrical system to be quickly and easily tuned to suit the weapon.
The new Nyx was designed to be fabricated using three sheets of material, 1/16" thick 6al4v Titanium for the top and bottom armor, 1/4" 7075 aluminum for the main structure of both the base and weapon modules, and 1/4" AR400 steel for the weapons. Limiting the design to these three pieces meant that there was less time loading material and setting up the waterjet cutting during fabrication. It also proved to be a good way to minimize scrap, as the individual components could be laid out in a way to minimize wasted space on the plates the components were cut from.
Design of the Lifter
For the lifter I selected a DeWut gearmotor from e0designs.com. The DeWut is a Dewalt drill motor in a custom mount with a custom output shaft that makes it well suited to robot combat. The DeWut is used in low gear with two stages of 3:1 reduction added to combine for a total reduction of 460.8:1. This combined with the current limiting on the Ragebridge being set to 50A means it would theoretically be able to lift 75lbs at the tip of the 23" fork arm. This was designed to greatly exceed the 30lb weight limit of the class because in addition to reducing the strain on the gearbox and motor, it also ensures that if the opponents center of mass is further from the axle than the tip of the arm it will at least be capable of lifting its weight. An additional design element meant to address the high loads on the lifter itself was the addition of 3 1/4" thick aluminum plates to each half of the lifter arm. The arm and all three plates are keyed to better distribute the forces incurred during a lift or sudden jolt to the arm.
With this module there is also an additional set of outriggers added. These outriggers help the robot lift an opponent without tipping. Without them, if the combined center of gravity of both robots was ahead of the front wheels they would fall forward. These outriggers move that point several inches in front of the main chassis of the robot, allowing it to lift most opponents with no stability issues.
Design of the Axe
With the axe the design started as a very simple idea: Build an electric axe with an A28-150 for the weapon motor. The A28-150 regularly sees use as the weapon motor for 30-60lb spinners and as the drive motor for robots weighing as much as 220lbs. An element that's often overlooked in electric axe weapons is that you have to dump as much energy into the axe arm as possible in 180 degrees or less. This means even with a fairly powerful weapon motor you'll likely need to add substantial gearing to maximize the energy output. The ideal situation would be to have the motor reach peak RPM as it contacts the opponent, however calculating the system to that level would take a great deal of effort, so I settled on the "good enough" option of an 18:1 ratio, which with the intended hammer size put the motor up over 4000rpm by the expected point of impact.
The question that follows that is how do you keep the motor from exploding due to sudden shock loading? There's no one right answer, however the approach I took was to integrate a torque limiting clutch into the hammer arm itself. The torque clutch is rated for 60ft-lbs of torque and with the planned 140A current limit and gearing providing approximately 66ft-lbs of torque at the shaft, it makes for a great fit. Now, instead of coming to a jarring halt, the motor is able to decelerate over a relatively large period of time.
The geartrain for the axe uses the same modified gears that the lifter uses for the initial 9:1 reduction, however for the axe there is an additional 2:1 reduction stage done via roller chain. Using chain reduction allowed me to keep the initial two stages of gearing lower in the assembly, lowering the overall system center of gravity.
The axe itself uses heavy triangulation of the weight reduction areas to minimize strength loss from the removed material. The main impact tip is designed to provide a balance between piercing and blunt force. While an extremely sharp point would be better at penetrating armor, it would also dull much more quickly than the style of blade used. This shape is fairly good for piercing, but has enough material at the tip that even against more durable materials it's not prone to noticeable deformation. The more traditional axe shaped head is primarily to move more mass out to the edge of the weapon but it is also a capable chopping weapon and should it be better suited to an opponent it can be rotated toward the front by removing the drive axle and flipping the torque clutch.
Design of the Crusher
The crusher uses a pair of Gimson Robotics GLA750-S linear actuators to power the crushing spike. Each actuator is capable of delivering 500 pounds of force and can extend at 2in/s. With the linkage ratios and a current limit on the Ragebridge that puts the motors at the maximum rated power (23A, 500lbf) the force at the tip of the spike is estimated at 670 pounds. To deal with this force, hardened steel axles are used both at the back of the actuators and for the weapon axle. The rear mount points on the actuators and the ends of the output shaft have been drilled out to 3/8" and 1/4" respectively to increase the shear resistance of the shafts they connect to.
The whole weapon portion of the module is designed to float. The benefit of this arrangement is that if the bottom of the opposing robot is higher than the lowered position of the arm it will automatically lift into the opponent, providing a firm grip on the opponent and keeping all four drive wheels on the ground.
The top of the mount portion of the weapon module is designed to limit the total vertical travel of the arm to ensure that it doesn't potentially get flipped over in combat. The shape of the mount also functions as a roll cage for the actuators, protecting them from damage should the robot be flipped. This is the only attachment that isn't capable of self-righting, and as such it is not intended for use against opponents that are likely to flip the robot.
Step 2: Major Components and CAD Files
NOTE: The majority of the fasteners and hardware used in Nyx are called out by part number in the CAD files provided at the bottom of this step. Most parts were purchased through McMaster-Carr. The 3" pitch diameter gears are stock McMaster gears that have been modified to reduce the weight.
Major Drive Platform Components
2x Dewalt Powerdrive Kit -Discontinued, http://www.robotmarketplace.com/products/0-TD-RCM5... , may still be available via private sale. Alternatives include the Robot Power Magnum 775, Vex Versabox, and BaneBots P60 gearboxes with minimal modification to the design needed.
1x MS-1 Power Switch -Obsolete, MS-05 can be substituted with minimal effort (two new holes in the baseplate) http://www.robotmarketplace.com/products/TW-MS05.h...
2x Ragebridge ESC - http://e0designs.com/products/ragebridge/ and http://www.robotmarketplace.com/products/0-RAGEBRI... Note: Ragebridge v2 is in development.
1x Spektrum DX6i - http://www.robotmarketplace.com/products/0-SPMR663... This can be substituted with the 6 channel OrangeRx T-SIX transmitter - http://hobbyking.com/hobbyking/store/__54822__Oran...
1x OrangeRx R610 - Discontinued, equivalent- http://hobbyking.com/hobbyking/store/__31224__Oran...
1x Turnigy Nanotech 6s 2650mAh Lipo - http://hobbyking.com/hobbyking/store/__20756__Turn...
4x Chaos Hubs with 3" Colson wheels - http://nearchaos.net/?p=90
4x 15T 7075 aluminum #35 roller chain sprockets - http://www.andymark.com/product-p/am-0167.htm
2x Gimson Robotics GLA750-S Actuator, 100mm stroke, 18v motor - http://www.gimsonrobotics.co.uk/GLA-S_linear_actua...
1x e0designs DeWut Gearmotor - http://e0designs.com/products/dewut-3-speed-gearmo...
1x Ampflow A28-150 - http://www.robotmarketplace.com/products/0-A28-150...
Step 3: Drive Platform
The drive platform houses most of the expensive, delicate, or otherwise important components in Nyx. The top and bottom plates were waterjet cut from 1/16" thick 6al4v titanium sheet and the rest of the structure was waterjet cut from 1/4" 7075 aluminum sheet.
While waiting to get the 7075 sheet cut I took care of the drive axles. All four axles started as shafts for the Dewalt Powerdrive Kit used in the drive system, two were cut-offs from a previous build and two were new shafts. The reason for doing this is that the layout of the bot is such that only two shafts are directly driven, while the other two are linked to their corresponding drive motor via chain. With any shaft, you have to do something to prevent it from working its way out over time. Shaft collars are a common solution, but for the external sides of the shaft they would be bulky and a relatively large target. I instead opted to weld a steel flange to the shaft where the collar would have normally been located and sharpen the excess shaft material to a point. This made for a light and compact means of preventing the shafts from being forced further into the chassis. The spiked shape also makes it much more difficult for the bot to be propped up on its side, as the balance point is much smaller.
Once the plate was cut I started on assembly with the LED power indicator lights being the first step. These are high power LEDs with a resistor attached to the positive side of each LED to allow them to comfortably operate when supplied with ~24v. Both LEDs were wired in parallel, shrink wrapped, glued, and zip tied to the front rail to keep them away from any moving parts and to minimize the stresses on the solder joints.
After that I spent some time with a single flute countersink putting all of the countersinks in the frame rails for the flathead bolts that are used in the inner side rails. Flatheads are used here to avoid the bolt heads interfering with the drive chains. The bushings are pressed into the inner and outer rails with the flange of the bushing facing the wheel and sprocket location so they won't fall out even when experiencing high shock loads.
Following that, the nutstrip for the side rails is added and the aluminum standoffs that set the rail width are installed to create an assembled drive pod. (Note: When sliding the wheels/sprockets on, it is easiest to remove the outer wheel guard with the shafts still installed in it and slide the wheels and sprockets onto the shaft)
The drive pods, front panel, and rear panel were then assembled and attached to the baseplate to ensure everything was fitting properly.
The next step was transitioning the electrical system from the old Nyx to the new Nyx. Since anyone trying to replicate the project won't have an already completed electrical system, I've provided a diagram showing how the bot is wired so you'll have some reference material. For the drive motors I used 4mm bullet connectors to connect them to their Ragebridge. The Ragebridges and power lights are connected to the power switch and ground via Deans connectors. The battery itself also uses Deans connectors. The weapon motors were checked and wired such that they'll also connect to their Ragebridge via Deans connectors. The polarization of deans connectors means that once you've got it right you won't have to worry about hooking things up backwards in the future. As far as male/female connectors are concerned, my standard approach is to have whichever connector is electrically closer to the battery be the female side. This dramatically reduces the chances of shorting the electrical system.
With all the internals installed the bot was ready for its first test and performed flawlessly, meaning it was ready for weapons.
Step 4: Crusher
From an assembly standpoint, the crusher module is by far the simplest.
The steel parts were cut prior before the aluminum parts, so assembly of the main jaw was able to begin fairly early in the process. The main spike was made by welding two 1/4" thick cutouts together with a tig welder. This could be cut from 1/2" thick material if preferred, as the choice of two 1/4" plates was simply to minimize the number of waterjet sessions needed for all of the components. The spike is bolted to two side plates, each of which have a bushing pressed into them and a shaft running through the two plates.
The actuators need minor modification, as the rear mount points need to be bored to 3/8" to accommodate the shaft and the holes on the shaft of the actuator need to be bored out to 1/4" to fit the shoulder bolt that's acting as a pin.
The mount shaft was inserted into the actuators and the two side rails were slid onto the actuator shaft and spike shaft with threaded standoffs used to set spacing. The actuators were then bolted to the top of the spike mount plates and thrust washers were used to make sure it all fit together snugly.
Once the aluminum parts were cut the two side plates had their nutstrip attached and were slid onto the actuator shaft and bolted together using a pair of standoffs, completing the assembly of the module.
Each motor is limited to ~25A with the Ragebridge to minimize the chances of the actuators shredding their own gears. Even at this relatively low current they'll put out a combined 1000lbf and can nearly push the spike through 1/8" aluminum.
Step 5: Lifter
The lifter mechanism was next on the list. It uses two stages of 3:1 reduction using modified 16 pitch gears from McMaster-Carr to connect the DeWut gearmotor to the lifter shaft.
Torque is transmitted to the lifting arm via 2" of keyway that is provided by sandwiching the steel arms with several 1/4" 7075 adapter plates. Both the arms and plates were cut on a waterjet with the keyway included in the profile. Due to the nature of waterjet cutting the cutouts did need to have the keyways finished to provide proper corners. This was done with a two ton arbor press and a broach. Bolting each half together prior to broaching ensured that the final keyway profile would be aligned across the entire assembly.
The first part bolted into the aluminum structure was the DeWut gear motor which then had the gear, keystock, and shaft collar installed. The 3" diameter, 48 tooth gear and shaft collar were attached to the far side of the intermediate shaft and were slid into place. While sliding the shaft through the bushing the shaft collars, thrust washer, and gear were slid onto the shaft while adjacent to the gearbox portion of the DeWut gearmotor.
Following that the rear shaft was inserted into the same plate and each piece of the lifter arm and gear was slid on with its keystock. 1/16" and 1/8" washers were used to set spacing. Once everything was installed and the spacers for the lifting arm were attached the other plate was added and a shaft collar was added to the outside of it to prevent it from falling off.
In testing the lifter has proven to be easily able to lift another 30lb robot when using the outriggers to support the shifting center of gravity.
Step 6: Axe
The axe was the most complex of the three weapons and as such, it was saved for last. With three stages of reduction and a torque limiting clutch it's also the heaviest of the three modules.
The first step was to attach the nutstrip and standoffs (Note: The urethane bumper for the rear of the assembly was notched to create clearance between it and the gear) to the two frame members. After that, the A28-150 weapon motor was inserted and bolted in place. The gear, key, and shaft collar were then installed.
Like with the lifter, the 48 tooth gear and shaft collar were installed first and slid into place. The positioning of the weapon motor meant that the collar, spacers, and 16 tooth gear had to be installed as the shaft was being slid through the bushing.
The second shaft was inserted in a similar fashion with the gear, spacer and collars being added. With this module this shaft also has a sprocket with additional spacers and another shaft collar that make up one half of the third stage of gearing.
After this, the hole for the chain tensioner was tapped and the tensioner was bolted into place.
The weapon shaft was inserted from the right side through the torque clutch which had already had the axe arm installed and tightened to a reasonable degree, then was forced out far enough that it was able to accept the sprocket and shaft collar. The sprocket and collar were then attached and a length of #35 roller chain was added to complete the assembly.
In testing the axe has proven to be quite effective, putting sizable divots and severely deforming 1/8" aluminum plate.
Step 7: Finishing Touches and Final Notes
Most events you'll attend will require some sort of weapon lock for active weapon systems. For the crusher, this can easily be handled with a large chunk of wood wedged between the spike and the forks. For the axe and lifter, there is already a weapon lock point provided in the design. The aluminum rails for each of these weapons have been designed to have a 1/4" OD, 1/4" thick magnet glued into them and will accept a 1/4" steel rod as a locking pin. The steel rod should either be bent such that it will contact the magnet or it should have a shaft collar or similar steel object added to it that will allow the magnet to securely hold it in place. This will reduce the chances of it falling out in transit but allows for quick and easy removal and installation.
The top armor has a mount point that will allow the addition of a camera should you desire it. The mount pictured here uses 1" UHMW round stock to absorb the vibration inherent in a machine with no suspension and has a safety cage made from 1/8" thick flat stock and 1/4" round stock which should provide a degree of protection from other robots while not taking up much space/weight. The mount position can be seen in one of the pictures. After initial fabrication I found that the mount was prone to rotating, so I added a second small hole and another fastener to keep it from rotating. When screwing into UHMW I recommend "Plastite" fasteners.
Fixing Tight Bushings:
Depending on the tolerances held during fabrication, pressing, etc... you may find that the bushings are quite snug. I had several with the same issue during the build. This can be dealt with first by drilling out the bushings to a true 1/2" (if they are pressed into undersized holes they will compress slightly and this clears that out fairly well) and then following that up by inserting the actual shaft you'll be using into the bushing and chucking it up in a drill (with the panel clamped to something firmly) and running the drill for a few minutes. This both lubricates the shaft/bore and ensures that the bore is a perfect fit for the shaft.
I would also like to give thanks to the sponsors of Near Chaos Robotics:
Pelican ( http://www.pelican.com )
FingerTech Robotics ( http://www.fingertechrobotics.com/ )
Equals Zero Designs ( http://e0designs.com/ )
Kitbots ( http://www.kitbots.com/ )
Step 8: Testing