Dane.Kouttron

[2.16.14] Atomic Bumble Prime [BATTLEBOTS]

What? This project documents the build of a 30lb featherweight battlebot for motorama 2014 in a relatively short timespan. 'Atomic Bumble Prime' was born as an upgrade from its older brother 'Atomic Bumble', my first foray into battlebots. Prime had some interesting design decisions, namely for the sake of time. Some worked better than expected, some, not so much. Build log follows below.

What?

Demo Video

CAD

Shell Assembly

Base Plate Build

Propulsion

Bill of Materials

War Wounds

Image Directory

First Spoolup Test Video

This was the first shell spinup at ~2330 on Thursday 2/13/2014, approximately 40 hours before the competition. Chaos was at hand, paint was barely dry and wires were merely zip-tied in place, yet the bots shell spun-up to 1/4 speed without any significant oscillation or perturbation. Unbalance on a large shell can cause a bit of havoc and potentially vibrate things apart.  This spinup was powered by a 4S lipo pack and 100A hobbyking ESC, with a linear regulator providing BEC supply to the receiver.

What is it that I actually want to accomplish here?

Upgrades from the original atomic-bumble: Finish before competition, not at the competition Have working propulsion Move shell-spinner contact point from outer-rim to top of rim assembly (less potential for impact deformation) Enclose shell to prevent penetration to inside.  Avoid aluminum, steel is far easier to weld Robot render from photoview360 (solidworks)

Background

'Prime's build  was put on hold to complete a rather extensive oral and written qualification exam, which, ended roughly 5 days before motorama 2014. It happened again! A week before competition and the robot has yet to be born! with some very late evenings, an overnight mcmaster-order and a hail-mary, last minute bearing distrubitor, the robot came into existence. I was fortunate enough to head to competition with the ever-fantastic Charles G, Bercu, Jamo and Cynthia, who supplied some fantastic last-minute ideas as well as incredible abilities to weld/bond things using miscellaneous tools in unknown states.

The haistly thrown together CAD model

Solidworks
The Shell The shell consists of six identical waterjet-cut pieces of 1/8" thick stamped steel diamond plate, 1/4" thick plate mounts and a 1/8" thick steel coverplate. The top of the shell houses a large bearing assembly and an innner-race track that functions as the friction drive for the shell spinner. The bearing is a encased in three waterjet housings, and mechanically bolted to the shell assembly. Six 10-24 bolts tapped to the shell pull the assembly together, encasing the bearing.
The Frame The frame is fairly meager, it consists of a baseplate, the central bolt-shaft mount, propulsion mounts and the ring-spinner drive motor. Shown right is the shell innards hovering above the baseplate along with the spring-tensioner assembly.
Spring Tensioner & bearing assembly The spring tensioner assembly consists of a large bolt welded to the frame and a central bearing, the bearing is trapped between the shell and the bearing holsers. The spring presses down on the center of the bearing, which couples the load through the bearing onto the shell. As a result, increasing tension on the spring forces more contact between the shell spinner motor and the bearing. A second nut is used to keep the nut in the same position.
Spring tensioner exploded view The tensioner shown above is shown (right) disassembled. This disassembled view shows the bearing holster, the spring loading points and the load bearing itself.
Shell spinner racetrack The racetrack lives inside the shell and is fabricated from 1/8" steel plate. The plates were intended to be pressed into the shell assembly and then eventiulally tack-welded in place.

   
Shell Build

Waterjets and welders ahoy
OMAX I was fortunate to have shop access to the wonderful waterjet over at MIT's hobbyshop. Its an early generation cantilever waterjet, but makes mincemeat of 1/4" steel plate.  To save time I pre-computed trajectories for the bot cuts and was able to complete all the cuts (shell skirt plates, top plate, inner braces, bearing holsters, over three 1 hour job timeslots. While 3 hours seems like a lot there is a bit of setup involved and
The inside of the exoskeleton The shell has six braces mounted in slotted holes on the top-plate and mounted to the skirt plates via T-nut bolts Welds were used to bond the two together. Initially i had wanted to leave the skirt plates bolted in place, for balancing, then weld, but due to time-constraints, welding came first. The plates were mig-welded together and to the support skeleton. Copious amounts of MIG welds pulled everything together. Ivan Vanko would be proud.
The racetrack The skeletal supports contained two mounting points for the racetrack assembly. These mounting points align the racetrack. Before welding, there were some minor hacksaw-cuts required to get everything in as snug-ly as possible.
Placement and welds After a final test-fit, the racetrack was haistly tack welded in place. The welds were intended to not breech the surface of the track as this would create stress spots for the rollers and bearings.
Adding a rubber footing I used a propane torch to super-heat the racetrack assembly, and then draped  'miters sparkly hot glue' around the track. The specific heat of the steel allowed for the glue to melt in and form a fairly uniform layer. While hot, and with the help of Peter Krogen, I stretched very heavily used Blue ESD mat over the track assembly. A large flywheel weigh was placed on top of the mat to help it cool in a flat fashion. Afterwards an x-acto knife was used to cut the mat into racetrack shape.
Cooling Time It took a fairly long time for the steel plate to cool. Afterwards the mat was surprisingly well bonded to the plate assembly. The mat provides a rubbery grip surface that the shell motor spins on. This worked far better than expected, as the rubber mat bonded well to the racetrack assembly and was a high enough durometer to not fall apart under high load (shell spoolup)
Paint As much as i enjoyed the rusty delerict steel look i decided a coat of mate black paint on the shell would add a bit of legitimacy to this haistly thrown together project. The letters DK were added on during the shell welding session, and ground flat after painting to help the letters pop-out. I used a self-etching primer (gray) coat and then followed it with a matte black coat of paint.

   
Baseplate Build

Buildtime
Waterjet The baseplate is waterjet 1/8" thick steel plate, the hole mounts are for the central bolt for the spinner shaft, the two wheels and the rear rollerball. Shown in each picture is the monsterous steel bolt, used as the central shaft A screwdriver is used for scale. I chose steel as its readily weldable.
WELD The ring and baseplate were mig-welded to the central bolt to give a fairly sturdy mount. The welding started with four tack-welds around the bolt-baseplate and then sewing sequential bits of weld to prevent uneven thermal warping / loading on the baseplate. The ring plate was welded in place to add in structural support of the central bolt.
Adding Accouterments the bearings that provide the two other mount points are mounted with pegs and secured with T-Nuts. The bearing is bolted on with an M5 Bolt and lock-nuts. The height of the bearing was pretty crucial, too low and it would make the spinning shell lopsided and remove tension from the spinner motor.

Propulsion

Buildtime

The motor control for the steppers and shell spinner is tied into the sporadic motor controls / shell spinner control setup, shown right. Everything is fed from a 4S lipo brick, and either regulated down for logic or run hard for the motors. An XT-60 connector is used as the 'on-off' switch, it disconnects the battery supply from everything else. The arduino nano receives PWM input from two of the 6 channel receiver, parses the pwms for each channel, and then modulates the stepper motors accordingly. This provides front-back / left-right control of the robot. The radio potentiometer sets the shell spinner speed, which is translated into a PWM used to control the shell spinner 3 phase drive

Rear roller wheel The rear wheel, detailed below in the BOM, was epoxy-glued in place in a pre-cut hole designed for it to sit in. the whole vehicle now was the two front wheels an a rear roller assembly. This makes turning fairly difficult as the whole bot wants to pivot around the center of the 2 wheel assembly, in tank-steering state.
last minute stepper upgrade Due to some issues burning out the initial steppers, radical changes were made, the mistakenly ordered 5:1 gearboxes were shoe-horned in place and epoxy/welded to the frame. Parts of the frame were removed to accomidate the increased mass of the motors and provide some clearance. Copious amounts of 2-part epoxy were used to keep everything in place. After some twiddling THERE WAS PROPULSION! It was slow and the stepper drivers were fairly anemic, but it was propulsion nonetheless.

   

Code

Buildtime

This was drafted on the car-ride to motorama. Due to a hardware issue, (I had blown out the FTDI usb-serial peripheral on the arduino nano), I was using an AVRISPMKII programmer. There was an issue with that route. I had 'arduino nano 3.3v 8mhz' selected as the compilation target, which, tells the on board ATMEGA328 to use its internal RC oscillator instead of its on board 16mhz crystal. As a result the timing is off by a factor of two :/ This was found by Aaron Fan, I was banging my head into the table trying to figure out this issue right before safety qualification occurred. Aaron saved the day and cleaned up a bit of my case-statement mess. CHECK YOUR FUSEBITS AND PROGRAMMING TARGETS KIDS. there's a few issues with the way this is implemented. It takes time to poll the RC receiver pins (2), and because the timing for the steppers is software generated, it results in a bit of a tradeoff between latency and max stepper speed. Ideally this should be implemented with interrupts and a counter to keep track of the input ppm signal from the receiver.

/* DKLABS AWESOME ROBOT PWM DO-DAD this is what was used on atomic bumble prime, note that the clock speed was 8mhz whereas the code expects 16mhz, so everytiming is off by a factor of 2 This example code is in the public domain. */ // These constants won't change. They're used to give names // to the pins used: int sensorValue = 0; // value read from the pot int outputValue = 0; // value output to the PWM (analog out) int ch1; // Value for channel values int ch2; int x; int ch1_v_forward; int ch1_v_reverse; int ch2_v_forward; int ch2_v_reverse; void setup() { pinMode(2, INPUT); // Set our input pins as such pinMode(4, INPUT); pinMode(7, INPUT); pinMode(13, OUTPUT); //-------------------- pinMode(8, OUTPUT); // stepper 1 direction pinMode(6, OUTPUT); // stepper 1 step pinMode(9, OUTPUT); // stepper 2 step pinMode(12, OUTPUT); // stepper 2 direction } void loop() { //digitalWrite(13, HIGH); // set the direction //delay(200); // wait //digitalWrite(13, LOW); // set the direction //delay(200); // wait ch1 = pulseIn(2, HIGH, 20000); // Read the pulse width of ch2 = pulseIn(4, HIGH, 20000); // each channel if (ch1 == 0 || ch2 == 0) return; // channel 1***************************** // channel 1 reverse is 1100 (full reverse) to 1300 (neutral) // channel 1 forward is 1900 (full forward) to 1300 (neutral) int delta = abs(ch1 - 775); if (delta > 300) delta = 300; int step_delay = map(delta, 300, 0, 150, 1400); for (int i = 0; i < 350; i++) { if (ch1 > 775 ) digitalWrite(8, LOW); if (ch1 < 775) digitalWrite(8, HIGH); if (ch2 > 775) digitalWrite(12, HIGH); if (ch2 < 775) digitalWrite(12, LOW); //delayMicroseconds(4); digitalWrite(6, HIGH); digitalWrite(9, HIGH); delayMicroseconds(step_delay); digitalWrite(6, LOW); digitalWrite(9, LOW); delayMicroseconds(4); } }

   
Getting ready for battle

Buildtime
Its becoming a real thing With the inner assembly and shell held together, it became apparent fairly quickly how frigging large this beast was becoming. The weight was starting to be a concern. A mass of the thing came to 29.98 lbs using the 'Charles g. BIG SCALE of science' that lives in the IDC. At this point, the weight limit was a continued point of focus, keeping everything inside of 30lbs was a bit of a headache.
The scale of the machine Its surprisingly thin and long for something kludged together. So all the motor and battery is crammed into a tiny space with little clearance. EXCELLENT.
It also makes an excellent hat / air guitar audience <Yes that's electrical tape holding a bandage in place>

  
Bill of Materials

Description	Part number / link	Description
Mondo Washer	91201A039	Plain Steel Extra-thick Flat Washer, 1-1/8" Screw Size, 2-1/4" Od, .3"-.33" Thick
Main compression spring for	9657K409	Steel Compression Spring, Zinc-pltd Music Wire,2.00"l,1.460"od,.135" Wire
	91268A125	High-strength Steel Cap Screw - Grade 8, 1-1/8"-7 Thread, 6-1/2" Long
	90502A430	Plain Grade 2 Steel Extra-wide Thin Hex Nut, 1-1/8"-7 Thread Size, 1-13/16"w, 39/64"h
	91290A252	Black Class 12.9 Socket Head Cap Screw, Alloy Steel, M5 Thread, 25mm Length, 0.80mm Pitch
Rear ball roller wheel	2415T13	Countersunk Flange-mount Ball Transfer, General Purpose, 5/8"stl Ball, Stl Housing, 20#cap
	90576A104	Metric Zinc-plated Steel Nylon-insert Locknut, Class 8, M5 Sz, .8mm Pitch, 8mm Width, 5mm Height
	92703A353	18-8 Ss Flat Head Torx Machine Screw, 10-24 Thread, 3/8" Length
	94855A123	Grade 2 Steel Square Nut, Plain, 10-24 Thread Size, 3/8" W, 1/8" H
Secondary shell spinner roller bearings	6384K52	Steel Ball Bearing, Plain Double Shielded For 1/4" Shaft Dia, 11/16"od
Primary shell spinner bearing	5709K22	Steel Tapered-roller Bearing, Roller Assembly For 1-1/8" Shaft Diameter, #15113
A36 1/4" thick steel plate	Link	ASTM A36 Hot Rolled Steel Plate (shell braces)
A36 1/8" thick steel diamondplate	Link	ASTM A786 Hot Rolled Steel Floor Plate Print Print (shell top, shell skirt plates)
Hobbyking shell spinner motor	Link	Kv (rpm/v)     350, Weight (g)     266, Max Current (A)     53. Power(W)     1190, Shaft (mm)     5, Length (mm)     58,  Diameter C (mm)     42, Can Length D (mm)     33, Total Length E (mm)     80
Hobbyking 100A 3phase ESC	Link	90A, 100A Max. 7S LIPO max voltage.
4S lipo battery	Link	Capacity(mAh)     1300,  Weight(g)     155, Length-(mm)     73 Height-(mm)     35, Width-(mm)     31
Arduino nano	Link	spare, semi-broken from prior project.
2x mini stepper drivers	Link	From previous battlebot, crufted from lab cleanout. Under-powered for this application. Linear stepper drives with max current of 1A.

War Wounds

FIGHT ROBOTS FIGHT
Pre-fight Before battle the shell was completely black and the three 'whacker' posts were tangential to the central shaft.
The fight with MEGATRON The first battle of the day was against megatron (created by the seasoned battlebot veteran JAMO). The bumble survived a number of circ-saw cuts to the shell shown right. Fortunately there is a lot of shell here. There was a bit of mechanical jostling and somehow the internal 4S lipo went kaput on the bumble. After the match it was determined that a portion of the shell had to have collided with the lipo during an impact. This cut power towards the end of the match. The lipo was haistly hot-glued in place and it was more than possible that mechanical impacts could have dislodged it.
Lipo impact Here's a view of the impact with the lipo. Its at an odd angle and probably came about from the shell being dislodged / loose during an impact. If the tensioning spring is loose, an impact will shift the shell over temporairally. Its a good thing the cell wasn't punctured! The vulcanizing-tape tire treads also seemed to come apart after a bit of rough use. Legitimate tires would have helped significantly instead of slippery delran plastic, but it was enough to move.
Last minute battery. With the 4S lipo mechanically 'derped' I was a bit upstream without a paddle, fortunately a sketchy lipo, which had been brought as an emergency backup from the 'atomic stockpile'. After a bit of surgery, the 5S lipo (with 1 dead cell) became a 4S lipo, and was shoe-horned into the chassis. It had about 1/8'th of an inch clearance to the shell racetrack.
Upgrades for the Canadians I had the advantage of knowing that i would be soon combating the canadian bot 'ROBot FORD', so, with the help of the wunderbar welder bercu, screws were welded to the mighty steel posts on the outside of the shell. The logic was "Tires really dont like screws" While all 3 screws were lost into the tire, the canadians, added a 'push from that tall bar thing' addition to their bot, which allowed them to keep me at bay. only a few hits sunk in, but some bit of tire bled off so I was happy.
Impact with CANDY PAINT & GOLD TEETH In a grudge match with the SUPER INTENSE INVERTED LAWNMOWER candy paint, piloted by Bercu (shown below) the 'whacker' arms got a bit 'bent'. Keeping in mind that this is 1/8" steel, there was quite a bit of imparted energy here. It take a bunch to bend that stuff, next time, more reinforcement!
More shots of the Damage Thats a lot of bend! More reinforcement for the next shell.

  

Candy Paint vs Atomic Bumble Prime

What happened: I started the fight with the weapon at 'full' when I woke up the remote. As a result the controller defaulted in a safe mode and i had to spin it down to 0 to restart. By the time i had the weapon back the first blow was struck, and due to the shell angle on candy paint, the bot spun away pretty fantastically. Huzzah.

Concluding Remarks:

TIME! It takes a while to get a working bot ready for the northeast robot superbowl. NOT 3-4 DAYS. Practicing, learning the performance of the bot and working out kinks are activities that should not happen first on the battlefield! HEED MY WARNING!
Spending time on propulsion could have made a significant difference in the robots performance. Steel worked really well for everything, although, I wish I had spec'd thinner stronger steel alloys, the floor-tile, while strong, wasnt nearly as sturdy as some chromoly steel alloys. If more time was spent on mass allocation I would have probably taken the time to add an inner skirt assembly to give the shell more torsional rigidity, and also anchor the 'smasher bars' into the shell frame better.

This could have not come together without some awesome folk: The Bercu, joiner of metals, Brian Chan and the Hobbyshop-ians, for letting me use the OMAX SLICER OF METALS, Charles G, for providing transport, design feedback and a metric boatload of connectors, wires and whatnottery. The folks in the land of FSAE for welding access, and of course the MITERS, whom i nicked the floor fairly well when spinning up this beast. You folks are awesome.

The  'walker' classification also peaked my interest, it allows for an extra 50% mass for the bot, future revisions of this bot might benefit from being a shuffler bot, although I'm slightly concerned that the 'bouncy' nature of walker/shuffler bots might cause a shell-spinner to become unstable. There were 2 'walker' class shufflebots at motorama, both were pretty darn awesome and worth oogling at.


If you have questions or comments, ask below or send over an email.

Comments:

HTML Comment Box is loading comments...

Dane.Kouttron
Rensselaer Polytechnic Institute 
Electrical & Electrical Power
631.978.1650