Sumo grap­pler robot: (brush­less hipster)-squared met­ric 12 O’Clock­off fea­tur­ing belt dri­ve skate wheels, real hot dog1 grip­pers, ser­vo-actu­at­ed clam­per, and too many screws.

Liftlord low 3/4

It was­n’t the idea, but what it became as I rolled with the punches.

Also, it’s not a beetleweight (3 lb)!

Held on Sun­day-before-Labor-Day, the micro bat­tles (1 and 3 lb class­es) event at Robot Bat­tles at Drag­onCon grew over the years in com­peti­tors while shrink­ing in sched­ule, turn­ing into a sin­gle-elim­i­na­tion, no rum­bles affair. That means half of the bots entered get exact­ly one fight: less than ide­al odds for me con­sid­er­ing I had exact­ly one win over the course of like ten events. So, it’s time to go big(-ger). Big enough to put me into the 12 and 30 lb fights on Labor Day!

Maxon gearmotors

The con­cept start­ed, as usu­al, with sourc­ing motors too big to be use­ful in a 3 lb. Nor­mal­ly I’d shove them into a Big Data any­ways and call it a day2, except this turned out to be a phys­i­cal impos­si­bil­i­ty and w‑wait… why am I still think­ing beetles?

These Max­on gear­mo­tors each com­prise a 25 W EC-Max 22 sen­sored brush­less motor cou­pled to a 19:1 GP 22 HP (“high pow­er”) plan­e­tary gear­head. I scored four plus a spare motor (no gear­head) at an aver­age of $45 each.

Maxon EC-max 22 283858 motor 25W datasheetMaxon GP 22 HP 370689 gearhead high power datasheet

Now, “too big for bee­tle” does­n’t auto­mat­i­cal­ly mean I can go up to the next weight class. The gap between 3 lb and 12 lb—a fac­tor of 4—is the biggest change in scale between any two adja­cent weight class­es in US com­bat robots3. Com­pared to the pair of 550-ish drill motors com­mon­ly used by many suc­cess­ful 12 lb Drag­onCon robots, my 22 mm diam­e­ter Max­ons look pos­i­tive­ly puny. I mean look at them com­pared to the Stan­dard Rage­Bridge 2.

Liftlord traction motors mounted

Why, then, did I think this set­up could work?

  1. The wind­ings on the motors, rat­ed for 24 V, are rel­a­tive­ly “cool.” Even as two-pole inrun­ners, they clock in at a low, low Kv of 549 RPM/V, a fig­ure you might expect from a 22-pole 50 mm hob­by out­run­ner. As a result, I can expect them to pro­duce a fair amount of torque with­out run­ning insane cur­rent through them.
  2. The motors have Hall-effect sen­sors that allow their con­trollers to deter­mine to read—at any speed and with­in ±15°—the rotor angle and thus the appro­pri­ate sta­tor cur­rent to pro­duce torque. This helps close the gap between brush­less and brushed trac­tion sys­tems at low speed.
  3. You don’t have to even peek at the datasheets on the plan­e­tary gear­heads to tell they’re srs busi­ness. The 6 mm diam­e­ter shaft sup­port­ed by sealed 696 bear­ings looks com­i­cal­ly over­sized rel­a­tive to the gear­head body.

Run­ning the num­bers, I can tell that with a 6‑series lithi­um-poly­mer pack, this set­up can push out sim­i­lar peak mechan­i­cal pow­er to drill motors, but in a small­er, lighter pack­age4. Unfor­tu­nate­ly, a small­er, lighter pack­age has less cool­ing capac­i­ty and ther­mal mass, thanks to low­er sur­face area and actu­al mass. So aver­age pow­er dis­si­pat­ed dur­ing a three minute fight was gonna be a problem.

Liftlord top traction motors highlighted

By the way, the absurd aspect ratio of indus­tri­al gear­mo­tors make it look like I’m only mount­ing them by their face. In actu­al­i­ty, 41% of the over­all length of each gear­mo­tor (the entire plan­e­tary gear­head, adapter plate, and change) is sup­port­ed by frame material.

How do you resolve design uncer­tain­ty and in the process dip your toes into the 12 lb pool? Build a brick bot with four of these motors for trac­tion to see how things would work out.


except lol jk no I did­n’t. Too much pride. Due to the Wang-wedge arbi­trary neg­a­tive dou­ble stan­dard (“oth­er peo­ple can build wedges but I can’t”), I had to put in some kind of… active weapon?

Liftlord render

Giv­en that exact­ly one event in the coun­try runs 12 lb sumo fights and it’s dom­i­nat­ed by the lifter/flipper–clamper/grappler spec­trum, I fig­ured I’d build this new bot as a clone of the most icon­ic of that event, the 30 lb Ãœber­clock­er. I even named the CAD mod­el as such:

CLOCKOFF model on Dropbox

But ego struck again. I made it a design goal to NOT be like Clock­er aside from the over­all con­cept of a four wheel dri­ve grap­pling bot. This meant no look­ing at ref­er­ence pho­tos. It meant inten­tion­al­ly stuff­ing in dif­fer­ent com­po­nents from what I can recall Charles using. It also meant replac­ing entire sub­sys­tems with sil­ly shit: ser­vos instead of lead­screw to actu­ate the clam­per, no spring legs instead of yes spring legs, belt dri­ve lifter instead of gear dri­ve lifter. After all, it’s not tru­ly Clock­er with­out its lift gear, right?

Liftlord lifter drive pretendo

Turns out “not look­ing” and “don’t rip off” are mutu­al­ly exclu­sive cri­te­ria. Yes, I avoid­ed the gear dri­ve, but still end­ed up with a hol­low alu­minum lifter live shaft run­ning in sleeve bear­ings with clamp shaft col­lars for shaft-to-arm torque trans­fer, because I lit­er­al­ly for­got that’s what’s on Ãœber­clock­er but had enough sub­con­scious fan­boy mem­o­ry to design it like that anyways.

Final­ly, every­thing had to be met­ric. At this point, this has almost noth­ing to do with round num­bers and stan­dard­iza­tion than to do with my grad­ual switch to build­ing almost every­thing with a hefty help­ing of “Chi­ne­sium,” the bot com­mu­ni­ty’s term for materiel sourced (near­ly) direct­ly from East Asian sup­pli­ers through the likes of eBay, Hob­byK­ing, AliEx­press, and Bang­good. At first I thought the term implied low qual­i­ty (for some folks, they’re syn­ony­mous) but more accu­rate­ly it describes the uncer­tain­ty of these items’ ori­gin and dura­bil­i­ty than a clear state­ment against them. Regard­less, Chi­ne­sium’s all met­ric, since lit­tle of it is nec­es­sar­i­ly intend­ed for direct Amer­i­can consumption.

"I wouldn't put it past you guys to use real hot dog in a bot."

Met­ric means hard­ware with met­ric screw threads, belt pitch, bear­ing bores, every­thing but wire gauge and raw stock (don’t give me ideas). It also means com­po­nents that fit met­ric stuff: thus 80 mm inline skate wheels made to mount 608 bear­ings and 775 motor lift motor with a 42 mm gear­box (10 mm out­put shaft with 4 mm key­way ;). Side note: Rage­Bridge 2’s heatsink mount­ing holes will clear M3 screws just fine.

Liftlord filing bottom plate

After mov­ing to Cal­i­for­nia and thus away from the Inven­tion Stu­dio, I’ve been in enough of a water­jet drought to have for­got­ten how to prop­er­ly design for it. I put in 0.05 mm clear­ance between tabs and slots in my part, which was­n’t enough for Big­Blue­Saw’s OMAX qual­i­ty 3 cut for 1/4 inch 6061. For­tu­nate­ly, I con­sid­er man­u­al fil­ing an excit­ing sub-hob­by of com­bat robot­ics (and of life, even), so this is A‑alright.

As a aside, Big­Blue­Saw is ludi­crous in the best way. The web quote inter­face lit­er­al­ly dumps into a full DXF ingest/process pipeline (writ­ten by Simon IIRC) that com­putes cut­ting length, num­ber of pierces, image pre­views, and I assume some approx­i­mate part nest­ing to get you an exact price near­ly instan­ta­neous­ly. If that’s just my expe­ri­ence with the fron­tend, the automa­tion going on from site to machin­ing must be some­thing delightful.

Oh and Julie runs the cus­tomer ser­vice so if you order parts through BBS, it’s like order­ing from team Chaos Corps. ^_^

Liftlord milling bearing bore

The bronze sleeve bear­ings for the lift shaft need­ed an exact cir­cu­lar press fit, so those were water­jat to exact dimen­sion and fin­ished on the Google Work­shops CNC mill, although not before I ruined two bear­ings by press­ing them into the rough water­jet bore anyways…

Liftlord turning down wheel spacers

Sim­i­lar idea here: these spac­ers (hubs?) go into the 22 mm bore nor­mal­ly occu­pied by a 608 “skate” bear­ing, so I’m fin­ish­ing their out­side diam­e­ters on the lathe to clean up water­jet taper and non-cir­cu­lar­i­ty (water­jets are X‑Y machines, after all). The live cen­ter was not a good idea. I should have used pure­ly axi­al force to jam the hubs against the jaws.

Liftlord milling keyway lift gearmotor

Kind of sil­ly but the 4 mm key­way was under­sized on this gear­head out­put shaft, so I expand­ed it real quick. I should have clamped the shaft, not the motor.

Liftlord traction pulley tapped

We don’t talk about this.

Liftlord wheel, trilobularly expandified

Or this.

Liftlord milling side armor

Needs more chmafer.

Liftlord clamper and lifter

All the black plas­tic you see is Onyx car­bon com­pos­ite parts that Jamo 3D print­ed for me on a Mark­forged Mark Two. These parts are unbe­liev­ably stiff and light. The qual­i­ty of the print­ing is insane; most of the cir­cu­lar bores were designed to exact dimen­sion and end­ed up per­fect press fit. I had orig­i­nal­ly intend­ed for all the plas­tic to be milled from UHMW but this oppor­tu­ni­ty came up and saved me a lot of machine time.


Liftlord left high

One thing I think is rea­son­ably unique to Lift­lord is the ser­vo clam­per mech­a­nism. Nor­mal­ly hob­by ser­vos are too wimpy to use for secur­ing anoth­er 12 lb bot. But I fig­ured if I add a slight­ly com­pli­ant link—in this case, ultra-high mol­e­c­u­lar weight poly­eth­yl­ene (UHMW-PE) strand­ed fish­ing line—and remove the clam­per from self-right­ing forces by using a spring to open the jaws, a pair of “stan­dard” hob­by ser­vos might sur­vive com­bat load. The orig­i­nal idea involved adding even more com­pli­ance to the fish­ing line link with a short high-rate spring on the ser­vo horn, which you can see in the pho­to (not attached). How­ev­er, I could­n’t run them as far past their spec as I want­ed to. They stretched lit­tle before plas­ti­cal­ly deforming.

Liftlord opposing clamper servos

I for­got that all this would take up five out­put channels—two for trac­tion dri­ve, one for lifter, and two for the servos—on my FrSky X4R radio receiv­er (RX), which had only four: one more than I had ever pre­vi­ous­ly used, except for Gyro King which had a lap­top link and PS3 con­troller.

Why could­n’t I run both ser­vos from the same sig­nal with a split­ter? Well, the ser­vos are fac­ing in oppo­site direc­tions. Run­ning off of the same sig­nal, they’d each spin in the same direc­tion that’s actu­al­ly oppo­site direc­tions. I need one ser­vo to run on a chan­nel that’s reversed in trans­mit­ter software.

Or do I? What if I were to reverse one of the ser­vos? In hardware?

After all, these ser­vos are just a neg­a­tive feed­back ampli­fi­er around an ana­log poten­tiome­ter (“pot”) and a brushed (DC) motor. The poten­tiome­ter lin­ear­ly maps rotary posi­tion to ana­log volt­age. If I were to reverse that map­ping, say by swap­ping the pot’s out­side ter­mi­nals, then I’d reverse that map­ping, so that the ana­log volt­age now maps to a rotary posi­tion mir­rored across the pot’s mechan­i­cal range.

Final­ly, I’d have to negate the gain of the con­troller too, say by swap­ping the leads of the motor to reverse its polar­i­ty, so that the feed­back to angu­lar posi­tion change remains negative.

Now I have a reversed ser­vo5!

Rewuzzled servo comparison

I test­ed this on some cheap ser­vos, tak­ing care to mark the ser­vo I reversed. To my sur­prise, this hack just worked.

Less to my sur­prise, ser­vo revers­ing turned out to be a fair­ly com­mon hack, with docs all over the inter­net. It’s almost like some peo­ple pre­fer to change things in hard­ware rather than in soft­ware or something.

Liftlord rewarzed servo mounted

But of course, the high-pow­er ser­vos I intend­ed to use were insane­ly dif­fi­cult to reverse like this. Its motor, con­trol board, and pot were all one beefy unit hard sol­dered togeth­er. I can’t even bring myself to describe the hor­ri­ble next lev­el hack­ing I impro­vised to make them work.

Epic ship day extrafirmware

The day before I left for Drag­onCon, Lift­lord’s state was rough­ly “the parts fit and are holes are tapped.” This left open the prob­lem of how the hell does this move, which, as I expound­ed on above, was the most urgent and impor­tant prob­lem in the whole design. Goes to show how great I am with priorities.

The orig­i­nal least com­mon denom­i­na­tor for trac­tion elec­tron­ics was to use cheap Chi­nese con­trollers (CCCs) flashed with SimonK. This was easy and shit­ty (because it did­n’t take advan­tage of the Max­ons’ sen­sors): the absolute worst case back­up plan, which is to say it was plan A on account of all oth­er plans being not even conceived.

Unfor­tu­nate­ly, SimonK’s default start­up rou­tine did­n’t “take” on my motors. I did­n’t want to fig­ure out why or do the tun­ing because I was­n’t con­fi­dent I could make it work in 12 hours. So time for plan B, which as I men­tioned, did­n’t exist or anything.


HÄRDBÖRD controller

I start­ed going through ran­dom box­es on my shelf because I fig­ured, WELL DANG I should be able to dri­ve some motors. What do I find but the control/power dis­tri­b­u­tion board from HÄRDBÖRD, com­plete with PIC24, XBee mod­ule, and most impor­tant­ly Max­on DEC 50/5 motor con­trol mod­ules.

This was maybe the sec­ond or third print­ed cir­cuit board (PCB) I had ever designed, from way back in the heady days of 2010, and part of the first thing I ever worked on with Jamo. It even has a Batch­PCB ID mark on it, from the times when hob­by­ists were will­ing to wait two months for PCBs and we were hap­py to do it because they had green silkscreen just like the pros and only for like fifty bucks my god this is so amaz­ing we’re liv­ing in real time like all effect­ing a desk­top man­u­fac­tur­ing REVOLUTIONNNNNN.


Once I came down from the Bre-stol­gia, I dug up the docs for the mod­ules and found that they have a 0–5 V input for speed con­trol, an enable pin, and a direc­tion pin. This meant that I need­ed a micro­con­troller to con­vert the dig­i­tal pulse-width mod­u­lat­ed (PWM) sig­nal of my receiv­er (RX) into what the DEC 50/5s under­stand. The PIC24 code on the board actu­al­ly did some­thing sim­i­lar: con­vert Wii Nunchuk data received over XBee to Max­on DEC ana­log. Unfor­tu­nate­ly I could­n’t find a 16-bit PIC pro­gram­mer any­where so I could­n’t just hack up that code; I’d have to find a small, gen­er­al pur­pose micro­con­troller to use in the PIC’s place.

You might expect me to be trip­pin’ boards. But I don’t nor­mal­ly need small, gen­er­al pur­pose micro­con­trollers because my projects tend to be the oppo­site of gen­er­al pur­pose. So I don’t have any Arduinos Micro chill­ing about. Or mbeds LPC1768. Hell, I did­n’t even have an STM32FxENDEAVOUR.

Teensy LC front

What I did find was a Teen­sy LC pur­ple edi­tion in a fiber gro­cery tote that I scored in some swag bag at a com­put­ers con­fer­ence (both the board and the bag). This… this is my San Fran­cis­co life.

This is the “low-cost” (LC) ver­sion of a Teen­sy, using a ARM Cor­tex-M0+ core Freescale micro­con­troller that has non‑5 V tol­er­ant input pins. This means that the Teen­sy LC can’t direct­ly take the 5 V “READY” sig­nal from the DEC mod­ules. More crit­i­cal­ly, it also means I can’t slap a pull-up resis­tor to 5 V onto an open-drain PWM pin and expect 5 V PWM out­put, with which I’d use to syn­the­size 0–5 V ana­log. I need a lev­el shifter for each input and a buffer (or lev­el shifter, which is like a bidi­rec­tion­al buffer) for each output.

Liftlord dead bug controller

I end­ed up wiring things up “dead bug” style on the back side of a blank HÄRDBÖRD PCB. The red board is a Spark­Fun Log­ic Lev­el Con­vert­er — Bi-Direc­tion­al with four lev­el shifters. The Teen­sy LC itself also has a 5 V‑buffered PWM out­put, which I used. The DEC mod­ules’ dig­i­tal inputs have 2.4 V thresh­olds, so they did­n’t need buffer­ing from the Teen­sy LC. But wait, why is the the 5 V dig­i­tal PWM out­put con­nect­ed straight to the DEC mod­ule’s speed val­ue input? Don’t you need to fil­ter it to ana­log? Isn’t the 10 KΩ out­put of the lev­el shifter too high com­pared to the 107 KΩ DEC input imped­ance? The answer is the same as always: STFU about my hard­ware bug and fix it in your firmware.

Liftlord Arduino code

Teen­sy uses the Arduino IDE and libraries. They most­ly don’t get in the way of writ­ing nor­mal C++, except for egre­gious novice mis­takes designed into the Arduino library like min/max macros. The libraries are extend­ed with addi­tion­al calls by Teen­sy. The Teen­sy-spe­cif­ic exten­sions saved my bacon this time, because they include analogWriteFrequency and less crit­i­cal­ly analogWriteResolution, which allowed me to run PWM out­put at CPU fre­quen­cy / (1024-step res­o­lu­tion), or rough­ly 47 kHz.

Liftlord disassembled back high

This was appar­ent­ly fast enough to be cut out by what­ev­er low-pass fil­ter was on the DECs’ speed inputs.

Now you can see that all of my belt ten­sion­ers are M5 shoul­der screws with sleeve bear­ings, bolt­ed into frame slots with nylon lock nuts. Also, jaw/spider shaft cou­plers because why not.

Liftlord controller with IMU

Once I real­ized it all worked, I threw on an IMU (why not?) and mum­mi­fied it all in elec­tri­cal tape. Oh, remind me to nev­er again use high strand count sil­i­cone-insu­lat­ed wire for board-to-board con­nec­tions because you have to twist and tin stripped ends before sol­der­ing them to pads. Huge PITA.

Weapon testing

Around this time I also dis­cov­ered that my attempts to engi­neer out the uncer­tain­ty around the 27:1 775 Chi­ne­sium gear­mo­tor might have land­ed this bot well into overkill ter­ri­to­ry. It was no longer accu­rate to call it a lifter, since it was more than hap­py to toss my 3 lb bots into the roof at 50% trans­mit­ter rate gain and low­est Rage­Bridge cur­rent lim­it. It blurred the lines between lifter and flipper.

DC 2016 taking over dining table

Time to install goo­gly eyes.


Drag­onCon 2016, held in down­town Atlanta, GA on Labor Day week­end, was what­ev­er Drag­onCon is. It’s hard enough to describe Drag­onCon rel­a­tive to what I think oth­er cons are like, but it’s impos­si­ble to explain in absolute terms. Instead I’ll just point out that it has the sec­ond-old­est robot com­bat com­pe­ti­tion in the world and fas­ci­nat­ing cosplay.

Actu­al­ly, I’ve been giv­en plen­ty of chances to plug Drag­onCon cos­play and nev­er did. I should­n’t pass up this one.

Drag­onCon’s lack of focus on any par­tic­u­lar genre—though it’s heavy on fan­ta­sy and sci-fi—gives it a mul­ti­fac­eted iden­ti­ty that bleeds into its cosplayers.


This and the par­tic­i­pants’ devo­tion to their prod­uct results in very eye-catch­ing cos­tumes that don’t nec­es­sar­i­ly repro­duce char­ac­ters from any par­tic­u­lar intel­lec­tu­al prop­er­ty (IP).

Iron Man King Triton

In fact, some of my favorite cos­tumes are the ones that jux­ta­pose unre­lat­ed IPs, gen­er­at­ing humor by sur­prise and some­times détournement. Iron King Tri­ton is pret­ty emblem­at­ic of DC’s “why not” remix cos­play performance.

My favorite three ele­ments of Drag­onCon came in one hel­la splen­did mish­mash this year: bat­tle­bots, cos­play, and my friends. Check it outtt: cynaes­thet­ics: FEMALE HANZO + BATTLEBOTS COSPLAY

They did some out­stand­ing work. I’m in total awe at the hus­tle they and oth­er folks at Con put into their var­i­ous projects, all of which I con­sid­er per­for­mances: parade cos­tumes and vehi­cles, knowl­edge exchange pan­els, even prep­ping them­selves for par­ty­ing is a planned endeav­or. I got some sharp regret for not walk­ing around with a cam­era; after all, I had less to work on so I feel I was oblig­at­ed to cap­ture it all. Instead, I have three pho­tos of food. Thanks, every­one else who did document!


Liftlord fight prep

From team TRIGGER WARNING… this bot’s AGGRESSIONS are any­thing but MICRO. It’s LIFTLORD!

After all that, I became uh, indis­posed. So Aaron drove Lift­lord for me. He had brought a 3 lb flail bot, which fought in the pre­vi­ous day. He’s bet­ter at dri­ving Lift­lord than I am with­out hav­ing even tried it pre­vi­ous­ly, while I had at least 20 min­utes of dri­ving practice.

In fact he even beat Drag­onCon Bron­co, Dale’s T‑Boner 2.0. I can’t help but admit that the win, as well as the first win against Jason “Evil” Brown, were aid­ed in part by mal­func­tion in their bots.

I think Lift­lord even­tu­al­ly exit­ed the brack­et as a quar­ter­fi­nal­ist? I’m kind of hazy on that one. Oh did I men­tion that the tour­na­ment was sin­gle elim­i­na­tion, just like the insect (1 and 3 lb) class­es? There goes 50% of my moti­va­tion for build­ing a stage bot.

12 O'Clocker and Liftlord

I was gid­dy see­ing this side-by-side because I hon­est­ly did­n’t expect Lift­lord to be so sim­i­lar to Charles’s 12 lb ver­sion of Ãœber­clock­er, 12 O’Clock­er, giv­en my require­ments on “not look­ing at ref­er­ences.” It’s not as adorable as 12 O’Clock­er though.

Mean­while in the 30 lb class, Andrew’s Pusheen, a wood­en bot (also brush­less hip­ster, maybe even more so), won against Near Chaos’s chain­saw bot. It’s even bet­ter than the time Wedgee took out team Inven­tion Stu­dio’s ass­bot. #drag­oncon

Final­ly I got up to the stage for rum­bles. As I under­stand, I won the sec­ond one but as I con­tin­ued to be uh, indis­posed… I don’t real­ly remem­ber it. I’m feel­ing too embar­rassed to watch the video since I was prob­a­bly dri­ving like derp.

Dri­ving this bot real­ly made me real­ize what a non­sen­si­cal amount of coor­di­na­tion is required to pull off a Charles-style grap­pling per­for­mance. And Lift­lord’s clam­per has posi­tion con­trol while Ãœber­clock­er has first- or sec­ond-deriv­a­tive con­trol. I had decid­ed while prac­tic­ing before the tour­na­ment that I was­n’t even going to both­er with the clamp­ing. I just was­n’t able to do the whole maneu­ver-scoop-clamp-lift sequence with author­i­ty. It also helped me appre­ci­ate what a huge hin­drance the stage ris­ers’ joints were to bot mobil­i­ty. If you don’t raise the lifter wedgelets and keep them raised, they will jam beneath the cheese-grade steel frame. I’m impressed with the prac­tice that the vet­er­an bot builders must have put in to put on the show that they do.

This is actu­al­ly my first win at Drag­onCon in I think six odd years of try­ing to com­pete in it. I felt kind of val­i­dat­ed that I could sup­press my urge to build nut­ty curios and put in the atten­tion to detail—Loctite every screw, wrap all electronics—to enter some­thing com­pet­i­tive. It point­ed out dri­ving as some­thing to work on: my shit­ty tour­na­ment record and fre­quent mal­func­tions means I’ve had rel­a­tive­ly lit­tle are­na time in the years I’ve been doing this hob­by. Maybe it’s time for a strat­e­gy rework? ∎

Shouts out to Jamo for the Mark­forgery, Aaron for the hacks/driving/photos, Wes for goo­gly eyes/tape work, Big Blue Saw for their help, and every­one else at Drag­onCon 2016 who made it an incred­i­ble experience.


  1. “I would­n’t put it past you guys to use real hot dog in a bot.” —Lucy []
  2. The Portescap trac­tion motors in Big Data DVT and PVT are more pow­er­ful than these. []
  3. Actu­al­ly on the West Coast, the 15 lb class replaces 12 lb for an even big­ger “class divide.” []
  4. Also does­n’t run at sig­nif­i­cant­ly high­er effi­cien­cy, since push­ing high cur­rents into a cool-wound core cre­ates crap­py cop­per loss­es. []
  5. IRL this involved zero engi­neer words and sev­er­al min­utes of star­ing at ser­vos before going, “OH what if I just swap these and these wires?” []
  6. This is a rebadged Chi­ne­sium “PG45” 775 gear­mo­tor that I would­n’t buy from Andy­Mark at their 2.5× markup except for their US stock. A less over­priced but wimpi­er US-stocked 775 gear­mo­tor is avail­able from Lynx­mo­tion. []