The Susquehanna Boxcar: A Return to Motorama

For literally over a decade now, Motorama and the NERC Robot Conflict event associated with it has been a sort of winter robotting tradition for me, minus only a few years; in 2020, I was in the middle of moving and so didn’t go back north for a while, and there was no Motorama in 2021 for Reasons.

However, the last time I really fielded a bot was in 2018 with the outgoing 30haul; I just went in 2019 for funsies and to help out. This year, with the Reasons beginning to wind down and with everybody in the robotting community ready to get competitions going again, I decided to to take a break from vans and enter something again.

But I had one rule for myself: No spending money on the robot.

This bot, whatever it’s going to be, was just a short detour from Operation IDIocracy, and it was going to just be a shitpost entry. Cleaning all the grunge and getting the engine apart has taken longer than I was counting on, and on top of that, it’s cold so I kind of lose motivation to wrench on things. Robots can be done indoors.

I had to build this entry as much out of just the materials and parts in my midden. Screws are fine, buying motors and ESCs or ordering custom cut parts is not. So what am I going to make? Initially, I thought about an extremely dirt floor rendition of Stance Stance Revolution in a 30lber, maybe using lawn care blades as the weapons.

To this end, I have plenty of weapon motor candidates in the form of 40-60mm brushless motors from Overhaul. I actually had a “preview” sketch model of what the next SSR beetleweight was going to be:

Actual stanced wheels! I had a few bright ideas on how to drive those while I was mocking the design up. This is definitely going to happen at some point; however, now getting close to the new year, I was getting less inclined to make a 30lb SSR as it was looking like more and more effort. Effort was to be avoided.

So why not a 30lb Sadbot? After all, Sadbot is my most successful heavyweight. I had plenty of 3″ C channel and 1 x 3″ rectangular tube left over from Vantruck add-on candidates, which was perfectly half the size (height) of Sadbot’s 6 inch channel stock.

We begin with a master sketch of the outline of the bot which I used to drive Inventor’s “Frame Generator” design tool, which is more or less the equivalent of a Solidworks Weldment with its libraries of standardized profiles and extrusions.

Weight was going to be a serious concern; my first pass frame measured out at 21 pounds out of 30. So having the frame size be driven by placing components and easily adjustable was important. I was basically looking at stuffing a 12lber inside a ring of steel.

The chassis material inside the beltline was going to be good ol’ UHMW plastic, a staple of my high school, pre-machine tool building. It behaves like a very dense bowl of grits and is easily workable with woodworking and handheld tools. This was really shaping up to be some kind of accidental retro-build.

Adding to the retro vibe was picking some 18V mixed-heritage cheap drill motors out of my bucket of motors. I would occasionally over the years “sample” cheap cordless drills from Harbor Freight, Walmart, and the like, so I have probably over a dozen of these which are almost all interchangeable. The newest lithium battery models have been deviating from the classic conical gearbox shape as manufacturers try to make the drills smaller/lighter while having 2 speeds as a default option.

The drill motor is to actuate a pokey stick similar to sadbot, and I was planning on a simple chain drive to an axle above the frame.

For drive, I decided to keep digging through my motor pile and found some mild-wound 555 sized motors. They’re identical in size to the usual 550 size R/C car or truck motor (and drill motors) but are typically wound to run at higher voltages, like 24 volts. The ones I have are surplus from who knows when, and have a Kv of 450 RPM per volt (Compare this with the typical 1200 to 1500 of a 550 class drill motor). They can therefore be run with a single stage of open gears, trading speed for torque.

In this bot, space was more the issue than anything, and using a single-stage spur reduction let me push the motors into the UHMW frame rails to clean up space in the middle for the drill motor and eventual electronics

So why not brushless? Well, that would involve finding 4 matching motors and/or ESCs in this size range, which I didn’t have. And no spending money!

I began sizing up the gear drive using a 4 inch wheel as a reference, using the spacing of the 555 drive motors and the drill motors as a guide for what gear size I could use. I’m going to exercise some impure nostalgia here by just 3D printing the wheel and hub. The gear pitch is 24 DP (or about Module 1), since that’s what my 3D printer flock can consistently print and have it work out well.

Initially, I was designing this hub to use the same wheels that 30haul did, made using the same methods and having the same dimensions as the Vex hubs I was using – hence the 1″ nub sticking out of the gear. I was going to hole-saw some 1″ gum rubber sheets and retain them with standoffs in the manner of its “disposable” duallies.

However, on a trip to Harbor Freight, I serendipitously found a new avenue to explore…

My only real complaint about the gum rubber wheels for 30Haul was that the sheets don’t come in any thicker dimensions than 1″, at least not with my minimal familiarity with the rubber indutry. Ideally, I’d be able to get a single 3-4″ thick wad, or laminate a few sheets together and be able to cut out wheels for actual full-size Overhaul.

I could probably get away with running singles for this bot, but I’m also keeping an eye out for thick chunks of rubber foam. And I found it by accident at Harbor Freight in the form of what I call the “Harbor Freight Yoga Mat”, actually a kneeling pad. It’s part number 56572 (as of now, anyway). It’s kinda exactly what I wanted – a big slab of rigid-feeling EVA foam. The texture felt right to make into wheels, and if the traction wasn’t good enough by itself, certainly I could coat them in latex or silicone (a long time favored builder trick still in active use at BattleBots!) for more traction.

So I violated my own rule, as usual, and spent money on the robot. One Harbor Freight Yoga Mat, for science!

I started prints of the hub design on a Markforged Mark Two (hi Markforged!) as well as a new-to-me Creality Ender 3 V2 (hi Naomi Wu!) that I picked up for $40 in a “Curated Wreckage” state. The pink material is PETG, which is my preference these days for non-critical and experimental prints that nevertheless will be used in the final application, despite my denial.

Those slots in the gears were added for a wheel retention feature that I dreamed up as well, and which will be seen shortly.

One issue I ran into with making the gum rubber wheels for 30Haul was keeping the hole saws centered with respect to each other. I decided to tunnel my way through the problem this time by opening up the center bore of the 1″ hole saw to fit on the 4″ hole saw’s larger arbor thread.

These cheap hole saw sets aren’t hardened meaningfully, so I blasted the bore out to 16mm with a metric step drill I had. 5/8″ probably would have been fine also. After this, I was able to jam both of the hole saws onto the larger arbor!

It was then just a drill press job away from making wheels that were at least concentric one one side; I didn’t want to go buy 2″ deep hole saw for this, so I had to just flip the Harbor Freight Yoga Mat around and find the pilot hole.

Here is what the deal is with the six slots. I didn’t want to just adhere the surface of the foam to the printed hub face because I thought the foam was just a bit lacking in integrity and would separate under the “skin”. I decided to add some interference elements, not unlike the standoffs of 30Haul, but without the intention that I could take it apart again.

So out came these ABS trim strips I bought for something way back when. I decided to use them as quasi dowel pins or driving keys. They get cut into chunks, pressed into the hub, and then trimmed with scissos.

With six slits cut into the foam with a knife, I could press these in with adhesive covering every surface, thus ensuring that plenty of surface area grips the foam on the inside.

E6000 contact cement is my go-to for these kinds of robot shenanigans. I slathered each edge of the “blades” and the hub face with it, and left it to cure under a weight.

That’s what one wheel looks like. The next day, I was sufficiently satisfied with the roundness and rigidity, and hit the go button on 6 more wheel hubs.

Alright, experiment over. Time to flesh out the rest of the design. I needed the drill motor shaft to stick into the center of the bot, so I decided on just using some fat standoffs.

The drill motor is mounted in a slightly unusual but also traditional way, using 4 of the of ball bearing clutch holes as mounting holes (and the other 4 still serving as clutch ring locking holes). All of these holes are to be tapped #10-32. This approach is stronger than using the two very small #4-#6 holes in the nosecone area.

The poker weapon drive is as simple as it could be. I’m using the same tactic as I used on Sadbot’s latest poker: Just welding a steel tube to a sprocket. The center bore is made for a 5/8″ ID bushing, which will just be a drilled piece of Delrin plastic in a 3/4″ drilled hole. The sprocket I plucked out of my Tomb of the Unknown Power Transmission Part is a 30 tooth #25 size.

25 isn’t my go-to for lifter/hammer style weapon in a 30lber. I’d much prefer #35, but I only had very small sprockets in #35 at the time, and something something no spending money. The same applies to the 5/8″ shaft. I would prefer at least 3/4″, but had some 5/8″ 4130 chromoly tube that I figured would stand a better chance than thin-wall 3/4″ regular buttery steel.

One of the next things to settle was where to put the weapon axle. I ran through a few options for this, including having it in-line with the top surface of the frame and making a half round shaped clamp to hold the axle in place.

The “overkill V2” iteration is where I decided I also wanted the ability to tension the chain, so let’s make a fancy set of towers that have slots so I can slide it back and forth for tensioning purposes?!

The third and final iteration is the “No, that’s too much effort” revision, which was a compromise. I needed more space between the sprockets anyway, so the axle had to move up. I decided to just make a simple block mount from the same offcuts that the frame will be made from, and use shaft collars or spacers to retain it axially instead of clamping

And that….. is all the CAD that I did before starting to cut metal and plastic. After all, I had 95% of the parts in house and ready after searching around for the day.

The electronics of this thing were to be equally found-object. Namely, I “found” two brushed Ragebridges, and decided I was going to break up one of my knockoff 18V Milwaukee Fuel batteries, which contain 2500mAh 18650s. The “9Ah” knockoff contains 15 cells (5S3P) and so I was going to run 7S and make two packs out of the one battery.

Next up: Fabrication, then the Motorama trip!

The Belated Equals Zero Christmas Special: Restomodding a Tomy Omnibot and Vintage Futaba AM Transmitter

Here’s a funny little side road I went down over the course of the 2020-2021 holiday season when nothing was open, nothing was shipping even if it were open, and I had otherwise nothing build-wise going on. It was a good trip back to the days when I did these little hacky things more often, building and making something without really a purpose except to see it happen because it needs to happen. Even better that I ended up with a mobile, displayable simulacrum of the current trendy 80s-90s nostalgia that has fully engulfed pop tech culture.

It all started during one of my random van adventures (of course) up to North Georgia. Having passed a lot of what I call “knick knack stores” on the way to and from the Blue Ridge and Smokies, I decided one weekend to just spend my time perusing their wares.

You know what I mean – the “antique markets” and “junk stores” that pepper state highways and county roads, well outside of metro areas. I’ve some times collected vintage tools or neat mechanical things from them.

But this little place on the side of Highway 515 was special. That’s because when I walked in, this thing was just sitting right there on the main display counter:

What is that!? Well first of all, it’s a Dr. Inferno, Jr. After that, it’s a Tomy Omnibot, one of the outfalls of the 1980s robot toy craze after the first microcontrollers became economical enough to put in (at least high end) toys. This was, it seems, an over $1000 (in modern dollars) toy back in the day, and it had all the 80s computer trappings – a tape drive that you could record movements onto and play back, for instance.

There was no remote or other documentation or original packaging to go with it. Without the paired remote, there wasn’t really anything it could do besides “Be Collectible” which isn’t really my thing. I thought “Well, I already have several 80s vans, so why not add an 80s robot to the mix”? The thought popped into my head then that it would be fun to drive around at shows and meets, like #RadwoodBait but for robots.

And so this is the story of how I turned this Omnibot and a vintage AM radio into a contemporary dance duo that never actually was.

There’s really two independent parts to this story: The AM radio conversion, and the buildout of the robot itself. I feel like the radio conversion is actually the neater technical contribution, but nonetheless, here’s the index:

1. Repairing and Upfitting the Omnibot

2. Converting the T4NL AM 4-channel radio to a 2.4Ghz 7-channel radio

Fixing up the Omnibot

I heard lap belts in the back rows are no longer the recommended way to keep your children safe in the car.

Anyways, besides the Omnibot, I also picked up a few old knick-knacky tools, including the pictured brass hammer for the Master of Benches.

The robot itself was in good shape – it had one of the arm guide links broken off at a plastic boss, and the wheels were missing small chunks of tread. Nothing major exterior-wise besides a few blemishes. The drivetrain took a little bit of wiggling to become ‘unstuck’, frozen grease and grunge if I had to guess. Much like my M.O. with vans, if it’s vaguely robot shaped at all, I can probably get it to drive.

The back lid pops off with a screw to reveal the battery bays. The main battery for the drive system and lights was a 6V sealed lead acid battery (which is long dead). The ‘computer’, timer, and tape deck were seemingly operated from 3 volts of AA batteries, the terminals of which were also long corroded off.

It’s a bit gross everywhere, like it clearly sat in a basement/attic with bugs for years. At each juncture I took the opportunity to try and get the detritus out with some bleach wipes and rubbing alcohol.

This thing actually pops apart very quickly once the six or so screws holding the body together are removed. There’s a LOT of what appears to be hand-wiring inside. I can imagine the assembly line for this thing being a jig holding all the body pieces in position as technicians install the wiring, and then the whole thing is closed up.

More corroded wiring and bug detritus abounded inside the shell. How’d y’all even get inside?!

To my dismay, the arms were not motorized. They’re all held in place by friction washers backed by springs. I was at least expecting the shoulder to be a lifting axis or something, as that’s all you’d really need to make the thing more animatronic. Alas!

Maybe this can be a future upgrade path if I choose to dive in further – adding an arm axis that never was!

The shoulder joint in particular is a mounted on an insert in the main body and has a set of O-rings to give it some more friction over just the spring loaded washer.

Two follower linkages keep the forearms parallel; one of these, as you can see, broke off its stud. It didn’t affect the operation all that much because there were three others, especially the one on its same side. I’ll likely just epoxy this boss piece back on.

As I was taking all of these joints apart, I was also cleaning up any worn plastic debris (powder) and then lubricating each joint with silicone grease as I was reassembling it. The grease didn’t make each joint looser so much as less grindy and prone to squealing.

This is what each arm looks like once disassembled. The claws are also unpowered, and just have a open and close lever that snaps the ends apart or together. I could see this end having a small servo embedded in the wrist joint to pull on this lever, if I went the full animatronic route.

I decided that the most immediately helpful replacement for the 6V lead acid battery was a 6.6V bank of A123 cells! Conveniently, two cells in a row was basically the same width and height, so this was a no brainer.

I whipped up a quick battery pack using the original plug cut off from the lead-acid battery, and a repurposed 3 pin servo connector as a balance tap connection so I can throw it on one of my 2S dedicated mini-chargers.

To get the bot base driving independently of the onboard controller, I decided to embed two of my leftover Vex 29 ESCs into the motor power path. This meant I had to tap power from the main board, and it was actually pretty nontrivial to find where.

Unlike most new PCBs, there’s no obvious “power bus” because the components are all through-hole and the power has to snake around to get places, using jumpers. I had to follow the original power input wire around on the board until I found where on the multi-position switch terminals it ended up.

Now when I turn the robot on, my newly appended JST battery lead gets power.

I had these de-husked Vex 29s left over from beetleweights of long ago, either Colsonbot or the first Stance Stance Revolution. They’re still my go-to for “Random small bidirectional DC motor driver that doesn’t need more than 12 volts and like 2 or 3 amps”.

The wiring was deceptively simple here. Because the receiver needed power as well, I just stuck the battery connector through a servo connector conversion into the receiver. The Vex ESCs are supposed to receive power and signal through that same connector. It’s only 6.6 volts anyhow!

(I’ve done the exact same thing for Stance Stance and Colsonbot, mind you. The receiver’s power trace is thick and short enough that on the order of single digit amps isn’t a problem)

I’ve closed the robot itself back up now after polishing the dome a little to get rid of blemishes.

For extra cheekiness, I gently peeled the original label off the lead-acid battery and then applied it to my newly made pack!

Converting the Futaba Conquest T4NL Transmitter

Here’s what I think is the bigger meat and thicker potatoes of this little holiday boredom project. A long (long) time ago (in a MIT far far away, at least now it is!) I collected a box of cruft which had a Futaba Conquest T4NL transmitter in it. It hailed from the era of “boxy and chrome” transmitters, and I always figured one day I’d shove a 2.4Ghz radio into it as a restomod. This was probably at least in the 2010-2011 timeframe, and it traveled with me through my multiple great cruft moves.

Well I think finally it’s time to do it! The idea came to me immediately when I brought the Omnibot home.

Like the Omnibot, it was in good mechanical and cosmetic shape, even if the batteries were long missing.

The radio of choice I was going to swap in is the “Microzone MC6C“. It’s one of many lower-end market 2.4Ghz basic model radios hailing from the vast ungoverned high seas of Jack Ma. I originally picked up two last year when I was putting Trashcopter together, since I wanted to see what the evolution of cheap radios has been like over the past roughly 10 years since I’ve owned two HobbyKing T6Av2s (which are still working, mind you, and exist for sale still!)

The upside of this MC6C radio, which made me select it over just getting another T6A/CT6B model to use as the transplant, was that it had mixing and servo reversing switches right on the front panel, something the FlySky/Hobbyking radio lack.

I’ve always been annoyed that you had to use a computer and USB programming cable along with a horrible VB6 application (or an open source one like DigitalRadio) just to do servo reversing – like mixes I can understand, but that’s some serious cost cutting to not even put reversing switches in. I’m hoping that maybe my exposé about these will help push them a little more mainstream in the robot community as a starter radio or a “class set”, like we used the HobbyKing radios for back in the 2.007 days.

So I went ahead and cracked open the spare one. There sure isn’t much going on inside the MC6C compared to the old Futaba. Miniaturization! Push the problem into software!

On the left side of the Futaba unit is the “RF Module”, so to speak. The larger motherboard on the right performs the sticks to servo PPM encoding. That encoding is done by a dedicated IC, the OKI MSL9362RS, specifically made for 4 channels of digital proportional RC PPM encoding. The combined PPM signal is modulated with the 72 or 75mhz carrier that’s generated by the RF module, and off we go to the races.

Before I started unsoldering things, I decided to do a little poking and prodding with the MC6C. See, there were unpopulated connectors inside on the main board, and unpopulated connectors mean “expandable features”. I figured I would see if the radio could be capable of more. By following traces, I determined that the unpopulated connector was an unimplemented Channel 7, and that it was 3-position (or potentiometer) capable, even. This was confirmed by putting a servo on the receiver’s Channel 7 as I fiddled the contact with tweezers.

I determined at this point that my plan of attack was to go a lot more in depth than the usual “Old Radio Revival” which seems to largely entail putting a modern radio’s RF module’s PPM input on the PPM output stream of the old radio. That would limit me to 4 channels only because of the fixed-function ASIC handling the encoding.

With the T4NL’s chassis containing a whole lot more holes than it actually had buttons and switches (as the same chassis would be used for several different models, such as 6, 7, and 8 channel ones, with dual rate settings or other functions), I decided to do it the hard way and fully embed the now 7-channel MC6C control board and transplant all the switches over, to make a retro 7 channel radio!

The problem to solve now was to decypher the wiring of the potentiometers that make up the joysticks. There was an additional complication, too. The MC6C is a modern computer radio and has digital button trims (click the button, move the servo center a little, save it in flash). The Futaba T4NL had analog trim potentiometers.

Separate ones, even. Most cheap radios that still have analog trims just have the potentiometer body on the trim lever so you can manipulate it separately, but it’s the same pot. This thing, however, had the separate potentiometers on each channel feeding into an analog voltage summation circuit on the motherboard.

The dual row of resistors on the right near where the wire to board connector comes in is the voltage summation. Between the three resistors in each junction are four pads, making the four channel voltages. Through some poking and probing, I discovered values and voltage levels for this circuit while under operation:

I dunno who else would do this, but if I’ve learned anything from this website, it’s that someone else will. So here’s the handy dandy wiring guide I’ve made! It’s interesting to note that the servo reverser switches just flip the polarity of power supply to the channel in question.

The next step is going to be deconstructing the T4NL’s joystick circuit and making it compatible with the MC6C inputs.

So here’s the wiring cleaned up into something that’s spliceable. I’ve made a few wire jumps to power the pots from the same wire, for one.

The next step was to discretize the voltage summation circuit of the mainboard onto the joystick’s trim potentiometers. They now feed the main pot via a high-value resistor.

What this does is let the trimpots influence the voltage output, but not by that much because the main pots have a much lower resistance and dominate the output range.

Through some trial and error, I found that a 3.4K resistor piping the trimpot into the main potentiometer gave me about 20% adjustment range in the signal, which is more than enough as a “Trim” function.

So the “hard electricity bits” of this conversion are actually done. By this point, I had 4 channels of joysticks which had known voltage output ranges.

The next steps were largely mechanical integration. I decided to take a hint from the T4NL case and stick the servo reversing switches out the back of the transmitter (Those 4 little slots were their locations). This area would interfere with the sticks, though, so I decided to hang them out of the battery compartment since I was planning on using a much smaller lithium battery.

The cut was made with a Dremel and a regular carbide burr bit in multiple passes to slowly whittle the line down further each time, until it broke through.

The MC6C reverser switch board used to be attached to the main board using some rigid standoff pins. Because of their move to the back of the radio, I replaced these pins with some jumper wire so they can make a 180 degree turn and fold over.

A little 3D printer nugget will hold the top of the MC6C board level, and its bottom will sit against the former telescoping antenna mount.

I also made two little nuggets to space the servo reverser board apart from the battery compartment’s rear panel. It turns out the switches stuck out really far into the battery compartment if I didn’t. This spacing gets them basically flush with the underside, enough to still manipulate.

See? As another cheeky relabeling, I removed the servo/mixer panel sticker from the front and glued it to the backside.

At this point, I had the transmitter powered on for power testing. Everything seemed to check out, so it was safe to proceed with adding the additional channel switches in. However, what I found was that the servo centers and travels were way off. Obviously, there’s not going to be any guarantee the potentiometers are going to be the same value, or in the same position at all between two radios separated by around 2.5 decades.

Here’s what the final integration looks like on the inside. I ended up removing the board spacers for the servo reverser board and letting the switches stick out again, after finding that it interfered with closing the case all the way. I transferred the switches from the MC6C case, and added a third 3-position switch. Maybe in the future this would be my flight mode select switch if I made some kind of retro-drone.

My last “Hmmmmm” moment was trying to sort out the stick calibration problem. What would I do if I were the enterprising managing designer/engineer of a board that had to take four analog inputs from potentiometers which weren’t guaranteed to be 1. in the exact same center position once mounted or 2. give the exact same range of travel, and 3. can’t be tested and calibrated individually for assembly/expediency reasons?

I’d make the final assembled device capable of running a calibration. My more expensive radios have this as a function, selectable from the questionably designed UI of the LCD screen. But what are the chances the MC6C just has a “Hold this button or close this jumper to run a stick calibration”?

100%, really. I noted a set of mystery jumpers when I first took the thing apart. At the end of the day, you want the factory calibration method to be accessible and easy to use for the assembly workers. A single big 0.1″ jumper is pretty damn out of place given everything else on this board is small surface-mount components.

So why not. I plopped the jumper in and booted the thing up. Immediately, it went into a flashing and beeping mode, which is obviously special. I swirled all the sticks and switches around a few times and left the sticks and trims centered, then powered it off and removed the jumper.

And that was it. All 4 channels now gave the same outputs for my test servo.

To really sell the look, I reached out on the robot combat Facebook groups to see if anyone had in the depths of their cruft piles a “rubber duck” antenna for use with these old 72/75mhz systems. It turns out someone did, and it was glorious. It looks extremely period correct when off, but of course once I hit the power switch, the orange and blue LEDs of the MC6C control board shine through the old battery meter.

Of course, this antenna isn’t connected to anything on the inside. The MC6C has a small internal PCB antenna, which I dropped behind the mainboard once it was mounted in the T4NL chassis in the same orientation. On my other MC6C (used on Trashcopter) I actually unsoldered it and added an external WiFi antenna using an RP-SMA fitting.

And that’s how we got this seriously #RadwoodBait photo at the January “Not cars and definitely not coffee” show. I’ll just say that the Omnibot is excruciatingly slow outdoors – it’s a fine speed for a house toy, I mean, but it took a good long while to get anywhere out here in the parking lot of the mall.

Plans for this thing? I’m not sure If I actually want to go back and try to engineer a movable shoulder joint. While it’ll be neat and all, this is a very reversible hack and it’s nice to just have. This project will, for the foreseeable future, just live in the house and be a thing I can pull out and demo. When the Radwood events come back around, I’ll bring it by!