Archive for May, 2018

 

The Overhaul Design & Build Series, Part 4: Do You Want to be Gooey?

May 13, 2018 in BattleBots 2018, Bots, Events, Overhaul 2

Wasn’t that an insane season premier episode!? If you missed it, it even seems like they’re distributing the episodes in more creative ways this time, up and including Prime Video. That’s good news, including for me, who can’t be buttsed to TV like 99.5% of people near my age group and lower, and so can barely watch his own damn TV show. I’m fairly confident Overhaul will first be on the 3rd episode, so I think it will time well with the conclusion of this series.

The bulk of the physical construction took place around the first and second weeks of March. Actually, let me rewind the clocker just a little bit, back to the last weekend of February.

I got another shipment of stuff from Markforged, which is returning this season as one of the team sponsors. First, a bunch of Onyx filament to print wheel hubs with, as well as two large molds made on the Mark X series machines which have a bigger build volume. The Mark Two is limited to around 5″ in the width dimension, and guess what has 5″ wheels?! I printed a pair of 3″ front wheel molds in-house from Nylon, since that’s much smaller than the build volume limit.

Printing each pair of large wheel cores actually takes an entire day (22 hours, anyway) so it’s kind of a long process to make a dozen wheels. However, it was easy to pipeline everything once I got the prints going, as the polyurethane also happens to want about a day before demolding.

The resin of choice was Smooth-On Reoflex 60. I had plenty of good experiences with Reoflex 50 in Überclocker, but thought it wore a little fast and that Overhaul’s overpowered drivetrain would make that worse. So I elected to move up on the durometer rating, and 60A is similar to Colson wheels.  I got a small pack from the local distributor around here, Reynolds, to test my process and also the amount of liquid pigment needed. See, the native color of the Reoflex resin is a pleasant poop brown color, which is actually too dark to turn MIKU BLUE. So whatever, black wheels it is.

I’ve gotten a lot of questions on how the hell these wheels are supposed to demold. The molds are one-piece with zero draft, so it should be some kind of physical impossibility…. but then you realize that is what the screwy tread profile does!

I went light on the mold release here, and subsequent wheels actually popped out easier than that. Have I mentioned it’s also awkward trying to hold a camera at the same time as keeping yourself upright AND applying several Torques to something? At least a few torques.

They didn’t all work out though. The first center wheel mold I got from Markforged seemed to have some extrusion problems for the exposed surfaces, leaving them porous. We figured I’d just try slathering on the mold release as they reprinted it.

Nah, this one was stuck for good. Later on, I actually cut this mold open and discovered the resin had seeped entirely down through the floor of the mold and even through the inner walls due to its porosity… Yup, not unscrewing this one.

The reprinted mold was fine.

The problem with a robot with much larger wheels…. is that little sample pack pretty much only lasted those three initial wheels. So guess who now owns an entire gallon of goo? There is no intemediate size between the small trial-size and the full gallon.

These buckets are kind of crappy to use without dispensing equipment setup, but luckily I managed to get the workflow down for pouring them, and only got everything slightly gooey.

All of the frame rails now have brace plate holes-to-be-tapped drilled into them, so frame reassembly can begin in parallel with the remaining operations on new drivetrain and clamp/fork parts.

The first things to go back together are the liftgear and new lift motors.

One assembled front 3″ wheel… I’m liking these already.  The tread adhesion is outstanding – I can’t begin to tear the sidewall away from the face of the wheel. That and the mechanical over-molded interface means short of getting these things cut off, I’m not going to lose the tread.

A little bit more progress on reassembly, now with added drive motors.

Going on in parallel with the wheel casting and reassembly was lots and lots of welding. This damn thing almost has too much welding on it. I also know that I only say that because during this build, we didn’t have a MIG welder, only a TIG.

Here’s why – TIG welding is a very slow, methodical process which gives the welder maximum control over the weld composition. For the things we’ve been doing for “work” and consulting projects, this has been great! What it’s not good at is making large amounts of obnoxious fat welds quickly, for things which are only meant to run into each other over and over. Really, a lot of what you’ll see in the arms was designed for MIG welding, but I couldn’t gain access to my usual one back at MIT until nearly the very end of the build season. Putting Overhaul’s arms together, and Brutus’ wedges and plows, were processes which took up an entire day, or days.

The combined builds of Brutus and Overhaul made us go out and buy a MIG welder because of how bad it was. So that obviously won’t be a problem again, since now we have a Millermatic 211 in the arsenal.

However, I will begin with pointing out that a TIG welder is great for performing an act of terror I learned during my MIT career: TIG bending. Hey, it creates a highly localized heat region! By gliding the torch over a line scribed into some metal, you can very easily get it up to formability temperature. The upside is also a smaller HAZ than (in my experience) with an oxy-acetylene torch.

To make these bends in Overhaul’s future ears, I simply dumped 200 amps of TIG into them for a minute or so and then quickly threw them in the brake press. The welded-like appearance is actually very superficial and was a result of the metal surface liquefying somewhat.

The clamp side plates required some cleaning and standoff tubes machined. I actually didn’t have to buy any new tubing for this clamp design – all of it was either from some other tubular object on Overhaul, or could be slightly machined to the needed diameter. The machined tubes were advantageous since I could control the width of the assembly precisely using the turned shoulder.

SSAB’s Hardox comes with a paint-like coating instead of the heavy hot-rolled mill scale that I see a lot on generic AR grade steel. It comes off very quickly with a flap disc, whereas last season involved several hours of grinding with a solid wheel to get the material to clean weldable state.

Other weld prep included fitting the new lift hub pieces together – some diameters had to be cleaned up and shoulders turned once again.

I had to do a rather hilarious setup on the ears which connect to the clamp actuator in order to clean up the internal bore. Yep, that’s 4 out of 6 jaws.

Here I am doing the first assembly tacks on the lift hub. I have a very strange welding habit: I like doing my setup with the TIG welder, then switching to MIG to finish out. This is solely because I have no patience whatsoever for TIG.

Remember those little flats that were cut into the actuator ears and endcaps? Check out the parallels on the bottom – they help align everything so there is no complex fixturing needed.

Blah blah blah… welder and paint, grinder I ain’t, etc.

Because the clamp arm’s aluminum pivot rings still need to go over these, I had to clean the endcap welds up on the lathe afterwards.

A finished lift hub with endcaps threaded and with bearings made of oil impregnated nylon. I actually found a blank that I had machined most of to the correct dimensions, so making more was easy. I had more unfinished blanks which I machined new arm bushings for from also.

It was now Pi day, and New England greeted us with like the 3rd winter storm in 3 weeks. But the build must go on! Never give up, never surrender (seriously kids, don’t ever move up here. it’s not worth it. it’s expensive, shitty, and cold). I set out to Mid-City Steel which was able to quickly supply plasma-cut Hardox 450 parts on short notice and for very low ruble. Combined in this order are more parts for Brutus.

With this order came the first DETHPLOW (out of 2 – I entered a 2nd supplementary order for more spares) and all the arm parts too.

Plot twist: The arms are mild steel.

Yes, yes, finally obtain mythical Hardox sponsorship, end up making lifter forks from goopy mild. I was ready to design the arms to be made from HX450 also, but couldn’t help thinking if the arms were extremely rigid, that something happening to them would just take out everything upstream – the lift hub, main shaft, etc. which are decided not very Hardy or Ox-y.

Therefore, mild steel arms it is. Depending on how they perform in the season, this might be changed down the line.

Setting up the arms for welding was a similar process to everything else – chop and turn some tubes, and clamp it all together. I for one don’t mind if we bought a CNC plasma cutter. Before these industrial processes (which themselves are rather “old school” and established) were “discovered” by the robot community, welding a frame together was a much bigger deal and required much more setup and skill. This was the environment I grew up in, so that’s why it took me so long to learn and appreciate welding.

Here I am putting the arms together with our Miller 200 amp TIG in the foreground.

Hey, wait… That’s not actually me! That’s…

Allen, a new team mate for this season, who is a ‘graduated duckling’ of my involvement with New York Maker Faire and the Power Racing Series. These days, he’s a mechanical engineering student at Stony Brook. I stole him for their spring break and basically trained him from-scratch on TIG welding, upon which he somehow dumped the entire tank of argon over the course of the week.

First of all, it was a lot of welding, if I haven’t made myself clear on this front. But I do think the regulator was set up for too much gas in general, since at one point the flowmeter had something heavy run into it and did not work properly, and we set it up by listening to it. Sigmas! We have none!

(We do now have a new flowmeter)

 

Allen put together essentially everything you see in this build report that wasn’t the lift hub. This is a photo of the two Overhaul heads under construction. It was jigged up using the lift hub on one end and the spacers for the tooth on the other.

Your Godfather horse-head moment for this build.

Connecting all of the welded bits together was actually very painless this time. Think this means I’m getting better at design-for-welding! This is a test-fitted complete lifter assembly. Not pictured are the spare set of long arms and pair of finished short T-Rex arms. And the other lift hub. And D E T H P L O W.

Mechanical re-integration of the bot progressed quickly from this point. Check back in next week for more original content!

The Overhaul 2018 Design and Build Series, Part 3: When Your Supply Chain is Missing a Half-Link

May 04, 2018 in BattleBots 2018, Bots, Events, Overhaul 2

It’s fab time! The following story takes place in the last week of February to the first week of March – yep, that’s just over 3 weeks, reaching into 2 weeks, until the bot had to be in the crate! In fact, a lot of the earlier photos here were concurrent with finishing out the drivetrain changes and electronics module, since the mechanical modifications were first priority.

Alright, the first thing to do after two years of stagnation and one entire shop move is to take inventory. Some of Overhaul’s parts had wandered off into other projects or been repurposed, and others had been sold to needy builders elsewhere. Oh, and some things instead were to be deprecated and thrown out.

I was first out to ascertain how many good drive motors I had remaining. I originally built a whole bunch of spares and only ended up using two or three. But the lift motors at the time were low on stock at Hobbyking, so I only had four lift motors total – one of which was burnt out and the other had a severely damaged gearbox from (likely) the rumble and post-season shenanigans.

There were also other parts being counted – sprockets and drive hubs, and hardware relevant to the liftgear which I only had a few of to begin with.

By the end of the day, I had a good idea of what needed to be ordered from McMaster the following Monday, as well as what parts to ask HobbyKing for!

 

I began the stripping down of the bot during this process. Poor Overhaul – it’s mostly been living on a lift cart behind my work area, usually causing me to run into it in some way on a regular basis, causing a pattern of injuries on my leg which would have been highly suspicious during high school.

Breaking the bot down was important since I would want to start over with all new fasteners and ascertain the status of all of the parts, such as where a frame rail might need to get cleaned up or if this or that drive hub was about to let go anyway. There were also quite a few parts which were going to receive lightening in my efforts to make weight for DETHPLOW. Basically everything above in blue is being replaced completely, and a lot of parts which went into the now-previous design head will have to be deprecated.

 

Ah, the sad electrical box. Overhaul hasn’t been operational since a year and a half ago when I sold the DLUX 250 reflashed ESCs to Ellis for Robot Wars. To my surprise, this damn thing still powered on. There is a single brushed RageBridge inside to run the former clamp motor, an A23-150 sized Ampflow motor, as well as a BEC module.

Oh well – everything which caught fire is considered automatically sketchy in my book, and this box had an entire unassembled spare if I somehow needed another one – so I salvaged the bus bars and some intact wiring harness parts and unceremoniously chucked the rest.

Overhaul is probably my current masterpiece in terms of design-for-assembly and design-for-service. I put more thought into how things go in and out into this bot than probably every other project I’ve ever made, combined. It’s actually very easy to knock down as one person, since the majority of the bot is supposed to be serviceable by 2 people in under 5 minutes. It takes on average only 2 minutes to release a set of drive motors and under 60 seconds to separate the upper and lower halves of the bot at the arm towers, after which the set of lift motors comes out with only 4 bolts.

The revisions will see some of this go away – for instance, the frame rail brace plate would add a dozen bolts and a different tool to the process. However, I was fine with this – at BattleBots with the current format, you usually have several hours of notice before fighting, if not days. It gets tighter around the playoffs and finals obviously, but the quickest turnaround I’ve witnessed was still on the order of 2-3 hours. If I was scrambling to replace frame rails that hard, it means I’m doing pretty damn well.

On the plus side, the actuator and upper clamp retainment strategies have been changed to be more easy to service.

Well, when you get down to the basics, Overhaul is just a series of gears.

 

All said and done – this is what my table and bench area looks like. I sorted the remaining spare frame rails by type, since each type needed a different kind of surgery or modification.

Chibikart makes a cameo in this photo.

There was about 8 pounds to lose in the frame, spread across a few parts. Overhaul weighed in at 247.0 on the event scale during Season 2, so I used that as a cross-reference to the CAD weight of 240 pounds. To make sure DETHPLOW took me up to the same physical end weight in replacing the separate heavy wedges, and also taking into account the new frame brace plates, I had to get the bot down to around 230 pounds. The rest of the ‘missing weight’ was to be made up in the battery and ESC assembly being smaller and lighter.

Above, I use an annular cutter (also some times called trepanning cutters) to empty out some of the interior of the Epic Lift Gear. If you’ve never used these before, they’re like specialized high-precision hole saws for metal, and can cut a large hole very cleanly and quickly. They used to be very expensive and specialized, but you can find Chinesium sets now that work fine for under $100.  This one was driven in low-gear on Bridget.

The arm towers themselves also lose a bit of meat, with each side getting 1/4″ removed. In a mild perversion of their use, I actually used the same 2″ diameter annular cutter to make the circular boss by simply leaving the interior portion and machining the rest away.

It was now the first week of March, and life suddenly got much more exciting.

I had been scouring the country for a more consistent source of the 4mm AR400 steel that makes up Overhaul’s clamp and fork profiles. Basically, the material seems to come and go at McMaster, who also seems to source it from the depths of a tropical rainforest as all of the AR grade steel I’ve ever gotten from them has been covered in rust and not a single one has really been straight.

Through a lot of calling around local steel companies, I was given an inside sales contact at SSAB – the international manufacturer of the UK/Europe robot fighting circle’s preferred armor steel, Hardox. The gist of the conversation was essentially “Hold on, what did you say you guys were building? Let me talk to the sales manager and we will see what we can do”.

Only afterwards did I do some research and found that SSAB was essentially the U.S. Steel of Sweden. To be fair, I’m not sure what I was expecting, since it’s not like some small mom-and-pop operation produces 8 million tons of high strength steel alloys a year.

So there we go – I’d like to formally welcome SSAB Americas as a sponsor of Overhaul this season! This explains the big Hardox logo on the Equals Zero Robotics Facebook page now.

They not only straight up sent me a diced up 4′ x 8′ plate of 4mm Hardox 450, but also helped find a local steel distributor who had a fast-turnaround plasma cutting service for even more Hardox. This will come into play just a little later.

Alright, presented with several hundred pounds of steel, I will obviously go waterjet. I paid some hush money to the MIT Edgerton Center and popped out a few parts on the Omax 5555 in an evening. I started with 2 full “heads” for Overhaul and basically all of the kibbles which go into the new lift hub design, as well as one set of titanium brace plates.

Here’s the finished parts in the middle of some post-machining.

Back at my shop, I cleaned up the actuator-mounting bore on the clamp side plates. The actuator trunions will ride directly in these holes, so I wanted a not-sandpaper finish here. I didn’t have a carbide boring bar that fit my old boring head, so I reground the old HSS lathe rool I used in there and just ran it very slowly and gently to scrape the little bit of Hardox off.

Harder steels might be horrifying to machine, but they do leave wonderful finishes if you know their weaknesses.

I made a design change to the lift clutch which entailed cutting out some new parts. I wanted to increase the clamping pressure significantly because Overhaul had a lot of trouble actually hauling stuff over during Season 2. To do this, I made new thicker pressure plates such that the tightening nut could exert a lot more force without causing bowing and decreasing of the contact area. I also switched to a higher-friction clutch material (what McMaster calls their high-coefficient of friction), from a medium one.

Additionally, I made over a half pound back here by machining down the Epic Lift Pinion and clutch gear! Technically, that shaft also never ever had to be made of steel – 7075 aluminum would have been highly reasonabel… but I already had slugs of steel sitting around in the right size back then…

Just a couple of days later, the new actuator bodies arrived from my Chinese CM.

Oh. My. Baby robot Jesus.

Probably the most gorgeous assemblies I myself have ever designed and carried to fruition (if I do say so myself). The story of the gear-nut is a tragedy upon itself. I originally threw it at my Chinese shop along with the billet halves, expecting them to basically tell me to quit it with my English Acme thread in the middle of a gear nonsense.  I was fully expecting to just buy two bronze nuts and shove machined stock catalog gears around them.

The problem is then they asked me for a sample of the 1″-4 leadscrew, which is obviously not easy to get in China, as an assembly fitment. They already made one. What?

Absent the ability to mail a chunk of leadscrew (which I didn’t even order yet) in a timely fashion, we settled on the next best thing: I would send them a 3D model of the leadscrew, and they will SLA resin print it and use it as a fitment test. These gearnuts have less slop in the Acme thread than my Bridgeport does.

In the end, it was worth it! The gearnuts are probably the highlight of this whole operation – on almost every project and bot, I have something I call the “penising piece”, referencing the unfortunate tendency for us guys to put a lot of effort into something very showy and impractical for the sake of one-upping each other.

You know what I’m talking about. Don’t deny it. Think about what sport this is.

Too many parts and assemblies like that and your whole project becomes very out-of-scope quickly. Trust me on this, I used to do entire projects that way.

That is not industry terminology. Do not dare try to make it standardized.

There was only one “quality control” problem, and it wasn’t the Chinese’s fault. You see, the thrust bearings I specified on McMaster had a nominal outer diameter, which I designed the pocket for. It in fact was a full 0.02″ larger in real life.

I’m technically not even mad – it’s a thrust bearing. What kind of dumbass tries to make a radially tight-tolerenced fit on something like that?!

The outer shell of those bearings is just a stamped piece of steel – it doesn’t acually Bearing anything, it just vaguely holds the upper and lower races together so you don’t sprinkle the rollers everywhere.

Sadly, I had to bore out my beautiful Chinesium to accommodate the ingrates.

Here is how things fit together. The big thrust bearings sit directly on the face of the gear. The pinion shaft is retained via snap ring and has another bearing that carries it, located in the half not shown here.

And this is how it fits together. I’m quite thrilled with this unit! It ended up weighing about 1lb less than the fiasco I designed last time around, is much smaller and also capable of much heavier loads due to the large trunnion diameter and thicker leadscrew.  The rod end is threaded into the leadscrew and retained with two cross-drilled 1/8″ roll pins each.

Next time on Overhaulin‘ – lots of welding. so much welding.