It’s serious Clocker time!

I’ve spent some time practicing my TIG wielding ability on small steel objects as well as reading back up on ways to make steel less temperamental (or perhaps more tempered). The results have not been scientifically verified nor even experimentally confirmed legitimate, but at least it looks kind of cool.

Last time, my TIG welding adventure left off with my little pins breaking off one by one because of my muscle-memory water quenching. I wrote those off to practicing the technique itself. Doing these pins was relatively easy, I think, because it didn’t actually involve trying to hover the tungsten over a path – just kind of wiggling it in place. Some kind TIG advice from friends also helped. After I went through 2 more scrappy shafts, I decided to do the real things:

Oh man, those just about don’t even need any machining to flatten the upper face. I had to cut short my desire to TIG WELD THE EVERYTHING such that I stand a chance of finishing the bot on time. These shafts were both preheated to the “nice golden brown” stage (260C or thereabouts), welded on 80 amps, then quenched in cold water. I then tossed them in the convenient MITERS tiny heat treating oven (interior dimensions of a 4″ cube) set to about 300C, for a nice blue finish, wherein they baked for about 2 hours before being quenched once more.

Now, it’s important to point out that there is absolutely no scientific basis behind the choice of temperatures. I don’t even know what steel these things are made off, past a crude scratch test with some known 1018 steel plate. The shaft itself is definitely a really really mild steel, and I’m hard pressed to believe the pins are that much higher carbon. Remembering some rumors and gossip on the CheifDelphi forums from my 2007 FIRST Robotics season when the Banebots P80s (back when they were just called “56mm” gearboxes) had carrier plates made of what was found to be essentially 1006-1008 (very low carbon) steel, I think I’m not far off base.

I based the preheat on the assumption that they at least resemble a medium carbon (.40% to .60%) steel in some industrial applications, as well as the tempering temperature, figuring that the mildest of mild steel shaft body wouldn’t do anything noticeable.

I really, really should take these things to the materials lab and do a real hardness test if they ever break. Even I’m a little curious now.

The next stage after my brownies baked to perfection was to machine the taper. Luckily, the taper turning attachment for the trusty Old Mercedes (the MITERS 1953 model South Bend 10L) has not disappeared since I made these shafts the first time 3 years ago. There’s not much I can say to explain the taper turning doohickey besides pointing to this video which shows a very similar conceptually (but different in layout) device.

I discovered after finishing these pieces that my previous tapers were not only not the size the drawing said they were, but were actually 2 different sizes. The gears apparently had different bores. Fortunately the bores were smaller than indicated, so I had to just keep making small passes while trying the gears on every once in a while.

Different setup, different machine, once again.

Tinylathe was used to make the retaining ring groove on the nontapered shoulder because its tiny adorable .040″ parting tool is precisely the width of one of the snap rings I’m using. The only parting tool available for the bigger machine was .125″ wide, which I really do not like – it’s really too wide for the machine’s age-induced lack of alignment and the tool holder’s non-stiffness.  I regularly machine on the big 10L and then part off on Tinylathe now.

Luckily, the “3 degrees” I eyeballed this time is the same “3 degrees” I eyeballed in 2009. I did quite a few verification fits to make sure the distance between the gear and housing was close to 1/16″. A little sandpaper-polishing was involved to very finely bring the diameter of the tapers down at the end.

Fully assembled and packed with not only grease but fresh gears!

I think this is the first time in 2.5 years that this gearbox has not made terrible grinding sounds when the lifter was being run.

Unfortunately, Clocker will only get more taken apart before it gets put back together at this point.

the upgrades

Several major changes will be happening to Clocker this time around, things which I should have done immediately after discovering it wasn’t doing the robot thing properly, but were neglected due to lack of roboner.

The first, and most simple change which could make it far more effective, is moving to 3 inch wheels:

Clocker was originally designed with 2.5″ wheels because I was still totally into flat robots then, and the ground clearance afforded by 2.5″ wheels,  half an inch, was enough for the Robot Battle stage at the time. Unfortunately, wheels get smaller over time, and stages also get torn up (on purpose) – last year, it got to the point where Clocker had too little ground clearance margin to even move around reliably. In a game where there are no arena walls, moving around reliably is like 90% of the match.

Going to 3″ “McMasterBots” 40A wheels will up the native ground clearance to 3/4 inch. While having your bot higher off the ground may make it more prone to getting wedged, it’s also easier to get away, and there will likely be very few wedges where I’m going anyway (cept mine of course)

I’ve designed a new hub which is made of a convenient stack of plates with a Delrin bushing in the center. This time, everything just bolt through the whole wheel instead of transmiting through the aluminum hub body and into a press fit as the current hubs are set up.

Going up in wheel size is going to require me to make some changes to the robot’s frame, as several elements were designed only with 2.5″ wheels in mind. The front ‘shocks’, for instance, have to be moved inboard more and the shoulder screw head can’t stick out any more. This is fairly simple – I’m just going to mill down that portion of the rod end. The ‘plate’ at the left which bridges the two front leg halves will also have a little chunk milled out of it.

The rear ‘corner block’ in the frame will also need to have its thickness reduced, since a 3″ wheel will be exactly rubbing on it.

These were the simple changes.

I’ve been dissatisfied with Clocker’s upper clamp arm since I built the thing. Two major reasons: First, the actuator I used on it is just a little 400-class motor feeding into a 20:1 Banebots 28mm gearbox (of which production has been stopped for several years, leaving me with no practical spares), driving a leadscrew. It has no real clamping force whatsoever, nor can it help the robot recover if it flips over while body-slamming an opponent (this has happened a few times).

Last, and worst of all, the motor is the hard stop for its own range of travel! The separation distance between the motor axis and the leadscrew axis, for some reason, was made as small as possible ,meaning the motor is the first thing to run into the leadscrew anchor by the main fork. This has happened once, and I literally bent my motor. This actuator was one of those things that I look at later on and can only go “…” at.

So why not just flip the motor around so it sticks out foward? Then it would have stuck straight up in the air above the clamping fork (and been the first thing to hit the ground when the robot gets flipped), or it would have stuck really far into the ‘clamping volume’ and been the first thing to land on the opponent when the fork came down.

Stuck between a rock, hard place, and its own leadscrew anchor.

All of this lead to a night or two of thinking of, sketching out, and just manipulating solid part models around a new upper fork and motor layout.

 

Here’s what’s going on. This is arrangement candidate number 1, top actuator with motor facing away from main pivot point.

Notice that the motor is actually Cold Arbor’s saw actuator? I decided this early on that I wasn’t going to make yet another new custom linear actuator when I had a perfectly workable one already kicking around. The 550 will add alot more force to the grip, and I’m going to use it as a chance to test out asymmetric current limiting on the Ragebridges – stronger release/lift current than grab/lower.

This arrangement was created purely as an iterative design step. It’s really quite terrible. First, the motor’s really exposed and if I grab an opponent, an edge of it might just punch my motor from below. Second, the center of mass of the very heavy actuator is quite far out from the fork’s main pivot. I’d want it as close as possible in the best case.

Finally, notice the much deeper fork. Clocker was designed pretty much to grab only flat, boxy pushybots, so it really could not grab anything over 4 or 5 inches. Well, there’s lots of bots that are 6″ and taller, so the deeper fork is designed to accommodate them now too. This, too, was on its first iteration and the following screenshots will show little geometric changes.

Part of the reason Clocker couldn’t grab taller bots was because the clamp could only open so far before, again, the actuator ran itself into the leadscrew mount balls first. I aimed to resolve that problem in the new geometry – here’s a shot of the new top clamp beign able to reach about 55 degrees of swing. This is actually physically impossible because the clamp arm would run into the main fork axle parts – but it can reach 45 degrees with no problem. Previously, the maximum angle attainable was only about 35 degrees…

Here’s arrangement candidate #3 (Number 2 was just a variant of #1 with the motor turned around). Now we’re getting somewhere.

The motor is mounted very low and aft, meaning my C.G. remains where it should be. The motor no longer travels a significant amount linearly when raising or lowering the clamp arm, since it is mounted to the fork instead. The leadscrew anchor, formerly at the end of the egg-cam looking thing on the right (in the profile of the main lifter gear), is now directly installed on the clamping arm.

And here’s how far back the arrangement can push the clamp arm. I briefly considered adding an ‘indent’ to the clamp arm so it can partially overlap the pivot hubs, and this may creep back into the design later on.

The motor, while it does intrude a little on ‘clamp space’, can be better shielded from things landing on it in this configuration. I like it alot.

A better shot of how the clamp motor will be mounted. Notice it now has its own little roo-bar structure in front of it – this will be a ‘hard stop’ for incoming robots. It won’t protect too well against other robots’ pokey things, so I may turn it into a plate (or have it swappable with a plate) for very pokey robots.

The summary of these changes is:

Here’s what else has changed.

1. The outer set of forks has been made thinner to compensate for some of the weight of the heavier actuator. The outer forks on Clocker are currently 1/2″ aluminum and have seen pretty much zero damage.
2. The clamp arm itself has been lightened by several ounces, again to account for the weight of the actuator. It should be just as stiff as the current one due to more vertical thickness (where it counts). I put a huge radius on the corner where the actuator is pulling on it so it doesn’t embarassingly break off.
3. The two little grabby-toes on the end of the clamp arm have been replaced with one.

Why? I bought replacements for them yesterday and I hit the wrong part number on McMaster. The ones I ordered are 1.25″ diameter with a huge 3/8″ threaded stud. I’m definitely not using 2 of those, because while the thought of force-teabagging your opponent with double huge rubber knobs is hilarious, it would be impractically wide. Also, using 1 is lighter and allows me to save some weight from the most important location – the very end of the long arm.

timeline

NOW.

NOW.

NOWNOWNOWNOWNOWNOW.

I’ve already done most of the preparation machining such as preparing the frame for new motors and modifying the Cold Arbor saw actuator for its new home. Clocker’s pretty much been broken down into component molecules, and I just need to cut out the new arm and drive pieces. It needs to come back together by Monday night.

 

With Null Hypothesis pretty much squared away except for the occasional motor detonation, and no real leads on that regulator problem besides “wait more than 1-2 minutes before turning it back on”, I’m going to start fixing up my current ‘flagbot’ (like flagship, I suppose) Überclocker Remix.

And I’ll start off like I always do:

poor Überclocker.

Having been mostly neglected during my focus on building more silly vehicles than silly robots, Clocker has always been on the receiving end of last-week hacks and fixes to address only the issue that came up in the competition previously. In 2010, to fix the problems from 2009, I replaced the shitty 775-type motors with DeWalt drill motors. In 2011, to fix the issue of the DeWalt motors being used with improper torque transmission methods, I just press fitted harder. And in 2012, to fix the problem with the chain falling off now that the motors are relatively reliable, I’m gonna…

Fuck, I need to make actual upgrades to this bot. Changes that should have occured in 2010.

The actual story, though, started a few days ago when I wanted to spar Clocker against Null Hypothesis just to shake down the bot a little more to see what else might have been creeping up on it. Long story short, I couldn’t get the lifting fork to actually lift anything – something, it seems, was slipping before the big gear clutch. Having seen this problem before, I quickly decided it was the tapered gear shaft on the Integrated Dual Frankenb0x – the retaining screw probably loosened or stretched or something. But it was an excuse to open the robot up and examine it completely.

Often times, what you don’t know and can’t see is better off being undiscovered.

Alright, so here’s the bot. It’s just looking worse and worse too – dirt and grime has settled into all the little metal dings and gashes. Shiny new cuts in aluminum look ballsy and cool, but dirty oxidation does not.

Clocker is a great example of one of my bad habits of designing things very sequentially. In order to take out the lifter gearbox, I have to execute the following sequence of events:

1. Remove the top lid of the bot and take out the batteries to access 2 of the 4 mounting screws
2. Unscrew the other 2 mounting screw from the front
3. Take off the front suspension legs so I can
4. Take off the left and right side
5. Remove the wheel and drive gear to
6. Loosen the drive motors just a little
7. Remove the top plates of the electronics deck to clear the wires and give space to the taper-retaining screws on the lifter gears
8. Push the whole lifter assembly, fork and all, towards the back in order to clear the slot and tab mate
9. Finally, wiggle it out of the bot while making sure none of the wires that are threaded through holes get cut or twanged.
10. Un-press the bearing fit on the gearbox itself and push it out of its mounting cage.

Basically I unbuild the robot to replace a critical component. If I just dive in and go, it takes about 10 minutes to go one way.

And after all that is said and done, it pops out!

…with one of the bearings permanently fused to it. Hurray?

Ah, the IDF, one of my finest machining examples. Combining the parts from what must be like 6 cordless drills, it’s a twin 3-stage 216:1 reduction all made of shitty sintered iron and maybe steel.

When I test ran the motors still attached to the gearbox, something sounded bad. Really, really bad. Time to take those gears out…

Uh oh.

Gearbox 1 is missing all of the pins ever on its output stage. This is a known failure mode of the cheap drills when overtorqued, because the carriers are seemingly made of mild steel and the press fits just explode instantly, causing the pins to fall out. Some times, if the loads are light, wiggly pins are enough when backed by the intermediate stage to not cause much noticeable trouble.

Not good at all.

Gearbox 2, unfortunately, displays the same failures.

The funny thing is, I bet these gearboxes have been running like this since at least Motorama 2010 when Clocker might have experienced the highest shock loading forces on the fork. I’ve literally never touched this gearbox since when I closed it up in 2009 and now.

Well, tits.

I went from possibly having something which was still working now to either a few hours of machining to replace just those shafts (which might just fail once again), or cooking up a new and potentially way more expensive solution.

Time to take apart the gearbox some more.

I really liked the tapered shafts for these gears. They take some amount of screw pressure to transmit a ton of torque using the full width of the gear box (unlike, say, a pin or partial keyway), and in the worst case they just let loose and slip. They took a lot of effort to get back off – I had to carefully secure the gears by their teeth in a vise and tap on the shaft with a hard punch and hammer.

After getting the gears off and appraising the condition of the carriers, I spent a little while brainstorming about what to do from here.

Solution 0: Just remachine the damn thing from one of n drill motor shafts I have sitting around. Prone to failing in the same fashion? Yeah, definitely, because cheap drills are cheap drills. It would take less than 30 minutes because taper-cutter.

Solution 1: Machine a tiny Wubba-wubba drive! Prone to failing: Probably not. Probability of me actually getting it right: Ummm….

Solution 2: Overnight myself some Banebots P60s and emergency-redesign the whole system to accommodate them.

Solution 3: Attempt to weld the pins to the carrier plate, just the way they are.

I decided to try solution 3 first, since it involved 1. trying to use our TIG welder because this is a very precise and delicate piece, and 2. The shafts were already broken and could only get broken even more in the name of learning and practicing TIG welding, or fixed.

 

Say hello to the blob.

I’ve only used our cheap Harbor Freight TIG exactly once, and it was to weld the tungsten to the random plate of scrap practice steel 15 minutes after we set it up for the very first time. This was going to be fun!

Since MITERS is used to MIG and very few people (save for Amy) had aptitude with the TIG machine, it was a little neglected. I resorted to using a chunk of mild steel MIG wire as the filler rod, and it definitely took a few tries to not weld the tungsten. But after practicing on some more angle scrap, I began to like it better than MIGing. It’s quiet. Deceivingly quiet – just an eerie blue glow. There’s no disgusting splatter and crackling.

Next, I practiced on a scrapped drill shaft from some project I can’t recall, but it was in my little baggie of drill parts.

Taking Solution 3 can actually have very indeterminate results depending on what kind of steel the Shady Chinese Power Tool Co. Ltd. had lying on its shop floor when the shafts were made. Welding different alloys, especially a very hard high-carbon alloy like the pin (….presumably) to a low carbon soft alloy like the squishy shaft steel will most likely cause cracking in the heat-affected zone. Welding high carbon steel is just terrible in general.

But the pin steel, as tested with some pliers anyway, didn’t seem to be that hard. I’m not exactly surprised – given how shitty the shaft steel is, I wouldn’t expect nice hardened pins. I decided to preheat the whole mysterious assembly to a nice golden brown color (approx. 250 celsius)  based on this site’s recommendation for a medium carbon steel.

The results of my first practice run is shown above. It’s definitely blobby – I didn’t let the puddle form for nearly long enough. I started with a fairly low current of 40 amps, too, so it took a little while to reach even that stage. I might as well have MIG’d it.

Attempt number 2. I think I’m getting a little better at this. I upped the current to 60A, and it just looked better. The liquifaction starts at the top of the pin (where I aimed the electrode) and propagates to the body. Once the puddle became a few millimeters in diameter, I dipped in a little MIG wire.

Since the IDF runs with just 1 0.5mm shim of clearance between the carrier and bearing, I decided to machine down this weld to see if I made it reasonably far into the steel. Interesting effect – the pin’s weld puddle is much harder than the surrounding steel.

Here’s 3 attempts on 3 scrappy drill shafts, all machined down. The first one on the left clearly shows some porosity and lack of filler, and I’m really starting to like the 3rd on the far right. Overall, I’m liking Option #3 as a potential saviour technique for my drill shafts (as well as preserving them from being damaged in the first place). While it may be metallurgically unsound and may earn me cold stares from certified welders, I’m going to guess that any metallic inteference is better than a slouchy press fit.In the best case, the metals join reasonably strong and I up the failure ceiling a few more ft-lbs and I win. In the worst case, I’ve made a mushroomed metal head on the pin so at least it can’t just slip out backwards, and I…. win?

Enough practice – hows about some production parts? These are the two drill shafts after welding and machining. The one on the right I kind of skimped on surfacing a little.

Unfortunately, my joy was short lived. I’m now going to admit a slightly embarassing mistake I made with all of these practice pieces: they were water quenched. Like straight off being red hot and into the sink.

Anything I’ve welded in the past has been gigantic, ugly gorilla welds in mild steel which can’t possibly heat treat in any way, and dunking the parts in water just became second nature to cool them down quickly. Well, with alloy steels and medium to high carbon steels, I think that was a dumb.

The pin there broke off after I put two stages of planetary gearing back into the gearbox and turned it by hand. Sadness.

Alright, it looks like I’m going to have to remachine these shafts anyway. However, before I do so, I will weld the pins in place and then more gently cool them down, possibly even applying a proper tempering process. MITERS does have a tiny oven which has been used by some members to cast aluminum.

So, how’s about them TIG welders?