Archive for the 'Project Build Reports' Category


The latest on the DeWut? project

Dec 15, 2012 in dewut?, Project Build Reports

It’s been a little while since I went “Make it your damn self!” on the DeWuts and left everyone hanging with the waterjettable pieces. Since then, the billet style design has evolved some. I’m proud to announce that it has been sent off for manufacturing by Sketchy-Ass Chinese CNC Co. Ltd., to return to me hopefully by mid January. This is a product which is in immediate need by robotland  ever since the old style 18v DeWalt “Team Delta” systems stopped manufacture, so, oddly enough, it might be my “launch product” instead of Ragebridge! Here’s what’s been going down.

This is the fully modeled design as of two weeks ago or so. As can be seen, I’ve actually bothered to model the DeWalt 3-speed gearbox! I’ve made the gearbox and motor available as a downloadable widget, if you want come up with your own design. The files are in Autodesk Inventor 2012 format, as well as a STEP and Parasolid.

While I tried to make a workable output shaft for the motor, I began to realize that it was perhaps more fruitful to replace the final output stage altogether. The 3 speed DeWalts have an advantage here because their antibackdrive (“that thing which makes it so you can’t crank on a drill’s chuck and have the motor turn”) system is very simple and planar. The idea would be to replace that ABD stage with a custom-machined ‘socket’ of sorts that wraps around the output carrier and has an integrated 1/2″ keyed bore, so in principle any 1/2″ keyed shaft can be used with the motor. If this is not clear from the above picture, then it will surely be elucidated by…

So, basically, the output stage planetary carrier has 5 little claw things. It’s easy enough to make a doohickey that wraps around those 5 claws. Normally, roll pins fit between those claws which are just barely smaller than the distance between one face of the weird decagon output coupler and the outer ring with 4 nubs on it (seen in the previous image). If you attempt to backdrive the drill, the decagon hub turn just enough to wedge the roll pins against the outer ring, locking the whole thing up solid.

This whole arrangement of course contributes much backlash to the system. While I could just say “take these 5 little derpy pins out”, that’s one more step in the instructions which, if not followed, would surely result in undesired behavior as the ABD rapidly alternates between locked and unlocked. A custom output coupler would also alleviate those concerns.

This is what the output coupler looks like, a 5-sided flower thing. In real life, this would be waterjet-cut from a high alloy steel like 4140 and moderately hardened. The shafting is a piece of stock McMaster 1045 steel shaft I bought to test fits.

The new output carrier pushes right against the inside of the inner ball bearing due to a chance alignment of English and Metric units. So, it truly is bring-your-own-shaft – the motor doesn’t provide any retainment force.

With this problem taken care of, I began addressing some fine details. With larger, heavier motors like this, face mounting screw holes are often not enough to keep the whole assembly planted under shock loads. A second set of mounting holes is provided at the rear to keep the heavy motor end anchored. These holes are designed to be 3″ apart and 1.375″ between centers. Why the weird dimensions? Because it’s compatible with a Banebots P80, just like the front mounting hole pattern.

This revision of the design also saw these little gearbox-retaining nubs on the inside, which help with setting the torque clutch tightness without having the motor installed yet. It allows more modularity in the assembly since previously the motor was the only thing pushing back against the torque clutch plunger (pressing on the spring steel wear washer immediately next to the gears, anyway).

The next logical step in the design was to combine the 5-sided flower thing with a shaft. This would fully constrain the output shaft, allowing direct coupling to a wheel.

Here’s what the whole thing looks like in mushy 3d printed plastic form:

This is the version I’m sending out to be manufactured. The integrated shaft is specified to be made from 1566 steel as-rolled 1″ (/25mm…) round, which should offer a yield strength in the mid 50s to 60ksi (400-ish MPa in Unamerican Units). So, the total setup if I were to kit this up would be:

  • Integrated output shaft
  • Output mount with 2 FR8 type ball bearings
  • Motor mount
  • Motor clamp
  • The nifty barrel shifter holder
  • 4 hex nuts to constrain the NBS holder
  • 4 long cap screws to hold the output and motor mounts together
  • 2 short cap screws to screw down the motor clamp
  • 1 set screw to adjust the torque clutch
  • A retaining ring, because retaining ring.

I’m wondering if I should make a version that has the “socket” output carrier such that the motor can hitch into any existing 1/2″ keyed shaft. The 5-sided flower thing will likely be available separately. I’m also going to pre-emptively make it available in downloadable form for your own waterjetting amusement (Inventor, ready-to-cut-DXF, STEP, and Parasolid). I strongly advise making it out of a high carbon or alloy steel for strength reasons.

For now, enough product development. I need to turn my attention to more pressing matters…

RageBridge Revision 3: Boost Converters and the Market Survey

Dec 11, 2012 in Motor Controllers

Continuing the story of Ragebridge’s quest to find a stable power converter design, I received the updated version 3 boards earlier last week, and got around to assembling and testing them over the weekend. I think RB is ready, at this point, to enter the release candidate & fine tuning stage of the development process. But first, I had an adventurous time remembering what boost converters did.

First, if you want to skip straight to the RageBridge potential customer & user survey, feel free to just click through.

To summarize, the RB project has seen me move away from a known stable (but not that vintage) power supply design to exploring other ways of generating logic voltage and gate drive voltage in order to push the lowest operating voltage down below about 18 volts. My general setup has been dropping a switching regulator on the battery input and buck converting to 15 volts (gate drive voltage) and then a linear regulator to 5v (logic voltage).

This arrangement has been reliable, but it pretty much limits the working voltage of the controller to greater than 13 volts or so. The switching regulator, a LM2594, can turn entirely on and pass the input voltage, but it has a 1.3 volt saturation voltage penalty. Add to that the other 1 volt or so of diodes (one before and one after), and the gate drive ICs turning off at 10 volts, means that operation down to that level or below is undefined.

My intention is to make this board also useful for 12v lead-acid and 3S lithium systems, of which there are still plenty, and hence I needed a way to make sure the board is logically determinate down to maybe 8 volts. The previous board version, as detailed in its post, explored an integrated buck-boost converter, the LT3433.

Unfortunately, while it could have worked in principle, it turned out the chip couldn’t supply enough current to run everything on the board. At around 20 volts, the output began sagging out of spec.

Turning my attention to a more serial arrangement of components, I decided to take the slightly less efficient (in principle) method of chaining a boost converter after a buck converter. In this case, I was going to convert battery voltage directly to 5v logic, and then from there, boost to 15v gate drive. This keeps the regulators (nominally) separate and out of eachothers’ affairs. The chip used was the LT3460, a small dedicated boost converter controller. There would be no more linear regulator, so I had to pile on the capacitors to keep ripple voltage lower and also added a transient voltage suppressor diode to the 5v bus.

That’s where the design stood as of 3 weeks ago, and the corresponding board layout is:

I pretty much followed the recommended layout of the 3460 exactly. This had better work. I sent out this design to MyroPCB, once again, but in 1oz only (since I was just out to test this converter, more or less) and…

in black. I’ve always wanted black boards – they just look so much more badass. As usual, I placed my Digikey order when Myro notified me of the boards shipping, but that seems to have gotten delayed, so my Digikey order appeared a few days before the boards.

The first thing I wanted to test was the power converter, so it was assembled first thing. It’s important to note here that I also bought a same-footprint-and-pinout equivalent of the LT3460, the MIC2288. The alternate chip has a much higher (1.3A) switch current limit than the 3791 at 300mA. This will become important later.

With the power converter testing out good, I finished one board fully and did a load test with a small 5v fan (0.15A) and a spektrum receiver (0.09 to 0.12A, because what?). This was just to make sure the thing could hold up under load. The gate drive, at this point, was not yet enabled (the receiver was in failsafe mode).

After that was done, it was time to drive motors. I remembered that I had Segfault’s old drive motors, which are CIM motors stirring a hydraulic brake a 27:1 Banebots P80 gearbox filled to the brim with thick grease. This thing draw like 15 amps just running freely, and an overvolted CIM can certainly push enough current  to scare most controllers. One of Überclocker’s batteries provided the bus voltage. Even though testing unknown controllers on a battery (a source of potentially infinite amps) is a bad idea, I did it because..

ouch. The sketchy switching lab supply has that as its output voltage waveform. I thought I was chasing down a noise issue on the board when I realized anything the scope probe touched was displaying this waveform or some linearly transformed version of it! Oops. I had forgotten that switching power supplies are usually pretty bad in terms of voltage ripple and noise, cheap ones especially.

It might be worth iterating to other aspiring secret board ninjas that you DON’T EVER USE A SWITCHING POWER SUPPLY TO TEST SENSITIVE LOGIC CIRCUITS. Linear power supplies, though lower in amperage, supply a much cleaner power rail that can be essential for correct operation of circuits, especially analog ones.

Anyways, after switching to the battery, I realized the noise issue was probably something else. The symptom was, like just about every disease that afflicts motor controllers, frequent and unexplained resetting. Maybe it was some weird positive feedback effect between 2 switching regulators? I tried scoping the 5v and 15v supplies to check for their stability, and it seemed okay. What was weirder was that the behavior was only present if the radio was on and the signal was valid. The board wouldn’t exit failsafe mode.

Leading me closer to the culprit was the fact that the behavior was basically the same as hitting the reset button. It took a little while of sitting and watching the 5V rail, but finally I saw a strange transient:

Wow. That’s my 5 volt bus dipping to about 2.8 volts, then flying back up to 6.5 volts. Something weird was happening.

The critical clue was seeing this transient occur whenever the board entered running mode (i.e. left failsafe mode). It also happened immediately after reset, before the program has a chance to load. The transition from running to failsafe mode was smooth, and by pure accident, I left the USB cable plugged in (while trying to mess with the code to see if I could get the reset to occur elsewhere) and noticed this transient did not occur. I could, in fact, “bootstrap” the board with USB power, then after it enters running mode, unplug the USB. And it will be fine.

It was clear to me now that it was caused by something suddenly wanting a ton of power from 5 volts. But what? The microcontroller doesn’t need that much of a load at the start.

The only thing that happens when the board leaves failsafe mode is that the gate drivers are enabled. Could that be the problem? I began to suspect that the sudden current demand of all the gate drivers enabling at once caused the boost converter chip to go crazy. To test this hypothesis, I ‘staggered’ the re-enabling of the gate drivers into 2 groups, separated by 5 milliseconds.

It was definitely the boost converter beating my 5V buck converter into submission now. The fact that there is a transient whenever the gate drivers turned on was the decisive piece of evidence here. What’s funny is, the board worked fine in this state. It seems the sudden pull from the 5V rail is no longer enough (since only half the gate drivers are demanding current) to reset the chip.

The explanation as to why the gate drives suddenly want a burst of current is that the first thing they do when enabled is turn on the low side FETs on all the H-bridges (i.e. feeding a pretty big capacitor in the form of the gates). The bootstrap capacitor for the high side also charges at the same time. A small dip in voltage occurs on the 15 volt bus as a result.

Because the boost converter is much faster than the buck – 1mHz vs. 150kHz, it responds by pushing more current through the inductor (briefly 1.3 amps worth at least, according to the MIC2288′s rating) and it does so quicker than the buck converter can respond. As a result, by the time the buck converter duty cycle catches up, the transient power demand on the 15v bus has already been fulfilled, so the boost converter backs down output current. Again, because the buck converter is slower to respond, it overshoots and then slowly adjusts itself again. That’s how I think it goes anyway.

So basically, I need a precharge circuit or something for my 15v bus. Well, one way to combat capacitor dumping is with more capacitors:

First, I replaced the small 10uF 16v ceramic output capacitor of the boost with a giant 47uF one. On this board, its brethren are buffering the 5V line.

This had very little effect. It seems that the problem is still the boost slurping from the 5V line when it has a sudden demand on the output (necessitating charging up the inductor with a burst of input current). So I moved it to the other side.

This made the transient less in magnitude, but unfortunately it was still causing the ATMega to brown out.

Facing another potential redesign of the board to add even more capacitors or something, I thought about other bad hacks like isolating the 5V power converter from the rest of the logic with a low-ohm resistor, such that even if the converter itself goes under a little, the local capacitors at the ATMega and current sensors, etc. will keep them afloat briefly.

Then I remembered the LT3460 has a 300mA internal switch current limit. The MIC2288 has a 1.3A limit. I began thinking that maybe putting a converter that can briefly eat 1.3A after one which can only source 0.5A (the LM2594) was a bad idea.

The LT3460 has the exact same footprint, exact same feedback resistor and compensator capacitor needs, and exact same output cap and inductor specs. Alright, why not… I’ll swap it in.

That fixed the problem completely. Even with the “staggered start” removed and the extra 47uF of capacitor gone, there were no startup issues. The transient on the 5v rail had declined to the point where trying to detect it was difficult with the scope’s noise floor.

Above is a picture of the 15v rail under switching conditions, when the gate drive is running. As can be seen, all of the FETs being switched on at once causes a brief dip in the power supply (on the order of 0.2 to 0.3 volts). Meh.

I zoomed way in on the 5V supply and AC-coupled the scope to catch the power converter’s actual output waveform. The 1mhz switching freq. of the LT3460 can be seen over the 150khz switching frequency of the LM2594. The 5V ringing is a bit concerning, but it doesn’t seem to be affecting the controller function.

In the interest of reducing this ripple, I’ve spec’d out a 100uF 6.3v ceramic cap  just for the 5V rail. It has the same package as the 47uF on there now, so it will simply drop into place. I’m also going to add more bypass capacitance at the ATMega and current sensors, upping it from 0.1uF to 1uF.

I waterjet-cut the updated heat sink plate design for RB and laser cut my remaining silpad sheet to make the insulator. This is basically what a stock RB will look like, without an additional fan mount or something. I’ve accepted a quote for punching out the heat sink design from 1/8″ 3003 aluminum, with a silpad die-cut piece bonded to it. The future is exciting.

Here’s a picture of all generations of RB so far!

I’m now going to throw this board into something like Null Hypothesis and ride it around drive it to obtain a reasonable robot drivetrain test for temperature and current draw.

In the mean time, I’d like to rally some help from everyone.


LandBearShark 2 is finished, but it isn’t snowing anymore.

Dec 05, 2012 in Land-Bear-Shark, Project Build Reports

Well, I can’t say that I didn’t already finish LBS2 like 2 weeks ago. I did, in fact, and it’s been around the block a few times already. But sadly, since then, there’s been no snow in the weather forecast. Meaningful amounts, anyway – a few flurries fell here and there last week, but what good is that? Do I really need to take this thing to New Hampshire!?

I was counting on some snowfall to let me combine a little testing video with the rest of the build, but given that seems unlikely to happen, here’s the build in its entirety.

The first step in rebuilding is to retire the old design. Here is LBS pulled out of under-table storage. A little dusty, but it still looks kind of glorious.

The big ATV style crashbar is not returning in the new design, because really, it was just too over the top. It was funny and contributed to the very unique look of the thing, but it also weighed about 10 pounds (made of surplus 1/8″ wall steel tubing…) and the mounting design was very much optimized for the old frame and I couldn’t quite make the design work with the new aluminum one.

Otherwise, electronics aside, many component parts of LBS are being reused. I don’t think I’m quite at the level of the Ship of Theseus yet…

LBS has been reduced to components. I took the opportunity to perform a mass cleaning of all the track parts because they had been subject to substantial dirt and grime buildup. The chains, especially, took a distressing amount of soaking in brake and carb cleaner and like an entire roll of shop towels. Chains are terrible things.

Its former frame was used as a collection bin for most of my other retired aluminum-bodied projects. I was intending on pitching it in a shop’s aluminum recycling dumpster, but ended up putting it up on Reuse (the free stuff mailing list) for people’s amusement. I figured that had I been a happy froshy bunny during this time, I would be so extremely excited, jumpy, and otherwise bunny-like if somebody posted a pile of waterjet-cut aluminum parts to Reuse that I might start building project of my own spurred by the simple fact of possessing these items. That was the plan, anyway. Paying it forward for the next happy bunny that bounces into MITERS.

(Some of it was used immediately in a productive fashion).

One of the first build exercises was spacer and standoff-making. Most of these short ones shown are parts that will go into the motor mount, suspension, and new bogies.

I’ve been gradually liberalizing from my usual hardline t-nutting policies in favor of using material more effectively when the design calls for it. In fact, LBS was one of my last major giant t-nutted plate assemblies (the other being Make-a-Bot). I actually can’t think of anything I built from scratch in 2011 and 2012 which made gratuitous use of plates where they weren’t effective.

I’m thinking it’s about time to update the How to Build your Robot Really Really Fast document I wrote some time in 2010 to be a more thorough treatment of various design for assembly manufacturing methods, rather than a cursory overview of standoffs and t-nuts.

With a few of the necessary spacers done, I turned my attention to assembling my Fake Andymark Gearboxen. I’m a fan of their very inexpensive (compared to industrial suppliers, anyway) hubless spur gears because for most robotics-related purposes, hubbed spur gears add unnecessary bulk and weight if you are just making cluster gears anyway, or using another method of power transmission like splines or direct coupling to the driven member. In fact I’m such a fan that I’ll plug them some more: SPUR GEARS!

(Some mixing and matching may be involved.)

The way my intermediate gear shaft goes together is simple. Take a hex shaft chunk, stuff a bushing into it (drilled and bored on tinylathe), then start piling hex-bore spacers and hex-bore gears onto it. A hex bore custom sprocket is in the middle somewhere. On each end is a bronze thrust washer to keep everything in check axially.

The whole assembly was purposefully made a few thousandths of an inch shorter than the length of the center standoff – involving shaving a bit of material from one of the hex bore spacers – that it could spin freely once lubed up a little, and kept itself in place. Bam, fast-build gearbox without machining complex shafts and retaining features.

I’ll also admit I am a slow convert to the hazardous, addictive, and self-destructive world of hexagonal shafts. They’re just so easy.

The two Fake Andymark Gearboxen completed. These have no mounting bolt pattern – they are hung from the center Big Shaft of the vehicle, and kept from moving by chain tension on one side and the chain tensioner on the other, upon which they brace against.

Onto frame assembly. Did I say I wasn’t going to make T-nutted things any more?

Nah, no way. I’ve not become that unprincipled. There are still a few holding the frame together, but they are no longer the majority contributor to frame rigidity. In this case, it’s pretty much just for making the right angle joint that will be backed up by long threaded rod-and-spacer preloaded columns (see Carly Rae Jepsen’s build style), which can add much more rigidity than an equivalent floppy plate span.

The bogie frames this time are much lighter in section, and maybe a little too flowy looking. The reason is that these never took subtantial structural loads anyway – recall that even in LBS version 1, they were hinged in the center at the Big Shaft. The load path goes from the rider weight into the Big Shaft, where it is met by reaction force coming from the ground, going through the track sprockets and into the track axles. But that load is expressed primarily in torsion (out of plane twisting) of the bogie frame sides, because the suspension is so damn stiff that it’s basically a truss member. LBS has always ‘sagged’ a little as a result of this torsion (think overloaded car), so there was no reason to keep using the huge heavy side plates.

The much lighter cross section of the new bogies saves about 3 pounds and cuts the design down to basically the bare amount needed to connect the dots in terms of mounting points.

Now it’s starting to look like some kind of cracked out Mars rover, or a kinetic sculpture. Everything gets slid onto the Big Shaft at once, secured by shaft collars. I switched to a 3/4″ diameter axle this time – LBS1 had a 20mm one, which is a dimension I have no clue why I picked initially, but 20mm shaft collars are espensive and I needed a few more of them anyway. 3/4″ was the closest size to 20mm that still allowed me some space to clear the CIM motors, and wouldn’t be bendy under rider weight.

After a brief game of Which One-off Spacer Is This?! I began to put the track wheels back on, and also slipped the motors onto the Big Shaft.

Without the outside bogie frames keeping the axles on center, slipping on the track itself was easy – the whole assembly just kind of bent in enough. I did have to spreader-clamp the axles to make all the screws go in, though. Overall, the track tension has increased over version 1, since it was known to be somewhat loose.

After the track pods were slipped back on, the project reached criticality. At this point, with the chains not hooked up yet, I was supermanning it and coasting down the hall – not very far, of course.

Moving onto electronics, I’ve punched together the electronics box and mounted the switch and cooling fan. Little rubber grommets have been installed in the bottom where the motor wires will enter. I also drilled, tapped the eventual RageBridge mounting holes and installed standoffs.

The box mounts to the frame using these little hanger hooks which interface with bolt heads sticking out of the electronics box. The arrangement is self-securing (the force vector of the box being loaded downwards tends to pull the hooks tighter together), or at least theoretically so – if not, zip ties will rescue it. The rear hooks were swung out, the entire box slid into position and rested on the front hooks, then the rear ones tightened down again.

Much of the wiring was recovered from the old LBS, and I made the RageBridge wiring harness to match it. Here’s one of them mounted to test for fit and wire clearance.

I took some time to finally repair LBS’s somewhat decrepit batteries. These were made 2 years ago and suffered a balancer cable short & fire some time afterwards. Since then, they have been wrapped in bubble wrap and duct tape, charged and discharged without regard to inter-cell balance.

To my surprise the cell banks were at most 50 or so millivolts out from eachother. While still alot, it’s quite a testment to the durability of A123 cells. Too bad the company itself… isn’t very.

I remade the balance harness, this time carefully routing the cables out the side of the pack, then wrapped the entire thing in some foam rubber with Kapton and fiber-reinforced strapping tape. After a night of balance charging, they were all leveled out and ready again.

The two RageBridges were stacked together with some 3/4″ tall standoffs, and linked via a small custom Y-cable going to the receiver. The bottom one powers the cooling fan through the 15V rail (which, incidentally, is going to be missing from RB version 3 to be replaced by a 5v fan output).

Each RB controls only 1 motor per side – the system is set up in mechanical parallel. Hypothetically, if one controller fails, the other can still move the vehicle at 50% power, but it of course depends on the mechanism of failure. If the failed controller becomes a short, then it would be very hard to power against the shorted motor, for instance. There was no intent of providing redundancy, just a convenient means of controlling more current than one Ragebridge can effectively put out.

With no custom software needed, it was drop the batteries in and go. LBS is basically a dumb ROV at this point, no different from one of the battlebots. The RageBridges were put into “mix” mode because the simple 2 channel Hobbyking radio does’t handle any of that fancy stuff.

After putting the other battery in, things started getting…. crowded. Batteries are retained on the bottom with a healthy dose of Velcro, and the virtue of being confined keeps them from jiggling around otherwise.

The wiring harness is admittedly a rat’s nest, but hey, salvaged wiring. It’s also be a good chance to test RB’s robustness under non-ideal wiring conditions.

Closing the top up… I designed this version to be way easier to service in case something goes wrong inside because the electrical box lid is removable through the top, after the board itself is removed.

And here’s the 98% complete shot. At this point, I didn’t yet receive my threaded rod to finish the two standoffs in the front and rear. Without those, the frame bowed a little when I stood on it. But it was functional enough for some superman-style hallway blasting.

There was one problem I discovered during this testing. The motors could exert so much tension on the chain that they were physically bending the rear bogie frame inwards, collapsing the hollow cutout and making the chain jump off the sprocket.

Well, that was dumb. The placement of the chain tensioner was pretty much in the middle of a totally unsupported span. I could, in fact, unbend it with some big channel-lock pliers. I definitely hit copy and paste a few too many times…

To address this issue, the rear bogie frame side was recut to be solid and the tensioner mount itself was made a little fatter and angled.

Here’s a picture of the bottom of the beast, showing the drive chain setup and the batteries. Still, missing standoffs (which have since been added).

I don’t have any testing video yet, since there hasn’t been exciting enough weather to do so. The new arrangement, however, has demonstrably increased torque and better steering response too. The Ragebridges are synchronous rectification drivers, meaning the motors exert a torque against any external changes in speed unless commanded to that speed, so the tracks have increased dragging ability on either side. It can alsofinally turn in place, even with rider weight on it. The top speed is right around 8 or 9 miles per hour.

Once the weather gets more interesting, expect some updates with videos!

The DeWut Motor: Next Steps, and Design Files for Making Your Own DeWalt 3-Speed Motor Mount

Nov 26, 2012 in dewut?, Project Build Reports

Alright, with the Long Weekend of Nobody Being Open coming to a close, I’ve pretty much finished the ‘version 1′ of this project and am moving onto another fork of it. As discussed previously, the DeWut is an effort to make the new generation DeWalt 3-speed hammerdrill gearbox actually useful. The ‘old generation’, according to my sources, has totally stopped production, and are getting harder to purchase spares for. DeWalt has been the foundation of so many robots I’m surprised they don’t straight up sponsor events.

The short story is that I am skipping over a waterjet-plate stacking version (… what? Are you out of your mind?!) directly to investigating an all-billet, CNC machined version. That seems a very not-me thing to do, and actually I’m going to build a few of the aforementioned waterjet-compatible mounts for my own amusement, but as a potential product, I am in the process of gathering quotes and potential suppliers.

The long story…

After completing the body of the first version gearbox, I went ahead and carved the future shaft out of some 1566 precision-ground steel stock. Luckily for me, the new generation DeWalt gearbox has a rather normal double-D shaped output adapter, so it was a short mill pass to turn the round shaft into the proper mating shape. Unlike the old style gearbox, which had a…. what the hell is that? Apparently a massive stress riser. This would be only a manufacturability test, because untreated steel is going to be far too weak to handle the torque output of the motor. Based on a few FEA torque simulations, to hold the torque of the motor at stall in low gear at 100 amps, the double-D area is going to see about 180ksi of shear stress.

Ouch. That’s some serious steel – comparatively few alloys can heat treat to the 200ksi+ range and not be brittle as glass, and we all know how my last adventure with “heat treating” went. This is a subject I’m not so familiar with and so will probably have to leave to the professionals when I make these into production parts. For my own amusement, though, I’m probably going to buy a sampler plate of steel rods, make the axles, and try heat treating the ends to various degrees.

I also came up with a new arrangement for the motor’s rear retaining plate. Previously, it only held onto that little nub that the motor’s tailshaft sticks into, and was fairly unstable. A more stable method is enveloping the ‘nub’ and pressing entirely against the motor’s endcap. Unfortunately, the brush holders have conductive crimps that end flush with the endcap surface, but can still short on metal. Hence, a ‘gasket’ of sorts is necessary, shaped to the endcap’s outline, and I’ve modeled that as the little yellow-beige-brown thing to be made out of a thin insulating material.

Additionally, the structure of the Nifty Barrel Shifter retainer has changed. It’s now removable from the side and the idea is to hold onto it with a nut from either side. The bolt slotting narrow a little before opening up to the proper center, so it ‘snaps’ into place and is less likely to fall off. This arrangement makes it easier to change gearing (if needed) but also allows the whole thing to be assembled before picking the gear ratio, since it does not have to be stuffed onto the gearbox while it is being lowered into position.

But the most important addition conducive towards kitting this assembly up is the alignment marks on the side. Starting with 1 little indent slot on the side, just line the plates up in incrementing slot count order. I figure the X shaped motor back end is obvious enough on its own. There’s 9 in the structure, then the gasket, then the motor retainer, for 11 total.

This is what the retaining plate looks like now. Of course, one option to bypass the need of a sketchy nonconductive shim which COULD conceivably become conductive by accident, is to make the thing out of a nonconductive material. Garolite (G10/FR-4 grade) was my prime candidate, but electrical grade fiberglass ought to work too. Anything in 1/4″ short of nonfiberous squishy plastics ought to be stiff enough to hold the motor in place.

Moving the retainer plate onto the motor endcap means the whole assembly becomes basically 6″ long minus the shaft length. A nice, even number to work with!

For the purposes of easy cutting, I arranged these plates into separate 1/4″ and 1/8″ assembly files with little “sprues” in between the plates. This ensures that they stay together and cut as one part, and can be extracted whole.  It’s a little more dangerous than individual parts, but machine time is billable by the second and precious seconds are spent cycling the nozzle, so continuous cutting is marginally cheaper too. For small parts, the sprues prevent them from falling into the tank.

However, I think that’s as far as I’m going to go with this design. Here’s why:

Huh. Well then.

That’s just for the 1/4″ parts! Note that I tiled 4 of the assemblies together, so the real cost is about $38 per set. Add to that the 1/8″ plates (another $20 per set) and the heat treated, custom machined steel shaft (estimating $30+ each), plus all the hardware and the laser-cut gasket. I’m basically looking at a hair under $100 just in cost. And only in quantities 10 and up, assuming I have like $8000 to drop on this right now.

As funny as modern digital manufacturing is, and as great as it is if I really just needed 1 of something, I think I can get a much better price by falling back to good ol’ subtractive machining. Hey, as it turns out, waterjetting in real life is expensive.

So, as previously mentioned, even though I could pop out a few for my own use, I won’t be able to introduce these as a viable product (in my opinion – feel free to differ). So what’s next?

DeWut? Design Files

I’m going to make everyone else do it. Contained in the above ZIP file is the two tiled parts to be cut (0.25″ and .125″ aluminum) and a 2D drawing of the gasket to be made from something nonconductive, in DXF version 2000. Also included are the original Autodesk Inventor 2012 3d models of the assembly and an exported version in X_T (Parasolid) which can be imported back into individual solids.

The assembly should only fit together one way, basically described in the first post.

There are a few #4-40 holes to tap in order to make blocks from the plates at first.

The mounting holes up front are properly sized to be threaded 1/4″-20, no more than 3/8″ deep.

The shaft solution, however, is left as an exercise to the reader (the shaft model is included).

The model as-cut should fit the new-style DeWalt transmissions (397892 series), possibly with a little stuffing.

There’s no warranty expressed or implied.

Next Stage

Based on my experiences sourcing the hub motor parts for Chibikart, I believe I can get billet mounts for the gearboxes contracted out through a Sketchy Chinese Dude With a CNC Machine for substantially less price. Leaving the thickness quantum domain also means I can make the gearbox mounts better fitting and have multiple threaded mounting holes, etc.

I basically began translating features found in the stacked plates into a single, solid model. There were optimizations for weight reduction made, as well as thicknesses and details changed to fit the manufacturing technology.

This is the end result. Now, it’s not totally free of potential laserjetted parts, because I’m keeping the Nifty Barrel Shifter holder. It will still rest on the tie rods that hold the thing together.

Check out that new motor mount. It’s a 2-piece clamp affair out of necessity because the DeWalt motors don’t face mount to the transmissions, they stick into them a good 1/4″ or so. Hence, the idea is to tighten together the gearbox and shaft portion, then stuff the motor in, then tighten the mounting collar. With a 1/2″ of support on the collar and additionally 3/8″ of support in the aluminum block, I think the motor’s not going to go anywhere short of having the output shaft shoved through it (by which time, terrible things have happened). This obviates the need for a rear retainer.

Overall, I’m looking at 3 aluminum machined parts, 1 heat treated steel part, and a derpy laser-cut thing. Because the NBS holder is not structural, I’m content with making it from Delrin or ABS plastic.

Incidentally, the mounting holes match a Banebots P80, at least on the front. This whole thing is basically a lighter P80 type motor with 3 speeds!

Of course, I’m not leaving the world of 3DRP forever. I wasn’t going to send anything out without a sanity check for the critical mounting dimensions. I turned, appropriately, to a Dimension 1200ES to produce these mostly hollow plastic shell representations of the gearbox. The Dimension printer is much better at making parts ….. on dimension…. than the Makerbot Flock.

Since it cost 3 times more than the aforementioned Makerbot Flock combined, I would expect nothing less…

And hey, everything fits! I verified that the little slots and cutouts to cater to the DeWalt housing were within reason – if they fit in slightly shrinky plastic, then they will be a little looser in aluminum unless Sketchy Chinese Dude with a CNC Machine is really that sketchy. The overall length has remained essentially the same – this was basically a direct plate translation, after all. The Nifty Barrel Shifter holder is not seen in this picture, nor are the #10 cap screws that will keep the two halves firmly locked together.

The next step from here is to pitch all of these parts to my favorite SCDwCNCM as well as (my new favorite snack) to compare prices vs. features. At this point in the year, I’m going to consider getting prototypes in before the holiday breaks a total not-happening, so expect some more news on these DeWuts in January.

By the way, this is what is inside an allegedly “empty” Stratasys Dimension ABS build cartridge.

…seriously? That’s like 50 feet. I only feel slightly ripped off.

Luckily, this filament is 1.75mm diameter just like the latest generation Makerbot Flock feed, so I’ve been making heart gears using the Replicator. Productivity!


The DeWut? Motor Project

Nov 22, 2012 in dewut?, Project Build Reports, Stuff

With the Thanksgiving holiday weekend in full swing (read: nothing is open) and with RageBridge revision 3 boards still out for manufacturing (read: nothing’s gotten shipped), I’ve been shaking up the old random project idea jar a little for quick, useful things that I can whip out over the next few days. Here’s one of them. It’s been baking in mind actually for a long time (like since Überclocker-original), and I really think it’s long, long overdue to be introduced. I present the DeWut? Motor Project.

This is a “new style” 12-18 volt DeWalt cordless drill gearbox found in their line of 3-speed drills. New, of course meaning a design that’s about 6-7 years old and not actually found in new DeWalt drills as far as I can tell. Correction! These are still used in the 14.4v XRP hammerdrill line (Models 984, 985, and 939) as well as the 18v XRP hammerdrills (models 988 and 989).

Photo shamelessly stolen from the Robot Marketplace (from which you should buy things). These gearboxes have seemingly never found a home in robot drives unlike the “old style” 18 volt drills, partially because they’re so hokey. The “old style” motor bolted to its transmission as a unit – hence making the whole thing relatively easy to mount, and for the longest time, the Team Delta style mount was dominant in the market and used in many different drivetrains. The “new style”, however, is totally dependent on being in the drill in order to be held together and function. Great if you’re DeWalt and aiming to vertically integrate and cost-cut or something, but it makes using these motors a total bitch. Typical solutions have just involved using the gearbox for parts and repackaging the ring gears (Überclocker ran a haphazardly built gearbox for the longest time).

I am probably one of the most persistent preachers for the Church of Power Tool Hacks, and DeWalt is kind of like the robot messiah because their equipment, no matter how hard to use, is generally well engineered. Recently, I’ve pretty much given up hope on the less expensive Harbor Freight class drills for anything bigger than a 12lber – cost cutting and reductions in material durability have rendered them marginally useful even for 30lbers, as I’m finding out in Null Hypothesis. With that option pretty much ruled out for future robots, I’ve been rumbling internally about coming up with a way to use these damn motors and gearboxes.

There’s two major design paths that the process of making the “new style” gearbox useful can take.

  1. As I mentioned before, just extracting the gears and stuffing them into a box of custom design. This is perhaps the most robust, but expensive and up-front engineering intensive cost, because it’s probably going to end up being a pile of CNC machined billets. It would probably be locked to one ratio (out of three) and optimized for space and weight saving. They already make this anyway (see RMP link under gear repackaging).
  2. Using the gearbox as a unit, Team Delta style, including the ability to use the nifty barrel-shifter gear change. The new-style gearbox no longer has any plastic ring gears molded into the casing, which is great. It’s steel rings all the way down, so the casing really just there just to herd the involute-toothed cats. It will be bigger because the case is so damn huge. There can be billetwork involved, but it can also be made from a pile of waterjet-cut plates.

It is probably obvious which way I’m going based on the above. Even if it’s an empty design exercise, I think retaining the ability to switch gears (if needed) is useful, and keeping the gearbox unit intact is also beneficial because you buy them in units, not by the gear, so if one takes damage or wears out another can be swapped in.

Incidentally, the “nifty barrel shifter” seems to have been the downfall of this style of transmission as compared to the newest DW offerings. They seem to have a reputation of falling out of gear or neutral-dropping under power, which probably causes shredding in short order. So whatever I cook up should keep the Nifty Barrel Shifter in place and stop it from moving or expanding. It’s also important to note that these drill transmissions arenot designed for shifting under power like a car transmission would be. They just crash gears into eachother, so shifting while transmitting torque results in sadness. The ability to retain speed selection would only be for switching between projects and duties, in my opinion.

So that’s the brainstorming part. The next step is the stare-really-hard-at-the-gearbox part, where I try to think of where I can use a plate of discrete thickness to retain the gearbox.

I had to start at 3 ends here – the front of the gearbox, the back of the motor, and the middle end, where they need to come together. This picture is already pretty late into the first revision proces.

I’m serious when I say ‘stare at the gearbox’ – I think I’ve managed to just look at the damn thing for 10 minutes straight. It must have appeared very disturbing to an outside viewer. What I’m doing while staring is literally running dozens of different design visualizations, mentally dressing up and stripping the gearbox and motor repeatedly. Consider it using a natural skill for the greater good of robots instead of… other things.

If a potential design vision looks good, I caliper it up and see if it’s even possible given my design constraints: using 0.25″ and 0.125″ plates, and ideally no other weird thicknesses, in as few of them as possible. If it ends up being bogus, I keep staring.

To the uninitiated, it looks like the most unproductive thing on the planet, but neural-net design is definitely a thing, and you can get pretty far with it. It’s also very, very hard to teach because it pretty much relies on I’ve done this before.

That’s pretty much how it’s going to go down. Notice that I didn’t even bother modeling the gearbox itself – I was just plucking dimensions from it as needed.

The constraint for the front end of the gearbox was that there had to be a feature to center it using a convenient circular shoulder, followed by a shaped cutout that prevented it from rotating using the features of the output carrier cap (the weird black endcap on the left in this post’s leading image). X = 0 for this design was actually that black endcap – everything is referenced off it and made to fit it. The octagonal shaped cutout is clearly visible in the pile of plates at the front in the picture above.

It was pretty clear to me that the motor would not be screw-mounted. I don’t even know why it has tapped holes in it – in the native 18v drill, the motor is just mashed a good 1/8″ into the end of the gearbox and retained by an external plastic structure. I had to replicate that somehow, and this explains the strange plate structure in the middle. It contains the weird Ferrari-sides pattern of the ass end of the gearbox (rightmost in the leader image), then bolts to a plate that actually grips the motor by its air cooling vents.

The entire thing has to be retained by tie rods, and that’s what the X plate at the back is for. Because there’s no way to secure the motor to the gearbox (the middle stack of plates is basically alignment-only), I’ll have to crank the whole assembly down using nuts and bolts.

Finally, in the middle, and intended to be adjustable back and forth, is the Nifty Barrel Shifter Holder. The intention is to pick a gear and then lock down the selection with nuts, keeping the plate in one place (and hopefully the NBS too).

That’s about as descriptive as I can get about the design process. Take every word up there and add in 10 seconds of staring at the gearbox between each one and that’s basically how it went down.


Because everything on this design is empirically calipered and eyeballed, I knew it would just have to get iterated until it worked. Sadly, I don’t have enough 1/8″ and 1/4″ aluminum on hand to just keep getting it wrong in metal, so laser-cut-something will need to suffice. Luckily, MDF is not only cheap but also made to exacting specification and tolerances (unlike, say, even plywood). I scrounged several panels of half-used hardboard (masonite, HDF, etc…) that was 1/8″ thick and decided to pile it twice in order to make the 1/4″ parts.

Laser cutting prototypes is a bit dangerous for duplication to metal, because lasers cut on the line and waterjets offset to one side. The advantage of the former is that holes are implicity larger, so you can totally make a part that fits over a protuberance (like a gearbox end) on the laser cutter and then totally have it fail in metal, because the more precise process means that your dimensions, which were wrong but compensated for by the laser kerf, are now totally wrong. I designed in some liberal over-estimates of kerf for these plates, so I was expecting the laser-cut parts to be super loose.

To minimize the kerf effect, I experimented with a few small changes in power setting such that the beam barely broke through the material, leaving the kerf almost zero on one side. A few test cuts later and I was ready to make the real parts. Splitting the 1/4″ thick pieces into two 1/8″ ones also helps with this, because it ensures that there’s at least 2 somewhat on dimension edges in the part.

Here’s a vaguely DeWalt-shaped cavity starting to emerge. The little socket on the right of the circle is designed to compress the torque-limiting clutch to almost all the way locked. The natural movement of the plunger that actuates the torque clutch is a little more than 1/4″, so it won’t compress all the way. If it slips, then I’ll have to add a shim somewhere.

The barrel shifter holder and gearbox stuff in next. Conclusion: Octagonal thing fits perfectly! That, of course, means I need to bring out the dimensions a little if I want the metal version to fit – the hole has to get slightly bigger.

While this assembly was occurring, I was thinking of how I’d write it up for someone else to do it. I’m probably still really good at designing impossible-to-assembly machinery, so if I couldn’t figure out a streamlined way to do this, something has to be changed.

Here’s the Hokey Motor Mount coming together. This part needed a little stuffing, meaning the tab that interfaces with the motor vent is too wide.

As I aligned the plates, I superglued them together to create a more monolithic piece. If the plates were shifting around, I’d have a tougher time confirming that my tolerances were reasonable. These two Dewalt-Shaped Cavities will capture and align the gearbox, and also prevent it from torquing. However, I suspect most of the anti-torque will still come from clamping friction.

Stuff’s starting to get piled together now. Seating the gearbox into the Dewalt-Shaped Cavity means the torque clutch plunger is pushing up on the ring gears. The motor must be clamped down in place to keep the whole thing together, again. Hokeyness.

With the motor in place, things are looking better.

I neglected to put any NBS-securing nuts on because… well.. I was out of 10-24 nuts. Completely. And so was MITERS, and so was the bin of alleged 10-24 hardware upstairs. What.

I could only find six #10-24 locknuts, and 4 of those were going to get reserved for the tie rod at the end. So, in the interest of being able to play around with the different gears, I left the NBS holding plate unsecured.

Whatever, here is the thing. Not bad for a day and some’s design and rapid prototyping work.

The shaft is probably not going to be made right now, because I need to figure out what to make it from first. The Weird Spline Output on DeWalt drill gearboxes has always been a headache for the robot world – usually you’d want a 1/2″ or 5/8″ output shaft to mount a sprocket to, well-supported by bearings. Problem is, that shaft has to get reduced down to a little 8mm flatted shaft in this case, or a very weird ~3/8″ 4-sided spline in the old style gearboxes. Hello, stress riser.

It will most likely be a one-piece machined deal from 1/2″ steel rod, with one heat-treated end for strength. There will be 2 retaining ring grooves on it in order to secure it to the bearings. For the final metal version, I’m thinking of just carving up some O-1 tool steel rod, but if I were to hire this out for machining, I’d probably order a higher strength alloy steel. A popular shaft steel solution for the old style gearboxes was ETD150 or 4140.

By the way, there ARE two bearings. They’re flanged FR8 for easy installation (and convenient 1/8″ spacing in between), so this thing should be able to take reasonable sprocket-belt tension loads or directly drive an overhung wheel.

The next step from here is to copy and paste it in aluminum. This thing is going to be chunky and heavy – 2.5″ square, 6.5″ long, and current weighing just under 2.5 pounds. Change the dense wood to aluminum and it will probably be at least 3 pounds. 2.5″ square is essentially the smallest it can get because otherwise I’d be dealing with very small wall thicknesses on the back end of the gearbox (which is already 2.3″ by itself!). No more 2″ bots for me…

Should this experiment turn out well, I’m strongly interested in kitting it up and offering it for sale through the site. One DW motor per side is a good match for RageBridge as Überclocker has confirmed. The plates would be waterjet-machined and shafting sourced appropriately, and the fastening hardware is like $10 on McMaster. Short of that, I’m hoping to convert Null Hypothesis over to quad-DeWaltWut? drive for even more pushybot horsepower, and it will also power the next generation Überclocker.