I made a precision gearbox - with NO GEARS.

If all you need from a

gearbox

is more torque and less speed, then a simple gear set is probably the way to go, but if you like it and absolutely insist on having five AIS CNC machining capabilities in your back shed, It probably will. Wanting a solution that takes care of the annoying backlash inherent in any toothed gear reduction system, those solutions cost a lot of money, so I designed and built my own zero backlash cial disc from scratch, but you and I know it didn't start Find out from That way how quickly this video is less about how you can build cyclical momentum yourself and more about an overconfident nerd spending 200+ hours charging headlong into an extremely complex and multifaceted exercise in technical design and execution, if you really do. want. build one of the things.

I have put all my sources in the description below. It's me. I'm the nerd. I go back to the video. As far as I know, there are two technologies that do a great job of solving this particular type of problem. which we know as reaction straight from the factory, stress wave

gearbox

es are the first and use a flexible inner spline ring and elliptical bearing to work their magic. The bearing forces the inner ring into contact with the housing, which has two additional splines. When the bearing is rotating, the splines act as wedges and the inner ring rotates slowly.

The simplicity of this mechanism is great on paper, but the material requirements for the flexible knurled ring along with other complexities in machining

made

my head hurt, so I moved on to plan B while it may seem and be more complicated, the The basic principle of a cyclic transmission is quite similar: an eccentric shaft pushes the cyclic disc against a ring of rollers and, due to the mismatched relationship between lobes and rollers, the disc rotates slowly driving the output pins as well, unless you We are putting together the most reconstructed magic wand in history. It is a good idea to add a second disc offset 180° to balance the vibrations.

Now I'm not an engineer but I'm learning and your comments are very helpful for this particular one. It helped me realize that when using commonly available Dow pins as rollers, all I would have to take care of are the cyal discs in the housing, which seemed quite reasonable at the time, so it was a cyal unit when I was trying to design a cycal disk. The unfortunate reality is that you can't just put a bunch of Lo in a circle whether you want to or not, fortunately the name offers a very useful clue as to the specific shape that a disc requires, but more specifically a cyal disc. what we really want and this should really be obvious is a troid.

I removed a mass as soon as I could, so I'm as confused by all this as you are, but suffice it to say that if you wrap a cycloid around a circle and an epicycloid. If you like, we're planning so ambitiously, you're going to get a shape with these abrupt transitions, and while that would work, I have it on good authority that using a troid wrapped around a circle and an epit troid if you want, will result in a working much more fluid. If you like the smooth running you should like the video and if you subscribe I'll know you want to see more once I know what I was actually trying to draw.

It was a simple matter of creating an equation-driven curve and designing a casing around it. I opted for a 25 to 1 reduction to maintain some speed on the output and keep everything a reasonable size. I'll explain what's going on with this mess of components you'll see later, but for now it's enough to say that Since I'm a perfectionist, I

made

sure I got everything right on the first try, so maybe not on the first try, but the second one was much better, for now we have to decide what not to do once I have my design. Grillo, shut up, once I had my design, it was time to start machining.

I started with a 25mm mild steel plate for the casing, cutting it to roughly the right size on this poor little imported band saw that was certainly approaching its limits. of its capacity, especially with such a fine-toothed blade, I quickly faced the piece on the lathe and then set it on the mill for a rather ridiculous setup. Once it's time to move the part back onto the lathe, I'll need something to hold on to. that is vaguely concentric with the roller seats. So the plan was to use the rotary table to mill the outside face nice and circular.

Yes, I know it's a CNC router and I know it can move in circles, but at the time I still had some major trust issues. about its ability to move in true circles just as I'm about to start a project, it requires circularity down to the hundredth of a mm, but first things first, this thing needs some holes. I started with a 3mm pilot hole because there are not enough pairs of underwear on planet Earth to give this machine free rain with a 7 1/2mm drill in 25 mil plate, then followed up with said drill and which is definitely an 8mm reamer and certainly not a bur. to adjust the holes, except these pins are slightly too big, so it's time for the most over-the-top machinist joke on YouTube.

The boring part, once I finished cutting the roller seats, I used the rotary table to surgically remove the debris. material in the center of the piece and added that circular reference I mentioned earlier, the cyclic discs were next. I was pretty sure that the tolerances on all the functional characteristics of these parts were equally critical, so I decided to create a locating fixture that would allow me to machine them all in one setup, preparing the material was just a matter of facing both sides. I'm drilling and drilling some holes, then some roughing and then finishing the holes with the drill head and the outside profile with a ramped outline to measure my progress. on the outside contour I made a stack of gauge blocks that fit perfectly between the rollers on the housing to find my target dimension, then I measured the cycloids from the lobe to the valley with the help of a pin and then I was able to gradually reduce the sock to let it until I had what I thought fit the housing well, I made note of that sticky spot, although it will come back to bother me later, then I went back to the band saw to get rid of the corners of the roller housing and make the next step was a little less painful on the lathe I switched to the forged jaw and marked the part that this setup and the interrupted cuts that followed were no fun, but that wasn't enough to turn me back even without talking about turning, maybe you've noticed . a new tool stand located at the top of the complex, which must mean it's time for a side project of the future, the future awaits, the future is.

I've now spent more than my fair share of time using a four-way tool stand in Los Angeles on work. I knew from the start that this Lae would get a quick turnaround. The only thing stopping me from getting one at first was the idea that I could make one myself and that could still happen because this is the cheapest example in the Allur style that I could find and it has its problems, that being said, once I got it gave a good clean deep ground and raised a proper mounting bolt, totally serves its purpose, mostly only comes with four tool holders, which means For the rest of this project I was using a hex wrench to change tools even more than Required with four-way tool post.

Now there is an obvious solution to this huge pain in my butt and that is to buy more tool holders, but I wasn't really excited about the options I was looking at online, both in terms of cost and the fact that I have some unique requirements that I would like. fulfill now, daring, I'm not going to try to squeeze the whole process of making a custom set of tool holders into this video because a I still have a cyc unit and a half to make B, it's been done many times and C PCB way has done all the machining By me, you may know PCB way as a leading manufacturer of custom printed circuit boards, but did you know that they have a wide range of CNC machining services, as well as from laser cutting and sheet metal bending to milling turning and 3 and five access milling?

PCV has the CNC machining capabilities to take on any side project. over the line so click the link below to get a free quote today what those tool holders must be now that I'm not an engineer. I wasn't very confident in my ability to accurately measure these dovetails let alone communicate that information through a technical drawing, so I made sure to leave a little meat in the files I sent to PCB, about half a mm. , this actually worked perfectly because when I scraped the tool, I have to say the posts on the tool holders fit perfectly.

I am incredibly impressed with these pieces, the surface finish is light years away from what I could have achieved and the Dov Tails look like they were cut on an EDM machine. That's stylish, the bulk of this collection are these extended stands based on the da13 smells but with a practical twist seen by adding threaded holes at the top and bottom. I've basically quadrupled the versatility of these holders, then there are some pretty standard boring bar holders and some not-so-standard oversized holders for those insert tools I showed earlier. I bought those cutters along with some good stock of this absolute drill press unit from a local engineering workshop that closed due to the madness of Sydney's current rental crisis.

I feel like I owe it to David to put these tools to good use, even if they look a little ridiculous in my little vice. I still don't have enough material or patience to make adjusting nuts for all the brackets, and despite getting lucky with this royal rescue from a hardware collection of course, I'm missing the only type of screw I need. I probably should have let PCB take care of them too, which brings us to the last bit of 4140 in the group, a solid mountable tool as you'll see later in the video I've had some issues with the surface finish of this lathe.

I'm sure many of you will argue that that's because I lifted The Thing up the spindle and rooted the bearings, but I'm not convinced and conventional wisdom seems to be that the compound is often the weakest link between the tool and the part. of work, so I'm more than happy to give a solid tool holder a try, but before I take this compound off the machine and while we're here in the future, I'll try with the new holders to cut the taper on this Tombstone blank for the fourth axis finished and maybe I'll make a couple more, but that's enough.

I still have a cyal and a half disk to thank. again to pcbway for sponsoring this video and helping me out once i did a final pass things got a little cool the idea was to create a space for the fat to live but the reality is i just created a bunch of strawberries to now we'll just add that to the growing pile of things I definitely won't do on the next try. Flipping the roller housing over in the drawer gave me access to the rest of its outside diameter. At this point I had the cycloids. and the housing and camshaft that I put together at the end of the last video and let me tell you, this was a dangerous place to be on a project like this.

I spent hours tinkering and a small part of that was denial, you see, I was. I wasn't getting the smooth cyclic action I had spent so much time fantasizing about and I wasn't quite sure why the shitty surface on the roller seats or the millions of tiny burs next to them was because the cycloids were too big and That the exterior walls were not square should have been the turning point, but after having spent quite a few hours machining these parts, not to mention the time wasted fiddling with them, I decided to go ahead with it.

You see, I had one more theory that the discs simply weren't constrained well enough and that completing the rest of the case would magically solve all my problems in retrospect; It was all of those problems and certainly a few more that were equally culpable for that incurable condition of current cyal conceptions, so don't worry, I'm not going to spend much more time on the rest of this failed attempt before we get to the good part. . I'll just skip to the end and point out the rest of the obvious when you think about it. flaws in this design due to the material limitations I imposed on myself in the form of a 10mm plate.

I didn't have room to mount bearings, so I convinced myself that a plain bearing built directly into the caps and preloaded by the outlet pinsThey were a workable solution, but getting the pins to the right length was a real pain in the ass, and that was because I'd done a good job of convincing myself that it didn't matter if things weren't exactly as they appeared. in printing, as long as I could machine them to fit later, I also didn't really have a plan for how I could use this. I knew I wanted to build the axle separately, but I really had it.

I didn't spend any time thinking about how the two would connect at the end of the day, all of these problems boil down to a root because I rushed the design and left a lot of the problem solving to my Future and now I was my future and Sucked, definitely I have a certain propensity to hold on to first attempts and failed attempts, but on this occasion I could see that the continued presence of this part posthumously was only going to hold me back, there was only one place where it belonged: the box of Shame, as well as He completed his final journey.

I started mine. I went back to the drawing board and this time spared no expense, only feeling a slight remorse for raiding the stock and bearings I had been saving for a certain second-hand lathe project that was now down the road. Swapping the improvised plain bearings for Japanese-made tapered roller bearings was a no-brainer and should give the cycloids more than enough restraint to function properly, while building everything in a single package reduces the number of parts and opportunities for misalignment, oh, and me. I remembered to put the glasses on this time, which certainly helped, the first try taught me a lot, but I think the biggest and most important lesson was when to cut my losses, so when I broke that ream 20 thousand and 45 minutes into the casing, I didn't even I didn't even hesitate to start over this time.

Using a different strategy, I skipped drilling and jumped straight into milling. This hadn't been a great choice for my first attempt because of the solid core, but now I was living in the tube world, baby, following the 8mm with a six and then the boring. head resulted in an excellent finish on the rolling seats. Things got off to a great start at this stage. I had spent too much time suffering the rhythm of my tiny banso nibbling on a solid rod. This was just the excuse, I mean the project I needed. get a Band with teeth what the hell having acquired the ability to cut steel like never before my strategy for extracting the cycloids changed from plate to bar which in turn simplified the job by holding a humble Ribbon Chuck this time I really took advantage of the fact that This is a CNC machine that runs both cycles at the same time.

I wanted to mention that I switched back to a manual pull bar a while ago, so I'm not as afraid for my life anymore when this thing is running in the shed with me in the process. The way I measured and checked the discs remained the same, but this time I made sure to remove a little more meat and not skimp on those spring passes. Once all the grinding was done, I was able to cut them free of the broth and use this handy step to hold them down. to get them to the final thickness on the W, just a word of caution here, be sure to take note of the end of the cross slide travel and if it fails now, when powered at that end, definitely don't panic and put the lever on in longitudinal position.

Feed what fortunately the only consequence was a small slip of the belt and after a change of socks and underwear the evidence of that brief but violent episode was erased in a flurry of splinters back to the casing. I turn the outer diameter true to the roller. The seats added a small step that will locate the cover plate and clearance for the drive plate. It was a little stressful flipping the part because it meant it was time to tackle another new challenge, cutting a Precision B to within 10 microns or so. To give tapered roller bearings a cozy place to live incredibly well, I was actually almost prepared for this challenge in the run up to building the spindle for that other old lathe, stocked up on inexpensive ball gauges and micrometers and put together my own version of Robin Ren Zetti's great lathe. ball lapping tool.

I think it's also worth mentioning that I'm doing all this work on the lathe without a digital readout, which is neither a boast nor an excuse, but rather a cry for help, but rather the low speed application of these bearing sets. . the bar a little lower than a placed axle, making it a perfect first exercise, at least that's what I tell myself. I brought the board to within 20 microns of its target dimension, which will hopefully end up being a proper transition for the bearing cup. and then I added a clearance slot to the bottom and took care of the threads for the bearing cup retaining nut, after a couple of finishing passes on the outside diameter I moved on to polishing this aluminum disc that has a cone internal so it expands as you squeeze it. that nut, all you really have to do is run the turn up and down on the B, letting the piece rotate from time to time to average out the influence of your hands that you will feel when it is no longer cutting, at which point that you simply expand the lapis lazuli. pinch and repeat the process until you get the target dimension.

Luckily I didn't leave enough material to fully polish the B, but that's just a lesson learned and room for the glue to start on that shaft since it's my first time working with 4140. It took me a while to figure out what feeds and speeds to use, but once I dialed them in, I was breaking chips like never before, even if they were a little blue. I was almost starting to regret removing the safety guards, but I'm not sure I could have done much to stop the rain of burning razor blades - seriously, these things are failing everywhere.

I completely machined my first attempt together with a corresponding nut before realizing I had made it too long, oh come on, and without a good way to reassemble it. On the lathe I started again, this gave me a great opportunity to stop and really consider how I was going to mount the faceplate fixtures to the NSE spindle. I decided on a short 5° taper with a pick and mix of threaded holes in the face and this time I remembered to cut all the radial features in a single configuration to ensure concentricity. I also fitted this bearing mounting surface with equally uneven results.

It's really difficult to achieve that balance between surface finish and the amount of time you actually want to spend lapping both. the housing and spindle got their corresponding nuts, wait is this a nut or a bolt? I put a cone and probably a few more holes than strictly necessary in the drive plate. I transferred the six internal holes to the spindle, then drilled and tapped them M3. the six outer holes will accept the drive pins which are just some 8mm shoulder bolts that ended up on the wrong side of Henry VII and the remaining three will lock the bearing preload nut in place.

The last piece of the puzzle was a new crankshaft to match the new cycloids, my first attempt ended up with half the required displacement, joking, so I did the second one twice as fast and that seemed to work if you're wondering why I flipped the tool for the second L cam. Since my grooving tool looks like this with all the pieces complete, I'd love to say it was as simple as just putting it all together, but I love a good fed and I was going to make sure this fit and Work before identifying all the parts with Loctite grease and retaining compound, so only after half a dozen dry fits to check clearances, find the best pin arrangement and shoot anyone who knows anything about how clean they are. need to maintain the bearings in a

precision

application like This was the time for final assembly, the bearing cups and their spacer came in first with a healthy layer of retaining compound.

They needed a little tap to get the ball started, but once I lined them up, they slid with the Of course, with the least amount of resistance, despite my many practices, I still managed to get this one in the wrong place on the first try, that's the way on the plus side, the panic of trying to get the thing out before it was locked in place. always really made me feel alive at least I remembered that the front bearing cone needs to be mounted before installing the retaining nut. I became an idiot. I lightly packed the rollers with lithium grease, put them in the housing and locked the retaining nut in place by turning the end.

This time I installed the rear cone without retaining the compound, as it needs to slide over the axle to adjust the preload and install the preload nut. A few good taps tightened it up and eliminated any play in the bearings, the drive plate went in. This O-ring is then expected to prevent the bearings from any bits that will inevitably wear out the cycloids and is pulled into its locating cone by 6 M3 screws. The three set screws make sure the bearing preload nut doesn't get any funny ideas about how to come. loose and ruining my day the drive pins guide and a squirt of grease on the roller seats helps keep the 26 Dow pins in place for installation of the cyl discs the alignment needed to fit the discs really is a A little more satisfactory grease and a quick test to make sure everything works as it should means I can seal this nightmare of an assembly once and for all, at least until it stops working, the project was really starting to drag on at this point, like in the video.

So for now, I quickly printed out a motor mount for testing and connected the steepest motor to my top-secret electronic lead screw prototype so I could control it with this rotary encoder. I can't express how amazing it felt to finally have it. moving under its own Steam and that feeling lasted all 90 seconds until I decided to try the kickback at the beginning. It looked great with about 40 microns of kickback at 200mm, that's just over 40 seconds AR, unfortunately this number turned out to be quite variable. and depending on the absolute position of the shaft, I got up to five times more reaction, this is almost certainly due to a roundness error in both the cycloids and the roller seats, but hey, less than four minutes AR of reaction It's quite good and I discovered another potential problem that will be much easier to solve in positions where the play is greater, the cycloids obviously have a looser engagement with the rollers, so backtracking is possible.

Driving gearbox is combined with this, very cheap and very flexible coupler and I'm sure it will cause a bit of a false reading but that will have to wait until the next video when I hope to even make some parts with it to