An eccentric wobbles a lobed disc inside a ring of teeth. One lobe fewer than there are teeth, so one input turn walks the disc back exactly one lobe.
The teeth are cut straight into the ring: one part, nothing to source, no teeth to index wrong. The disc profile is the same either way, because a tooth is the metal a pin was standing in. Between the teeth the valley is one arc tangent to both, so the whole bore is arcs meeting tangentially and a cutter can follow it in one pass.
There was a second ring here, with loose rollers dropped into half-round grooves, the way a bought reducer does it. It bought rolling contact instead of sliding and cost a ring of ground dowels, a groove each, and a housing that has to retain them - and on a printed drive the rollers were the part least likely to be to size. Gone, along with everything that had to ask which of two machines this was.
21 teeth of ⌀7.0 on a ⌀100.0 circle, giving 20 lobes and 20:1.
The reduction is exactly the lobe count, Np − 1, so it can only be a whole number — there is no such thing as a 20.5:1 cycloidal.
Eccentricity 1.4 mm, disc 8.0 mm thick, bored ⌀12.0 with 39.1 mm of rim under the lobes.
Curtate ratio E·Np/R = 0.588. Keep it between 0.4 and 0.8: the lobes cusp as it nears 1 and vanish as it nears 0.
Three diameters, one inside the next. Name a cam and a bearing of that bore is fitted, and the disc's centre bore becomes that bearing's outside - so with a cam named it is the bore that follows, not the other way round. Name a shaft and the journal bearings take that bore instead of the largest one that happens to clear the cam, which is what they do on automatic and what used to make the two look welded together. The shaft has to stay under the cam: the cam's bearing is pressed on and has to slide over the journal to reach it. The motor shaft is the hole up the middle, bounded by the shaft rather than by the cam, since that is the part being drilled.Stock bores are 3, 4, 5, 6, 8, 10, 12, 15, 17, 20, 25 and 30, and asking for anything else is answered rather than rounded. 0 hands a field back to the drive, which is also how to undo a size you have outgrown - none of these has a minimum, so a number that will not work is refused in words instead of being made untypeable.2 discs on the eccentric, 0% of one disc left unbalanced.
Residual orbiting unbalance 0% of one disc. The discs are forced 9.00° apart — half a lobe pitch for a 180° pair. That is not a choice: α = −Nl·ψ ties each disc's rotation to its own eccentric.
Each disc carries a group of counterbored marks in the web between its bore and its output holes: one dot on the first, two on the second, and so on. Count them to tell the discs apart, and put the counterbored face up — they are handed parts and the dots are the only thing that says which is which by eye. Assembled, the groups sit exactly 180° apart, and they stay that far apart wherever the input is parked — so if they are not evenly spaced, a disc is on the wrong cam or turned over.
Holes are ⌀15.00 mm for ⌀7.20 mm pins — oversized by exactly the orbit diameter 2E, so a fixed pin never binds.
Input turning at 120 rpm, assembled.
The parts come apart in the order they go together, from the motor end up: cap, base flange, eccentric, discs, housing wall, output flange, collar. Bearings and pins travel with the part that holds them, which is what says where they belong.
Parts, as exportedOne chip per file. Dashed ones are drawn and never written: ring pins and bearings are bought, and a printed one of either is how you get a drive that binds. Each part is a single closed solid, sections and all - the housing is its wall and the collar bored into it, the output flange is its plate, its boss and the pins pressed through it. The switches below still apply on top of this.
| Lobes | 20 |
| Reduction | 20:1 reversed |
| Curtate ratio | 0.5880 good |
| Disc outside ⌀ | 95.800 mm |
| Disc root ⌀ | 90.200 mm |
| Lobe depth | 2.800 mm |
| Pin-to-pin gap | 7.90 mm |
| Output hole ⌀ | 15.00 mm |
| Residual unbalance | 0% balanced |
| Stack height | 16.00 mm |
| Disc orbital envelope | 98.60 mm |
| Housing bore | 99.02 mm |
| Housing outside ⌀ | 115.05 mm |
| Housing length | 49.24 mm |
| Wall outside the ring | 7.40 mm |
| Mounting | 6 × M5 on ⌀107.6 mm |
| Flange outside ⌀ | 86.40 mm |
| Output pin length | 18.00 mm |
Eccentric 128, pins 6700, input 683, output RA8008.
The reduction is the lobe count, so it can only ever be a whole number — 47.5:1 means choosing between 47 and 48. Useful from about 8:1 to 120:1.
Nominal geometry, for machining or resin. Every surface moves in by this much, so teeth get thinner and bores get wider from the one number. It applies to STL only; STEP keeps the nominal geometry, because a CAD file that carries one printer's shrinkage is wrong everywhere else.