[DIY] Harmonic Pin-Ring: a new drive type - 3D Print

 

Just stumbled on a new type of drive similar to cycloidal but that doesn't use disc pins at all, simplifying the design considerably: the TQ-HPR50 Harmonic Pin-Ring drive used on a Trek mountain bike.

 

It replaces the cycloidal disc with a pin-ring disc whose perimeter is a ring sporting both an outer and an inner profile with the same number of teeth. As on one side the pin-ring disc engages with the stator thanks to the eccentric on the input shaft, on the opposite side its inner profile engages with the output disc.

This means that the pin-ring disc's wobble is never transmitted to the output shaft. No pins & holes needed. Detailed (but somewhat confused) explanation on the PinkBike site

 

Regular cycloidal drive:

In contrast, the harmonic pin-ring drive has only 4 parts, besides bearings: stator, dual-profile pin-ring disc (in red below), output disc (green), eccentric input shaft.

Genius, super simple, no more faffing around with disc pins / bearings and their inherent play, weight, complexity.

 

So I had to try to reproduce that mechanism, of course, to better understand it. CAD and 3D printing to the rescue, and here is the toy version:

  

 

The 13:1 and 12:1 tooth profiles were generated via a Fusion360 script from woodenCaliper on Github and with the help of this video from Howey 

To animate the rotation in Fusion360, select the axle to 12:1 gear joint, click on <Drive Joint> and grab the rotation knob:

3D print settings:

•     0.4mm nozzle, 0.24 layer height
•     3 walls, 3 top, 3 bottom
•     25% adaptive Cubic infill
•     no supports

Before gluing the stator's front piece, make sure to align the dots:

The print worked like a charm:

The measured reduction ratio is 6:1 for the output shaft, and 13:1 for the pin-ring disc. So, experimentally speaking, it seems that the total ratio is (N-2)/2, referencing the TQ HPR diagram above. But this needs to be confirmed, see the discussion in the comments below.

This print isn't backdrivable, it immediately locks up. But I suspect it's because of the high 13:1 intermediate reduction, no bearings, and abysmal friction & losses from my print. Indeed, regular cycloidal drives can be backdriven if their reduction ratio is low, which is a feature used in robotics. So I suspect that a proper metal + bearings version of the HPR might be backdrivable too, but likely not above 4:1 to 5:1 ratios ?

Another useful characteristic of this drive is that the input and output shafts can be on the same or opposite side, as needed by the application:

Also, if the output shaft is held fixed instead, the casing starts rotating, with a 7:1 reduction ratio, which can be very useful for some applications.

A disadvantage, of course, is that the reduction ratio is about half that of a similar cycloidal pin-disc drive, which here would deliver 13:1.

But this architecture is also stackable: add an eccentric on the output shaft and it can drive another pin-ring, and so on. In other words: make the stator longer, drop in copies of the same pin-ring + output discs, all riding on a long input shaft, and boom: multi-stage reduction. Mille sabords, j'en suis tout ébaubi... 

Example of a ~30:1 stacked version: 

I find it astounding that ~250 years after the start of the industrial revolution people can still invent simple and useful gear drives never seen before. I can relatively easily invent complex new stuff, but new discoveries based on simple and revisited-a-million-times gears are something else. Kudos to the team at TQ Systems !

Initial patent this drive seems to be a clever variation of: https://patents.google.com/patent/DE202013012416U1/en 

Follow-up patent similar to the TQ HPR gearset: https://patents.google.com/patent/WO2019229574A1/en

One thing I don't understand though is why the initial patent describes in lengthy details the math and shape of the teeth / profiles. It looks to me like a cycloidal profile as it is based on the same eccentric rotation + semi-round pins principle. If a more efficient profile was discovered, surely it would have been filed as a separate patent. Maybe I missed something. Das ist nicht klar...

Would love to see research papers / analysis on this type of drive: efficiency, loss types, balance, torque, longevity, etc. Especially versus traditional cycloidal and planetary drives.

 

The Fusion360 CAD, STEP and 3MF files are on Github and MakerWorld

Feel free to copy, remix, whatever (non commercially)

Happy printing to all !