Harmonic Linear Actuator / Drive Ive been enamored with harmonic drives for things such

3D Printing
,
1

Harmonic Linear Actuator / Drive

Ive been enamored with harmonic drives for things such as robot arms. But it’s patented and expensive. Models exist on thingiverse that allow you to print one but after printing emmet’s design, it’s not robust.

Then I found this a couple days ago…

http://m.machinedesign.com/news/scanning-ideas-novel-recirculating-belt-powers-linear-actuator

http://www.animatics.com/images/library/hld-animation.html

http://www.igus.eu/wpck/2232/app_dry_zerobacklash?C=DE&L=en

http://www.linearmotiontips.com/animatics’harmonic-linear-drive/

It’s patented but a simple way to make a harmonic linear drive to power Cnc / 3D printers with very high drive ratios and zero backlash! Just belts, different sized pulleys and gears

http://www.animatics.com/smartnews/59-no-gearhead-no-brake-animatics-new-harmonic-linear-drive.html

Has anyone used this setup in a 3D printer or Cnc application? I will build one to test. It seems that it offers improved accuracy, specifically no backlash. The drive rations mentioned are pretty low nothing like harmonic drives for robot arms, but the hardware needed is at our fingertips for cheap. I will probably try first with steppers. But if it’s really no backlash, it could work with servos (closed loop brushless motors).

Ive dreamed of open source hardest/firmware for years, haven’t we all?! But lacking this, I bought teknics servo killers :

They offer closed loop in a self contained package and are quite nice. I’ve tried them with belts, but I was using dual y belts and their support said this was impossible to tune… Back to drawing board. Then we tried them with acne screws but the torque seemed to be lacking since my y axis was incredibly heavy. This linear harmonic drive setup would offer increased torque with zero backlash and looks like a good solution to couple with the tekinic servos… Maybe there would be increased accuracy and faster movement with steppers alone.

Since I don’t know if any open source electronics/firmware that run servos (I assume the host software / slicer would be unchanged), maybe this configuration offers advantages.

The use of belts on Cnc, with very heavy carriages/hardware could benefit from both the torque and accuracy. But there would be a limit to speed with steppers, I would think. I found video examples of the harmonic linear drive in use with a large Cnc using servos, though. I assume it saves money to use belts over acne rods and acme rods are practically limited in length to 3 meters from what I can tell… Shipping sizes, I presume.

So is anyone up to design and build a machine with this setup to test accuracy and other benefits?

The example diagrams show a rather low drive ratio. I know there is always a trade off in design… So I assume the higher drive ratio you use, the less speed you could have access to. But you would get increased torque. I suppose that finding a sane amount of torque / drive ratio for a 3D printer or Cnc would be the goal that translates into increased speed… To meet the limits of hardware and rigidity that you have on that machine.

I think Cncs will run up against the limits of cutting speed for a given router, spindle or cutting bit, but cheap, 3D printable robot arms always run up against torque issues first… Ok, and arguably rigidity comes into play.

Robot arms are an interesting study for these linear harmonic drives… Kuka arms do use harmonic drives. They need high speed (low torque) motors but require low speed movement with very high torque at the joints… Thus, the combination of fast motors or servos (closed loop using encoders at the motor?) and high ration harmonic drives for converting that speed into manageable torque with the magic of (near) zero backlash.

I assume that you can license the patents for harmonic drives, but the hardware is very tough to build and therefore expensive. I also assume you can license the patents for linear harmonic drives, but the simple, lower cost hardware seems to give it an advantage if you are looking for cheap, high speed motors to deliver the high torque at the joints due to the potentially huge drive ratios in a cheSp mechanism you can build at home.

So why has no one, to my knowledge, build a home brew version of this? Or licensed the tech for a Cnc or 3D printer? Since we use belts and it’s all about linear motion, it seems a good fit.

I found a 2009 RepRap post asking the same thing but it didn’t go far.

I also found the patent. http://www.google.com/patents/US8516913

Of particular interest is also fig. 12, 18, 22 which show other embodiments - although I didn’t understand if it was prior art/inventions or derivatives of the new design. All are interesting. They specifically call out the use of this applying to robotic arms.

Note: I saw a YouTube video of a Kuka robotic arm tear down and was surprised at the extensive use of belts. I didn’t investigate further, but I assume they are used to get the weight of the motors off the arm… Not to use this type of system. Anyone who can confirm this?

Thoughts?
Brook

11871d186837e8c625905920f0538d80.jpeg

2

Pretty similar to a belt-driven Z stage design I’ve considered in the past. https://groups.google.com/forum/?nomobile=true#!topic/3dp-ideas/V2BzFiNq-vo I think the big issue here for CNC is that no matter what the drive ratio is, you’re still limited in stiffness by the elasticity of the belts. A ballscrew is going to be vastly more rigid than even a compound belt arrangement like this.

3

It’s only a minor variation on a differential pulley. https://en.m.wikipedia.org/wiki/Differential_pulley
The wrapping around of the slack end to a second pulley on a carriage is at least 100yrs old, iirc when I searched on this in the past. I think the patent examiner dropped the ball. I’ve been thinking on doing this with a spectra line setup. Endless belts that are ~4X your stroke get expensive and moving mass grows quickly. I think I have a tensioning setup for line that won’t slip. I just have to splice an endless line and test it under load.

4

I think you are missing the nuance of this design and how the drive ratio is controlled.

There are two different diameters of pulleys used. And the two groups of different sized pulleys are locked together, not spinning independently.

Standard block and tackle with identical pulley diameters only adds torque, but does not benefit from speedier motion. It’s a hybrid block and tackle with the two ends being static and the linear motion happening between the two static ends at a different rate of speed (depending on the pulley diameter difference) with added torque.

And block and tackle never used a looped belt (rope).

As the RepRap commenter mentioned, the Chinese apparently used this many years ago in gold mines (weight = the need for torque)… But prior art has to be defended do here we are :frowning:

5

Another name I recall was “chinese windlass”. I’m very familiar with the many variants. I still question how thoroughly the examiner searched.

6

@Brook_Drumm We understand it. Check the 2nd diagram in the link I posted. It’s exactly the same thing, except with a slightly different pulley arrangement. (https://lh6.googleusercontent.com/-DkEGAaXBokY/VLg5zBnZzrI/AAAAAAAABCQ/NOqVC7LAeaM/s1600/Capture.JPG) The one you’ve linked is more elegant because it’s easier on the belts, but the concept is identical.

If you read the patent claims (always read the claims!) you can see that what’s patented is an implementation that specifically includes the angled orientation of the carriage pulleys. That was probably novel at the time the patent was granted. Easy enough to circumvent – just make the carriage pulley axes parallel to the driven and guide pulley axes. You’ll introduce a little bit of resolution distortion near the ends of travel, unfortunately.

7

I see a couple ways of doing this that can bypasses the patent.

  • use any other pulley than the joined differential pair to drive the loop. That pair synchronizes the two loops and converts the linear force of the actuator to torque between the pair.
    -eliminate belt twist, thus angled pulleys. It would require orher elements but is possible.
    I’ll dig into the referred and referring links when I have time.
8

I hate patents. Can’t make sense of such language in the filings;) that and they are lengthy enough to bite me, admittedly. I’m still going to try it. I need some better testing equipment (other that my gut).

Cheers
Brook

9

Is then the motion of the carriage per motor revolution equals to the teeth difference times pitch (large pulley minus small pulley teeth) divided by two?

10

@Brook_Drumm ​ that is actually intentional. The wordings are vague enough to make it super broad. Also the claims must be a run on sentence, by law each claim must be no more than 1 sentence and it must be novel… and words have stupidly different meanings… like “contains” means exactly these things, no more no less, but comprises means “at least this”…
So all of the rulings about verbiage since the first case pollute modern patent language.

Also if you are not selling the item, or an item derived from its use, or distributing the files for others to make these. You are not infringing on the patent. It only matters of you impact their market share (ie you compete with them in any capacity) personal use would be fine.

11

@Camerin_hahn In the US, any creation or use is infringement, regardless of whether it’s for profit or impacts the patent owner. There’s no exemption for personal, non-profit, or research use (except for a very narrow research exemption on drug compounds). If you only make a single infringing device for yourself, the damages would not be terribly high, and the lawyers probably wouldn’t bother chasing you, but it’s still infringement.

EU/UK is different of course. Dr Bowyer’s RepRaps would have been massively infringing Stratasys patents if he had been in the US instead of the UK.

12

Ryan, I understand the spirit of what you are saying but it not technically true- there is a lot of gray area in the patent system. It’s broke in a lot of different ways. But instead of debating stuff I know very little about and have very little interest in…

I’ll just say: I think harmonic drives are cool and have neat advantages over gear trains or simple looped belts.

Btw: Conical drives are cool too (like the davinci hub for bikes, etc!)

13

What Ryan said was confirmed by my conversations with a $500/hour patent attorney, and all of my research on the subject.

14

Those in the EU will happily give it a go and report back :slight_smile:

15

I obviously don’t endorse anything illegal. And recommend taking your lawyers advice. I’m just saying, if I buy from the manufacturer, it’s my right to reverse engineer and if there are no known infringements, I can build what I want.

I know businesses that build whatever they want until they are sued, I hate THAT. But there is something to be said for building something that is different enough but inspired from a patent… And open sourcing or protecting that invention in some way.

I generally don’t let patents keep me from innovating, I see them as design challenges. I’ve had patent office employees at the highest level tell me they know the system is broken and they hope for change. Until that happens, my hope is that innovation continues… Somehow!!

Also, I’m happy to license if there is a real benefit. I just hate patents that block obvious things (automated build plate, anyone?)

16

Yeah, the ABP is my go-to example for a bad patent because 1) the inventor didn’t want it patented but his employee agreement with MBI allowed them to patent it, 2) it’s super broad, and 3) MBI abandoned the concept so it’s not benefiting anybody. The original kapton loop conveyor admittedly had major issues, but the patent is so broad it prevents people from innovating with alternate approaches.

17

@Brook_Drumm
I don’t like patents much either. I never persued anything on my own as I lack the funding and stomach to do so.
But my day job (we build airplanes) reqires me to submit inventions and work with patent attorneys and examiners. The legal language is very specific, where a misplaced comma can sink a claim. Kinda like programming that way, syntax and proper use of variables are everything. The attorney skill ranges from rockstars, hacks, and everything in between.