What are Harmonic Drives and Just How Much Tooth Contact Is There?
Вставка
- Опубліковано 5 лис 2022
- How does a Harmonic Drive work, and just how much contact is there between the teeth of the splines? They’re great for rotary axes on multi-axis CNC machines and other robotics. Let's take a close look at how the beautiful Harmonic Drive reducer works.
Instagram: / mach_super
Blog: www.machsuper.com
Music Credit:
Ever Flowing by ItsWatR on Pixabay - Наука та технологія
This is fascinating. I’m only into this for the initial assembly part so far but I’m already blown away. I love things like this and my brain is fully engaged, zero backlash if you will. I’d love to talk sometime, have a desire to build a robotic arm for loading pickups and interested to developments in prosthetics also. I believe there’s some overlap in between these goals. Anyway, I know I’m going to learn so much here!
Landed here after the harmonic drives are now being used in astronomy mounts. Great video!
You are doing a great job, love your videos!
Thanks mate, I appreciate it.
Great stuff again man, keep up the great work!!!
Thank you Drew! I’m doin’ my best.
Thanks for the video!
I've known about harmonic drives, and had surface level knowledge of how they worked, but because of how much time you spent going over them and how you showed it, it now has me REALLY thinking about these. I've build my own "measly" 3 axis CNC, so i don't see how i would use one. I also made the same mistake of getting SKR rails at first. You live and you learn. Great videos by the way. Hello from Canada.
They’re fascinating to learn about, aren’t they? What do you use your 3 axis for?
Cool channel, already can see you getting more confident in front of the camera and improving your narration. Hope you'll get the popularity you deserve!
Thank you so much Bruno! The confidence does come with a bit of experience. I was never super nervous thanks to my experience in public speaking classes, but it’s hard to know what to say and get it out clearly and fluently. Getting better at that makes it easier to get back to acting more naturally.
What an idiot on the Prussian blue comment, probably an artist. Definitely not mechanic, machinist, millwright, mechanical engineer, mold maker or tool and die maker
I look forward to the scraping video!
Many days spent sweating in the shop for it, it’s a big job, but I’ll be excited to get it out!
Great video!
Thank you Stefan!
Mid end cnc rotary axes use cycloidal drives, they are much more approachable for machining/diy as well
Very informative, cheers.
Pretty much this 👌👌
You’re welcome buddy! ☺️👍
Being that it is not a zero tolerance, there is going to be some amount of axial load engagement which will not really affect the radial backlash.
Hi, its great job, its good video.
First, loving the channel of course. Can't wait for more videos. I didn't even know what a harmonic drive was.
Second, small critique. On any of your audio that isn't voice over there is a very noticeable hiss in the background.
Vocal quality is great, but some noise. If you use Premiere, the denoise feature at 25-40% should clean it right up.
Thank you for all the nice words, dude! This video did have crappy audio, my recorder mistakenly had the auto-gain switch on. I did everything I could to fix it. 🙄🙄
@@machsuper Ahhh no worries at all! Was wondering why I didn't notice it as much in the newer vid. Hey, still great quality overall.
We designed & installed these in plants in America.. mostly, Harmonic Drive AG in Germany were the best we had.
The Harmonic drive mechanisms utilize a strain wave gearing design, a flexspline undergoes elastic deformation induced by a wave generator, engaging with a circular spline to minimize backlash. This system ensures precise motion transmission, ideal for applications requiring stringent precision.
Ótimo conteúdo !
Thank you Jesse! ☺️
The harmonic reducer has a backlash! This is indicated in the technical description. The machines use either worm gearboxes or bevel gears with a backlash removal spring.
just read the specs, i think 1 arc minute accuracy is more than fine. forbthe application
some machines may use worm drives, but most likely use
-Harmonic drives (commonly used in the latter axis of smaller robots and some of my cnc machines came standard with)
-or Cycloidal gearboxes (heavier duty and tried and true in the robot industry).
Hi Ivan,
no the gear does not have backlash! The Wave Generator slightly deforms the Flexspline to a conical shape and the Circular Spline deforms the Flexspline back towards the Wave generator. So there is permanent spring preload between the engaged teeth which eliminates backlash, even for the whole service life. We call it the "coning effect". Of course you have to look also at the hub-shaft connection. A keyway connection has minimal amount of backlash (maybe this is what you find in the manuals), which is lowered a lot towards the output side by the high gear ratio. However there are backlash free hub-shaft connections like clamping sets, which are commonly used.
Sebastian from and passionate about Harmonic Drive.
@@unlockedcomposites7086my real machine uses a roller cam for the b axis.
I’m not sure what the 2 c axis use
yep backlash is small but it is there.
Good content, I wish I knew more about the software side of 5/6 axis home gaming. How do you do the CAM?
I've been using Fusion, but how do you ad a custom machine with all the variables?
Most of the work I did with the machine was years ago, but it was always with Fusion360. I have a video explaining it in about as much detail as I can here: ua-cam.com/video/nFT4texmPQU/v-deo.html
Also, i am looking for compatibility for motor-harmonic drive, its not clear explanation on youtube. If you make video about which stepper/ servo motor suit to which harmonic drive....
I worked in the web offset printing industry. We used these to control tension and to change timing while a machine was running. The older presses used one on the plate cylinder (image). The press gears compensated to the harmonic ratio so that when the wave generator center race was held stationary the plate cylinder remained in time with all the other cylinders and printing units. A stepper motor was connected to the wave generator and this allowed you to control (advance and retard) movement of the plate cylinder to control registration of colors being printed in the other units. Similarly, driven rollers that pulled the paper to control tension had a motor driving the inner race. You could control the speed of the motor to control paper tension as needed.
Cool video. I have a couple of these that were tossed in the garbage. There's nothing wrong with them... but they were suspect as creating a printing problem.
Thanks for the comment Winky!
I don’t fully understand your explanation, but is it in appropriate to say the harmonic drive was used as a trimming mechanism for the timing of the press to maintain correct tension?
@@machsuper I probably didn't explain it very well. Register in this case is talking about how all four colors line up with each other on a printer product. If a product has good register all the colors are printed in the proper position. The harmonics drives the cylinder with the image which can be advanced or retarded in timing. The most amazing aspect it that even though the inner race is held stationary the cylinder remains in time with all the other cylinders. The gears in the units are made to compensate for the 80:1 ratio.
Anyway... no need to understand this but it is a very cool application. These harmonics are used in many areas of a press for tension and adjusting timing.
Your videos are so well done! Please make more!!!
That’s Matt! I’m working hard on it!
I have no idea what's going on.
But I am totally here for it.
Nobody except nice strangers on the internet know what I’m on about 😄
So the flex spine actually deforms when the wave generator is inserted? How’s the stress on that part hold up? And is there a possibility of stripped or chipped teeth?
Correct, it’s interesting, isn’t it? They seem to hold up quite well, I assume the the Flex Spline is made from a spring steel of some kind to be able to handle however many hundreds of thousand or maybe even millions of rotations. Of course, the fact that it flexes is blessing and curse that allows it to take up all slack but also limits its thickness and thus torque capacity.
And I suppose stripped teeth is a possibility, that’s very much why I did this test, to see how much tooth engagement is needed to resist stripping at the forces I’ve seen these handle. I don’t know which part fails first when pushed beyond its limit. I’d love to know.
@@machsuper I would "guess" it's the flex spine would fail first. Also I could not tell from the video, but is there only 1 area of contact between the flex and circular spline, or is there another contact area 180 degrees opposite?
@@MrConcord75 Oh yeah, I’m sure the Flex Spline would be the first to fail, but I’m curious to know if the spline teeth would strip or the cup would twist and crumble, or maybe something else.
Yes, there’s another contact 180 degrees opposite at all times.
Cool video mate. Just one question I was thinking in... how do you prevent the outer race of the wave generator bearing sliping inside the flex spline during high load (torque/speed) aplications? Just by internal friction??
Gracias Pablo! I think slipping isn’t a concern. It’s not operating on friction, it’s only offering a shape deformation to the flex spline. You could in theory, make the wave generator a solid brass bearing that’s oval shaped and it would function basically the same. The biggest difference being that it would just wear much quicker.
Prussian blue just dried a bit. Oil can dry fast if it's cold weather.
What is the damage on one of those things? Where did you get it? I've looked at them before and find them an intriguing way of doing gear reduction.
OK found the part number on the website and looked them up. Ouch you're right they're not cheap.
Yeah they’re not cheap. I bought three of these ones as unused surplus on eBay way back when I built the 5 axis. I got a great price on them, I think it was under $700 each.
@@machsuper still that's a lot of cash to drop on a "it might work" DIY. Although it does look like a sweet 5th axis setup. I'm not a risk taker, so for me it would be super hard to take that risk. But sometimes you just need to take a leap of faith. Guess that comes down to how greatly you think it is going to succeed.
What kind of motors did you use?
Thank you for the answer.
Just simple open-loop stepper motors. I can’t remember the torque rating but 200oz-in would be plenty.
@@machsuper Thank you for answer.
I see that for example nema23 has two parameters of torque. For example peak is 290 oz-in, continuous torque is only 58 oz-in. When you say 200 oz-in do you mean continuous or peak torque?
Whats the name of the bearing unit opposite the harmonic drive? Is it also sold by harmonic drive?
The one carrying all the momentary axial and radial load? That’s called a cross roller bearing. They a super cool, ultra compact way of getting axial and radial rigidity out of a single bearing.
@@machsuper Did you make the bearing clamp that goes in the center of the cross roller bearing that is attach to the harmonic drive?
Yeah, it’s just a plain steel plug shape that goes in the centre of the cross roller bearing. When the spindle mount gets mounted, the assembly clamps up to the inner race of the bearing. The flex spline attaches directly to this plug to transmit torque.
At 7:40 seems like there was some play in the spindle holder. Did you notice that @machsuper?
Thanks for the comment Amit. If you’re talking about play in how it stops spinning momentarily while I’m spinning the input, that’s just because the circular spline isn’t secured properly. I’m using my left thumb to hold it. Look closely and you’ll see that when the output stalls, that circular spline starts spinning.
@@christianlewis7055 at 7:40 you briefly touch the spindle holder and it appears to move a bit. Wondering if that is actual play or the whole unit moving?
@@amitika2234 oh I see what you’re talking about. There’s no slack besides what I mentioned above. The whole lot can spin freely if I don’t hold it with my thumb, so when I grab it for that brief moment, I’m just rotating the whole lot. When the back cover is on, the circular spine is held in place by two aluminium friction rings which allow slippage in the event of a crash.
I like how u said good day. I believe
Haha thank you! I normally will say that.
What if I combine two of this Harmonic Drives in a line. Is than the Reduction 1:6400 ?
Yeah that would be the outcome. What are you doing that needs such a high reduction, high rigidity, and low backlash?
If the wave generator is tapered axially, it can compensate for the different amount of deflection.
I suspect that the outer race of the Wave Generator would already be tapering slightly under the load of pressing the Flex Spline against the Circular Spline. Interesting to think about though, would a tapered Wave Generator improve it? I’m gonna think about that.
i assumed the flex spline would taper thickness to account for this
Are they backdriveable ? I need to make Robotic arm for industrial use so if not backdriveable can the robot hold position after power off without brakes. All the load or torque will be within the rated limit of Gearbox and ratio will be 50
They are indeed backdriveable. With about 10-20kg of force at the spindle nose, the B axis could be rotated against the unpowered stepper motor.
@@machsuper i had talk with one of the supplier of SMD harmonic cup gearbox, and he said that it isn't backdriveable so no need of brake, maybe their is different.. .Thankyou for your help .
Hmm… well unless I’m confused about what backdrive is, I stand by my claim that they’re backdriveable, because I did it on purpose to this machine frequently just with my hands.
@@machsuper Yes had a talk with dealer and he said that huge force is required to backdrive it , hence they are backdriveable.
Thanks 👍
What is better planetary, cycloidal or harmonic?
Better for what? A planetary will generally be cheaper, but has the most backlash. I hear cycloidals are used on big industrial machines as they offer better rigidity, while harmonic drives will be lighter and probably have the least backlash. Sorry mate, the question is incomplete, engineering is about making compromises to favour a certain goal.
Have you considered planetary gearing?
Planetary gearing usually has too much backlash for the rotary axes on CNC machines. Perhaps there's some unique solutions out there that I'm not aware of.
how much did you pay for those two, i so far only found some really expensive ones?
I got a good price on mine on eBay many years ago. They were industrial surplus and I think they were a special order too, making them cheaper for me. I honestly can’t remember how much I paid, somewhere between AUD $300 and $700 each.
What did rotation stop at 7:47? 🤨
Notice when the output stops, the circular spline starts spinning. One of them has to spin, and for that moment, my thumb was slipping. Normally the circular spline would be secured better than just my thumb holding it.
You could have a 50% faster, possibly 50% better backlash prevention, by going with a tri lobe wave generator, and having three less teeth on the flex spline instead of two, since you would have 3 contact points hopping per input revolution.
So basically:
Less torque ( still a very high torque multiplying gear)
More speed
More stability aka
Less backlash
I’ve thought about these ideas. I think it’s conceptually good, but I’m sure the engineers at Harmonic Drive would have thought about this before. I imagine it’s just a cost constraint. It might be cheaper to make a larger unit than try to manufacture a trilobe wave generator. Maybe the metal fatigues too quickly from the harsher radii.
Or maybe I’m just hopelessly ignorant, genuinely possible.
would be cool to 3d print thing like that, will be so much cheaper
what if I need an RPM of 80-500? and lots of torque.
500 is quite a lot for a harmonic drive.
Yeah I’m not sure what to suggest for that other than a two-speed gearbox. Do you need the same torque at all speeds?
more or less, yes.
what I realized, I actually need is a direct drive motor. with 20Nm. so I could avoid a drive (even drives are sexy in their appearance.)
its for a design project and cost is an issue. I just stumbled over torque motors and direct drive motors that don't need a gear, if the motor is strong enough.
thanks for your reply.
do you know about Axial Flux motors?
@@machsuper
bmw e vanos uses a Harmonic Drive, i was so surprised when i saw one in real live use.
Oh really?! So interesting to imagine one in use in a car.
Eccentrically Cycloidal Gears
They seem super simple, why would they cost so much?
Precision stuff usually is. I’m not sure of the splines are ground or broached, but if it’s the former, I can totally understand the cost.
That plus I assume the elliptical bearing is rather expensive. Sounds like a quite difficult shape to create with any precision.
@@Bruno-cb5gk yeah, I’m not sure if the bearing would be made circular and the races made flexible enough to fit over the elliptical center piece, or of the bearing is made elliptical from the start.
Great video. Thank you.
By the way, I am concerning about when loaded situation.
All kinds of forces will come from every direction.
Don't you think that is why harmonics can't be used in heavy duty areas?
The final output of the gearbox should be supported by appropriate bearings for the application and the Harmonic Drive should only be seeing the torsional forces.
Harmonic drives are seemingly limited by how thin they have to be though, so I suppose very high load applications aren’t well suited.
Nice video. Good work on the dissassembly. Another way to achieve the rotary motion is with cam rollers. Tünkers have some large positioners with the the rollers aligned axially ua-cam.com/video/uar5Rr4zTiw/v-deo.html
And i think a lot of cnc rotary tables use radially aligned rollers ua-cam.com/video/9Mmaf8w1MkQ/v-deo.html
Look forward to seeing your scraping video
I do know that Haas uses these cam rollers in their rotary axes, not sure how new or proprietary their systems are.
In that video you linked, it looks like the pins are tilted mid-way between axial and radial alignment.
It's a great nice, simple concept, but still requires fairly complex machining to make the 'worm' gear component, even if you're not going to grind it, it's still complicated to manufacture that component.
@@machsuper Yeah true. I cant even think how to model the worm part! But it is basically only one really hard part, whereas there are a minimum of 3 difficult parts in a harmonic.
There's no easy way for either designs tbh!
How are you thinking to make the internal spline and oval bearing parts?
@@lawmate maybe even the ring that carries the roller pins would be hard to make within tolerance. The location and orientation of the holes that they screw into would probably have to be as accurate as the grinding on the worm to ensure they make good contact but don’t bind up.
I have no intention to make my own Harmonic Drives, but I have a couple ideas I’m working on in my head that use somewhat similar concepts. I’m also working on a two speed gearbox that makes use of one of these Harmonic Drives.
@@machsuper Ah cool, very interested to see what you come up with. They are a bit too expensive to use all over the place but are super useful.
In terms of the roller wheel, it wouldn't be too tricky. It could be made on a mill with some sort of rotary indexing. The diameter, depth and x & y positions can easily be measured and adjusted in place on the machine, so can be very precise. The angular alignment of the holes would be dependant on the quality of the rotary used, but as long as the overall device diameter isn't huge, angular errors won't create that big a deal. With the worm piece though, I think you'd need a CMM to make it properly. I had thought of making a little machine just to cut the worm, with a spindle mounted on a rotary axis, but I realised it would need an XYZ stage on that rotary as well to get good precision. I don't really have enough need for it atm. Maybe one day...
@@lawmate yeah, I agree, the location of the holes on the ring could be sorted with some thought and attention to the challenge. The worm, a little more thought and maybe money.
I have been thinking about how I might make a machine specifically for cutting worms like that too. I was thinking to replace the spindle with a lathe tool, though.
Comment for the algo
Cheers Rob, much appreciated.
is someone doing a hula-hoop whilst slowly rotating themselves a reverse Harmonic drive?
Ey?
@@machsuper yeah, watch a video of a woman doing hula hoop, and the hoop moves round in a segmented way like the outer ring of the harmonic drive. And then sometimes the lady will slowly rotate herself as part of her dance routine.
Lol Prussian blue. It’s just blue. You aren’t a painter or artist. The specific pigment of blue has no bearing (no pun) here.
How do you know I’m not an artist?
I'm pretty sure tool and die makers call it that
gay
No such thing as 0 backlash clickbaiter
Sorry man, I’ll write ‘approximately zero backlash’ next time.
@@machsuper What do i need if i want something with
Near Zero is the same thing. Extremely low backlash.
There are zero backlash systems everywhere in CNC machines. Actually there's interference (negative backlash so to speak) and the metal is elastically deformed and acts as a spring that keeps the backlash out. Some fresh engineers will get angry and call it impossible but that's just how the things are built
Great video !
Cheers Spastic! 😄