Anyone here tried the cheap Chinese harmonic drive 4th and 5th axis off ebay?

Last time I asked about cheap Chinese rotary axis for diy builds there wasn't anything decent on the market in that "affordable for hobby" category. Recently I've noticed some new options that utilize a harmonic drive.

For example:

https://www.ebay.com/itm/19289023686...cAAOSwGJ5cigkc

And

https://www.ebay.com/itm/22451853762...oAAOSw-iNg3sbw

And

https://www.ebay.com/itm/15440777564...0AAOSw67hgcA4C

In theory, using a 50:1 harmonic gear solves the issues found on previous belt drive and worm gear versions. I'm wondering how they are in practice?

I.e. Do they operate backlash free and with enough torque and precision for continuous 4 axis or even 3+1 aluminum milling?

I have the harmonic 4th. Not the 5th. It would great. Amazing holding power and I've never had an issue with losing steps. I added an endstop too. It's a great addition

Sent from my SM-N960U using Tapatalk

Hi,
the only problem with harmonic drives is the deep reduction, 50:1 is the typical minimum reduction, great for high torque but SLOW as a wet week.
Cycloidal gear reducers work well at ratios around 20:1.

When I built my mini-mill seven years ago I use low lash (<2 arcmin) planetary gearboxes of 10:1. They have been superb. They allow very modest steppers to have
GREAT torque, but only at the expense of speed. To be honest I've never been able to measure or even notice the lash. So I would say that any of the cheap Chinese
made planetary gearboxes of less than 15 arcmin lash would be fine for hobby purposes, and they are available at ratios of 3:1, 5:1, 10:1 etc. Over 10:1 then your into
a two stage reduction and increase in lash.

Craig

The feedrate is reduced and if too many steps per min are sent it will not move correctly. Off hand, I cannot remember the actual said mine moves but, I can still cut at around 200ipm. Depending on stock used, obviously.
I'm not using turning tools, that's impossible. But, as an accurate and strong indexer, it's great. Continuous machining is possible using fusion cam now too.

Here is some trim I made for a kitchen window. I first used aspire and realized it's no good unless doing very simple rotary stuff. It so has to be wrapped if you do, so no pockets come out good. You could make multiple files to accomplish it.
In fusion you just create a new sub setup and change the axis to reflect what side you're using. Then you can travel in xyz to do pockets and clearing that don't end up using the side of your tool and leave a bunch of ridges everywhere.

One picture is showing that the left side one was done with aspire. The other with fusion. The aspire one looks like garbage. https://uploads.tapatalk-cdn.com/202...bef3f676f9.jpghttps://uploads.tapatalk-cdn.com/202...eafbf793be.jpghttps://uploads.tapatalk-cdn.com/202...89883d24c9.jpg

Sent from my SM-N960U using Tapatalk

Quote:

---

Originally Posted by ubergeekseven

I have the harmonic 4th. Not the 5th. It would great. Amazing holding power and I've never had an issue with losing steps. I added an endstop too. It's a great addition

Sent from my SM-N960U using Tapatalk

---

What size motor are you using with the harmonic drive? And have you used it for any aluminum parts?

I'm interested in where the limitations are re part size when working with aluminum. Could you, for example, use your 4th axis to make a 2" x 6" x 20" part?

Quote:

---

Originally Posted by joeavaerage

Hi,
the only problem with harmonic drives is the deep reduction, 50:1 is the typical minimum reduction, great for high torque but SLOW as a wet week.
Cycloidal gear reducers work well at ratios around 20:1.

When I built my mini-mill seven years ago I use low lash (<2 arcmin) planetary gearboxes of 10:1. They have been superb. They allow very modest steppers to have
GREAT torque, but only at the expense of speed. To be honest I've never been able to measure or even notice the lash. So I would say that any of the cheap Chinese
made planetary gearboxes of less than 15 arcmin lash would be fine for hobby purposes, and they are available at ratios of 3:1, 5:1, 10:1 etc. Over 10:1 then your into
a two stage reduction and increase in lash.

Craig

---

I have a large precision cycloidal gear from a Fanuc robot arm. It has a 10" turning face weighs a ton. These can also have higher ratios like harmonic drives -especially for the larger ones. Mine is 121:1.

My one is rated at something like 1 arcmin backlash with is negligible for a 100+:1 gear. I think I was just too optimistic about using it with a 3.5a stepper. The input gear shaft is 1.5" diameter so they don't even make the right coupling for a 1/2" stepper shaft.

On the speed issues, are you confident your cam set-up isn't the issue? I watched a tutorial on YouTube explaining how speed issues with a 4th axis can be related to how the cnc software interprets degrees as minutes (or something).

I haven't tried it yet but the tutorial apparently offered a fix.

Quote:

---

Originally Posted by Goemon

What size motor are you using with the harmonic drive? And have you used it for any aluminum parts?

I'm interested in where the limitations are re part size when working with aluminum. Could you, for example, use your 4th axis to make a 2" x 6" x 20" part?

---

I'm using a nema23 stepper. In the way I set it up, the maximum diameter I can use is 8 inches. The length is the width of my table at 48 inches. It could be mounted in the y direction for longer. Steady rests needed at that length for sure I made an entirely removable setup for my table top though. I've thought about building the same setup into a purpose built cnc that can travel the bounds of the build area in xyz. Maybe someday.
If I did build it into my table itself, between extrusions, I could get a 16 diameter or somewhat bigger.

I have not yet done aluminum on it. Only wood and plastic. I'm sure I could do it. I'm not sure the speeds and feeds I could use without moving something with the force.

Sent from my SM-G996U using Tapatalk

I should also say that the trim vertical pieces I made are around 30 inches in length

Sent from my SM-G996U using Tapatalk

Quote:

---

Originally Posted by ubergeekseven

I should also say that the trim vertical pieces I made are around 30 inches in length

Sent from my SM-G996U using Tapatalk

---

Here is the way I have mine setup. Aluminum plate and extrusions.https://uploads.tapatalk-cdn.com/202...8364942dec.jpg

Sent from my SM-G996U using Tapatalk

Quote:

---

Originally Posted by ubergeekseven

Here is the way I have mine setup. Aluminum plate and extrusions.https://uploads.tapatalk-cdn.com/202...8364942dec.jpg

Sent from my SM-G996U using Tapatalk

---

Thanks.

I'm curious if there is any issues with attaching the chuck directly to the gear box in terms of wear and tear on the gears.

I was thinking that the harmonic drive would connect to a rotary table with rigid rotary bearings made for cutting metal.

I guess it's probably fine for wood and an occasional aluminum job in a hobby setting.

Good question. So far, for my purposes, no problems. I suppose the forces in other setups could wear or break the gears. This is 1:100 so the forces would need to be pretty high, I think.

Sent from my SM-G996U using Tapatalk

Quote:

---

Originally Posted by ubergeekseven

Good question. So far, for my purposes, no problems. I suppose the forces in other setups could wear or break the gears. This is 1:100 so the forces would need to be pretty high, I think.

Sent from my SM-G996U using Tapatalk

---

I don't think the gear ratio would be the biggest factor. I'd guess that it would be the gear size and strength + the level of stress placed on them.

I.e. If the gears are thin, made of aluminum and you're taking deep cuts in hardened steel I'd expect more issues than if the gears were chunky, made of hardened steel and you were taking light cuts in aluminum.

It could be that the higher ratio may cause more wear as it increases the effective torque. Without a rotary table with bearings to take the axial load, that extra force is all on the gears.

Either way you'll find out eventually. If the gears start to wear out I'd assume you'd start to see some backlash where there was none before.

It probably wouldn't be that hard to add rotary table bearings if it ever becomes an issue. Ones like this are still cheap (ish) on ebay:

https://www.ebay.com/itm/19387656077...id=19044064172

I haven't gone as far as taking it apart so, not sure of the makeup of the gears. I'm also not sure if there is a bearing or of it's directly going to the gears. I took a picture of the attachment area for the chuck. Maybe someone here can let me know from that.https://uploads.tapatalk-cdn.com/202...27fad39813.jpg

Sent from my SM-G996U using Tapatalk

Hi,
if its a harmonic drive then the torsional rigidity is largely determined by the 'flex spline'.

The flex spline is a cup shaped object with gear-teeth on the external periphery of the cup. The wave generator rotates inside the cup but is larger than the natural
diameter of the flex spline and therefore forces the spline to flex into an elipitcal shape. This is the critical part of the harmonic drive. Given that the flex spline HAS TO flex
in order for the harmonic motion to happen it is also where you could get 'lost motion' or lack of torsional rigidity.

Having said that when I was investigating getting a Vexta stepper with an integral harmonic drive the rated torque of the combination was 22Nm, and that was from a 23 sized stepper,
very impressive indeed. The 'lost motion', ie torsional flex was 48 secs arc @ 22Nm. So there is torsional flex, but at under 1 min of arc still very VERY good.

Craig

Backlash wouldn't be the issue with a harmonic drive initially. The question is about the effect of wear over time.

Specifically if any wear is accelerated by connecting the gear directly to a chuck instead of using it to turn an intermediary rotary table with bearings to absorb the axial load. Worn out gears will cause backlash even on a harmonic drive.

I have seen some of the larger harmonic drives made for industrial robots that have a built-in rotary table with large bearings. From the front they look similar to my cycloidal drive:

https://i.postimg.cc/1VhPtZsh/IMG-4257.jpg

The bearings on this beast are the size of marbles. It's a different animal to the smaller harmonic drives that output via a small shaft.

I suspect that the cheap Chinese harmonic drives have more in common with the smaller shaft output drives. I'd also guess that they use aluminum gears. I can't be sure without looking though.

Either way, it would be easily solved by adding your own rotary table between the gear and chuck before any issues occurred.

Given that I'd want to use it for long parts, I'd want to add a rotary table on the other side too. I can't see there being much rigidity with one of those small tail-stocks.

Two rotary tables + two chucks would probably put the least amount of strain on the gears.

Hi,

Quote:

---

Worn out gears will cause backlash even on a harmonic drive.

---

No they won't, the only gear contact within a harmonic drive is between the ring gear and the flex spline. The ring gear has (typically) 200 teeth whereas the flex spline
has 198 teeth. That's correct a differing number of teeth and it fundamental to the way harmonic drives work. With the teeth numbers I've given this would result in a 100:1 reduction,
a common norm for harmonic drives. When the gears mesh, they are due to the differing pitch 'fighting' with each other thereby eliminating ANY backlash. They will not develop
any backlash until tens of thousands of hours have elapsed.

The output shaft is connected or integral to the flex spline. The bearings used are commonly required to be the main bearing of robotic arm. I see no need that you have another set of bearings.

Craig

Hi,
flex splines are made out of carbon steels, sort of have to be.

Craig

So at $300 or so, you could replace it several times before hitting the cost of a more expensive one that may last longer.
I do like the idea of using two instead of a tailstock.

Sent from my SM-G996U using Tapatalk