[ardenum2]

sorry guys and gals thats Efant Bearings not Ephant... Peter

I think all harmonic reducers come with cross roller slewing rings

[peteeng]

Hi Ard - Thanks this is a new area for me... Peter

[ardenum2]

the cross roller bearing is built into the body of the reducer so it's hard to discern it on some drawings. refer to the picture and red arrows.

I've been looking at harmonic drive drawings for some time until I found out that the torsional stiffness is quite low because of the flex spline. Having said that it's not that bad if you index and then engage an external break, simultaneous 5 axis is where the problem might start if you were to turnmill a bigger part, small parts should be fine.

The sizes I was looking at were at around 15Nm torsional stiffness, that might seem low but the 15N is at a distance of 1m from the center, if you put a part with a smaller OD that stiffness becomes greater, at 100mm distance from center you get 150N stiffness, at 10mm that's 1500N. I think that's how it's computed but I might be wrong, I'm not the brightest when it comes to maths and I wasn't able to confirm this from any source.

[peteeng]

Hi Ard - Where are you getting these numbers from? Does the supplier have a catalogue or manual? I'm about to ask Efant similar questions. Peter


Harmonic Drive Strain Wave Gear – Precision Gear Solutions | Harmonic Drive

Harmonic Drive Gear Solutions | Harmonic Drive Technology | Cone Drive

https://youtu.be/IXmCze1GsGU

thinking about bearing stiffness. The flexspline should not be in the structural load path. The bearing should be doing all the support. Some of these are preloaded which would be stiffer or a supplementary bearing may have to be used to improve the stability/stiffness. More research needed. Torsional stiffness is usually in the shaft direction, so the 15Nm needs a deflection in degrees at that load? or is this a static strength?

edit again - the torsional stiffness would be speced like 15Nm/deg ie at 15Nm applied torque the assembly would twist 1 deg. This would be the take up or deflection in the flexspline for instance...

[ardenum2]

harmonic drive yes that website too but from the gear unit menu not the component. eg:

https://www.harmonicdrive.net/produc.../csg-25-50-2uh

so the 15Nm needs a deflection in degrees at that load? or is this a static strength?

I'm not sure, there's a bunch of different T and K values I'm not sure what they mean. the chinese guys I asked about torsional stiffness always gave me values similar to T1 though.

The flexspline should not be in the structural load path.

if you rotate while machining at the same time or use the break on the servo, the flexspline is what takes that load, what exactly happens at 15Nm I'm not sure, it either brakes or you overpower it and rotate it against its will, if your servobrake is engaged it probably breaks

[joeavaerage]

Hi,
if you look on the Japanese HD site you'll find that they do not specify lash with a harmonic drive but 'lost motion'. The flex spline undergoes torsional flexure
under high torque loads. Having said that the lost motion is seldom more than 10 arc seconds, so very VERY good indeed, better than cycloidal or precision worm drives.

The issue with harmonic drives is their reduction. Most of the Japanese HDs are 50:1, 100:1 or 200:1. I've seen one brand on Ebay that advertised 38:1 but that's as low as they go.
Lets assume you have a 50:1 reduction, with a 3000rpm servo the output speed is only 60rpm. If you have a genuine simultaneous toolpath it's going to take forever to run. I promise you,
you will get heartily sick of the fact that your fourth axis is so slow.

My fourth axis uses a precision worm drive of 19.5:1 and with a 3000rpm servo it goes 150rpm, which is better......but I sure as hell don't want it to go any slower!!!

Most cycloidal drives I've seen advertised have reductions in the range of 15:1 to 30:1. I'm sure you could find specialized units outside that range, but typically within it. With precision manufacture they
can have las as small as 30 arc seconds, and have no lost motion as does a harmonic drive. Even with a 3000rpm servo they are still slow but have none-the-less found extensive use in CNC rotary axes.
Shame they are so pricy!

Precision worm drives can have reductions in the range of 4:1 to 50:1 with a choice of quality and lash. My fourth axis (19.5:1)_has <2 arc min lash whereas my fifth axis (6.75:1) has <1 arc min lash.
I am looking forward to the day my fifth axis is finished because then I'll be able to run simultaneous rotary paths at 444rpm with a 3000rpm servo, and that will reduce cycle times which are still
a bugbear even with my current fourth axis. I have gotten both off Ebay, both brand new (old stock), for $148USD and $180USD respectively. They have a new list price of over $1500USD. Having said that
Atlanta Drives are made in Germany, if anyone can get a deal then it would be Ardenum. They are truly superbly made, very much in the tradition of German quality engineering.

Craig

[joeavaerage]

Hi,

if you rotate while machining at the same time

That is what I call a simultaneous rotary toolpath....and should be what you are aiming for. Indexing is way less demanding.

An indexing toolpath is where you rotate a precise amount and then machine that face. It is common for indexing tables to have a hydraulic or pneumatic brake so that the worm/flexspline/cyclodial gear
and its driving servo are not subjected to the stress of remaining stationary against the cutting forces.

A simultaneous toolpath requires that the worm/flexspline/cycloidal gear can not only withstand the torque of cutting forces but generate the torque to engage the tool. Night and day difference.

Craig

[peteeng]

Hi Ard - I downloaded the "Harmonic Drive " manual. See attached. The T figures are strengths, the K figures are stiffness's. T1 is the acceleration torque limit, T2 is the constant torque limit, T3 is the deceleration limit and T4 is the dwell limit. These are values that will produce infinite life.... K1 is the accel stiffness, K2 is the constant speed stiffness and K3 is the decel stiffness. Details are in the manual.... stiffness is defined in Nm/rad

There are equations to adjust from infinite life if overloading and shorter life is acceptable. . They seem to be concerned with crash loads as well.... Peter

[ardenum2]

Atlanta Drives are made in Germany, if anyone can get a deal then it would be Ardenum. They are truly superbly made, very much in the tradition of German quality engineering.

I'd have to make a company to buy them, companies in germany won't even respond to emails if you're also not a business. I was considering cycloidals for some time but ultimately ruled against them, haas' still use them for some of their rotaries, but I'd be more interested in what dmg use, belts driven by servos driving a slewing ring with helical teeth that's backlash compensated by another pinion. think that's mainly how that works. would be better to find the patent for it, just to be sure.

main purpose of building my own stuff is to be independent from manufacturers. times change, prices change. I don't like feeling that I'm at the mercy of someone's greed so I'm trying to reduce my dependence as much as possible. Can't do anything about electronics but making a direct drive is easier than making cycloidal discs within spec, that's mechanical tolerance, harder to get right and the worse the tolerance the worse the lifetime. ironcore linear motors are just a plate with magnets and another plate assembly with wires, making diy ballscrews on the other hand is impossible. direct drive is as precise as the encoder you use and the drive that controls it.

[joeavaerage]

Hi,
the overwhelming majority of indexing tables are worm gears driven by a servo through a right angle gearbox.

https://www.ebay.com/itm/31419065187...Bk9SR9zT25rHYg

This is a good example as any. Note the circular cover on the left hand end, no doubt to retain the bearings of the worm gear. The servo and right angle drive are in the right hand end.
If you look closely most, not all but most, indexing tables are of this construction.

https://www.ebay.com/itm/28544469688...Bk9SR-DT25rHYg

Another worm drive, if I only had $18.5k USD.....I could have it!

Craig

[joeavaerage]

Hi peteeng,
those harmonic drives are characterised extremely well.

They are very often used in robot arm and thus axial stiffness under any conceivable loading condition would have to be known for the device to be of use to OEMs.
Have you priced some of these units?. I'd always found that Japanese quality harmonic drives were eye-wateringly expensive. Even second hand they are pricy.

This is just a baby one (size 14), but it is new:

https://www.ebay.com/itm/32531665971...3ABFBMrNnrp8di

Still $899USD....not cheap!

These ones are second hand, but there is still three of them left and much better buying at $213USD each and in a much bigger size, size 32:

https://www.ebay.com/itm/28524878954...Bk9SR6jZ66fHYg

Craig

[peteeng]

Hi Craig - The manual is from an American company that invented the harmonic drive. I'm about to start talking to Efant about their bearings. In another thread you mention 1:50 gears could be very slow. My calcs show that as well so maybe a 1:20 or 1:10 precision gearbox is the go for this machine. I haven't got that far yet. This machine is sort of a learner 5 axis to figure out what's what in that space. You have mentioned twice that Milli is tardy in getting built. The investigation has wandered and may never get built. At least the thread will be a legacy on materials available and how to use them and not to use them amongst other things. Peter

[joeavaerage]

Hi peteeng,

My calcs show that as well so maybe a 1:20 or 1:10 precision gearbox is the go for this machine.

Yes, I have found that to be about right for my machine also. I had a stepper (itself had a low lash 10:1 planetary reduction) on my fourth axis, which with the steppers slow (240rpm)
input resulted in 12.3rpm output at the fourth axis....and that was just to slow, in was in effect unusable for continuous rotation rotary toolpaths. For indexing fourth axis it would be fine,
but it had always been my intention to have a simultaneous four axis toolpath capability. I fitted a 3000rpm servo, and that perked it up considerably, up to 150rpm.

Johno, my work mate has been making parts with it this afternoon. Very satisfying.

What you really need to decide is whether you want continuous fourth axis, as I did, in which case I'd aim for rotation speeds of 150rpm or higher. If however you are solely interested
in a trunnion fourth and free running fifth axis....then a high reduction/slow fourth axis would be perfectly acceptable. It would be the fifth axis which now comes into consideration as to its potential
output speed, if you want continuous fifth axis rotation. Naturally if you only want indexing fifth axis then high reduction/slow is viable there too.

High reductions give great torque multiplication and are highly desirable as it allows a smaller cheaper motor....but that has to be balanced with speed. I can assure you that continuous rotation toolpaths
on a slow rotary is a prime cause of drunkeness in my neck of the woods! Devil and idle hands and all of that.

Craig

[ardenum2]

Hi,
the overwhelming majority of indexing tables are worm gears driven by a servo through a right angle gearbox.

https://www.ebay.com/itm/31419065187...Bk9SR9zT25rHYg

This is a good example as any. Note the circular cover on the left hand end, no doubt to retain the bearings of the worm gear. The servo and right angle drive are in the right hand end.
If you look closely most, not all but most, indexing tables are of this construction.

https://www.ebay.com/itm/28544469688...Bk9SR-DT25rHYg

Another worm drive, if I only had $18.5k USD.....I could have it!

Craig

those are not for us, they're for job shops. ones with dozens of machines and that is why the manufacturer prices them like that, they are not worth that much on their own, their price is calculated based on how much $$ a job shop can make using them.

[Buoyen]

The new Haas 350 HD has a max speed og 100 deg/sec for B and 80 deg/s for C.
DMG, Hermle or Kern might be in a different league. But I guess higher speeds are mostly (in the real world) only for faster positioning and rapids.

https://www.haascnc.com/machines/ver...umc-350hd.html

[joeavaerage]

Hi,
well you may not think they are suitable for you but they are to me. In truth both are probably a little large but otherwise they are what I want.
Indeed the fourth axis I have built is a somewhat smaller version of exactly the same thing. My fourth axis has over 40Nm torque, and that is a fraction of
the gearboxes rated torque of 130Nm. My fourth axis does not have a brake, but then I anticipate the majority use will be for continuous fourth axis
rotation, not indexing.

and that is why the manufacturer prices them like that, they are not worth that much on their own,

News flash, heavy duty fourth and fifth axes are uber expensive, tens of thousands of dollars. If you want a fourth and/or fifth axis of the rigidity to do steel parts, even
small steel parts you had best get used to the idea it's going to cost. How do I know?....I've been looking for three years. I believe I've come up with the most rigid combination
I could within my budget....at this time about $5000NZD ($3000USD). A good (Taiwanese made....not Japanese...they are more again) 4 inch self centering vice is $600USD on its own.
I'm not quite sure how you go about pricing things....I tend to look at manufacturers websites, enquire via email, look on Ebay for second hand or new old stock, but I always seem to come
up with an answer that you think too high. If you can buy as cheap as you claim then do so.....and tell us all where and how.

Craig

[joeavaerage]

Hi,

But I guess higher speeds are mostly (in the real world) only for faster positioning and rapids.

No, speed becomes important when you have a continuous fourth axis tool paths. Its nothing for a part to require 2000 rotations to run, and at even 150rpm that's going to
take 131/3 minutes. If your rotary axis can only do 100 degree/second or 16rpm such a toolpath is going to take 2hours. I promise you'll get really sick of toolpaths that take
that long to run just because your rotary is so slow.

Its common to see glossy advertising of four and five axis machines proudly proclaiming rotary speeds of 300 rpm, 500 rpm, and in the million dollar machine class, 1000rpm.
Why do you suppose manufacturers go to the expense to design, build and market such rotaries?

Craig

[Buoyen]

Yes of course. For some the increase in productivity justifies an increase of 5x in the price of the machine.
And of course many are doing really high end high speed manufacturing in 5 axes and do utilize that speed.

My point is just this: you can do everything the big boys can it just takes a bit more time - and most toolpaths are slower than this and not limited by the B and C speed - I think this is clear when Haas can sell these machines and have happy customers.

[joeavaerage]

Hi,

and most toolpaths are slower than this and not limited by the B and C speed

That is contrary to my experience. Having tried continuous toolpaths with small parts I realize that I want decent rotary speed. My upcoming fifth axis will be 444rpm,
which is what I consider the sweet spot. Fast enough to make a reasonably quick job of multiple rotation toolpaths and yet enough reduction to have some useful
torque multiplication. The C axis will have 16.2Nm at rated servo torque and 48.6Nm at overload. It certainly won't be swinging engine blocks or anything like that,
but I do expect steel parts of 5kg-10kg and maybe 150mm in diameter to be within the capacity of the C axis.

I think this is clear when Haas can sell these machines and have happy customers.

I'm sure they do, and I would guess that those customers are utilizing 3+2 toolpaths, for which modest rotary speeds are entirely adequate....its continuous rotation toolpaths that show
what a bottleneck a slow rotary can be.

Craig

[Buoyen]

That is great.
I was just going to mention it that it seems many think slower speeds are also okay. And that opens up som opportunitets when it comes to building the rotary.

This is a Hermle C22
https://www.youtube.com/watch?v=WiD56u0h_XA

Spec sheet: 25-55 rpm on A and 40 to 80 rpm on C depending on configuration according to the spec sheet.