[jsheerin]

I'm thinking about making a C/A head for my next router (rotates the router about the Z and X axes to give 5 axis movement). It needs a large gear reduction between the servo and the head - something around 64:1. Going with timing belts and pulleys commonly available from SDP/si, it looks like that would require 3 sets of pulleys. I've been reading a bit, and all I can find is that the GT2 belts are the lowest backlash commonly available. However actual backlash is never really specified... Does anyone know or have a feel for how this would work out? Would backlash add up to unacceptable levels with this many reduction stages?

In comparison, I've been looking at the Thompson Micron Ultra True gear drives and they typically spec backlash in the 10 arc second range, I believe (although that's just from memory). When I was doing the calculations, I think that translated to a 0.015" movement (or lack of movement) of the tool tip on a long tool which didn't sound particularly great to me.

[JLSD]

Hi,
Just a thought, if you're basically tilting the cutter axis, What total angle do you need? Up to around 90 deg. total movement, a ballscrew/lever mechanism could be one answer. I know the output's not linear, but that could be fiddled in software, or by simple calcs if you're moving to preset positions. If you need continuous movement, it really needs a math function input to the controller, but that only needs someone smarter than me! (not hard!) It ought to be way stiffer than any rubber trick, especially 3-stage!
Regards,
John.

[jsheerin]

Thanks for the idea. It is interesting... It seems like it would be a tough way to go for continuous motion using non-custom software which is an eventual goal.

I'm looking to match (or better) the basic capabilities of an HSD head as seen on this machine:
http://www.cncmotion.com/titancanada.htm
Head specs are here:
http://www.hsd-hitec.com/eng/cont/2_assi.asp
Basically the C turns +/- 210 degrees and the A turns +/- 120 degrees.

To start with I'll only be positioning the head and then cutting, but eventually, hopefully after the machine has made some money, I'll be looking to upgrade my cam software to be able to do continuous 5 axis motion.

[stangtjk]

With that high of a gear ratio it might be better to go with an anti backlash worm gear assembly rather than a series of belts and pulleys.

[jsheerin]

Yes, harmonic drives are used in the HSD heads. Is that what you mean by anti-backlash worm gears, or is there another technique there to reduce backlash?

Overall, the goal is to reduce cost. So far almost the entire project is an ebay or industrial auction special. That is why I am thinking about the belt drives. If it turns out to not be practical, then I will have to hunt for some surplus gear drives.

[jsheerin]

Or do you mean two worm gears on opposite sides of an output gear with one driven to remove backlash? Like in US patent 7552661?

Would it be possible to use a separate servo motor with a drive running in torque mode to constantly apply a specified torque to remove the backlash on one of the worms? This seems like it would be possible, but I don't know enough about the electronics side of things to know if this would be practical... So this motor would always apply a constant torque, but the sign of the torque would change depending on which direction the primary motor was moving. Anyone know if this is possible? I'm running out of movement axes if I stick with Mach for a controller, so I'd probably have to be able to do this by using the direction pin for the rotation axis - when the direction pin changes, the direction of the torque would change.

[stangtjk]

I was referring to something like this. I'm not sure how much backlash is actually present with something like this, its basically just 2 gears face to face with a spring loaded between them to remove backlash. It was just a thought don't know much about it.

[jsheerin]

Okay, that's a better (or at least simpler) idea than mine - passive instead of active. So that's basically the same as a rack and pinion split pinion, but the worm has the advantage of preventing back driving, right?

So the question then would become if I could find a reasonably priced commercially available model or if I'd have to make my own up-sized version (which might not be too hard). I sent an email to SDP asking if they had any torque ratings for the parts they sell. The anti-backlash gear is only ~$50, so that would be pretty reasonable. If it won't support the torque I want, then I could do my own with cast iron and steel parts from Mcmaster-Carr for around $270 plus springs and machining. It seems like the biggest problem here might be getting springs in there that would be stiff enough to resist any torque trying to move the head within its backlash. I suppose that would be the advantage of the active version...

[jsheerin]

I suppose another way to get around having to have springs inside the gear (and a split gear) would be to have a second worm gear (as in the active version) but couple it to the driven worm and then spring load it against the gear to take out the backlash.

Another concern I had was efficiency of this type of setup. Looking here:
http://www.roymech.co.uk/Useful_Tabl...fficiency.html
and here:
http://www.engineershandbook.com/Tab...efficients.htm
it appears I could get 70% efficiency with greased steel and cast iron before applying any loads to remove backlash.

[Al_The_Man]

I have traditionally picked planetary gear box on ebay, Bayside is one nice make, lately the prices have been getting a bit steep, I found it sometimes cheaper to buy when a motor was attached for some reason.
Planetary you can get up to 100:1.
Al.

[wildwestpat]

Hi

The main consideration I suspect will be ridgidity and the forces inposed by the cutting action with an alowance for the odd dig in! The use of spring loaded half worm wheels will give you an elasticity in the axis drive. In designing radar aerial mounts I have used tightly meshed worm / wheel centers with the tooth form such that the worm can not bottom on the gear tooth root diameter. This is a common arrangement and is very effective as the adjustment mechanism is easily arranged. The worm and the gear should be made of compatible materials - stanless steel and brass for example with a good lubricant as the tooth pressures are higher than for a normal mesh. The rotation speeds for this sort of mechnism would be low so the lubrication problems should be minimal.

Hope this helps regards Pat

PS Tigter meshing also works with spur gearing but you would not get the back drive protection again there is an increase in friction losses which will limit the maximum speed.

[jsheerin]

I finally got back in front of my computer (well, the one with all my design info on it...). The HSD head has backlash at the end of a 15.7" long tool (or at least the tool tip is that far from the center of rotation) of 0.0016" with a torque of 118N*m applied if I'm reading their spec correctly. If I use a 100:1 planetary gear head reduction, those typically have around 0.5 arc minutes of backlash (Thompson Ultra-true) which would give about 0.0023" of backlash at the same distance. So that could be a possibility.

Looking at the stiffness of the head, HSD specs a stiffness of about 4e-5 in/N. This is actually for a load applied perpendicular to the tool bit when the bit is pointed straight down, but I wanted to compare that to the rotational stiffness of what I'm designing. So I looked at the Ultra true planetary drives again. I'd need a size 18 head for a 100:1 ratio. This would give me a rotational stiffness of 1.1e-6. For the continuous cutting force I'll have available (with the head at full acceleration), that would give me a deflection of 0.0014" at the tool tip. At the peak cutting force, I'd get 0.0055" of deflection. Those seem pretty reasonable. Then again, a size 18 gear box weighs 45 pounds, so it better be pretty stiff... Somehow I doubt I'll find one of those cheaply.

HSD also specs positioning accuracy at 0.004" at the tool tip. This is still something I would need to figure out. However, after reading here:
http://nsa.kpu-m.ac.jp/gijutu/kousak...qnotes.htm#wga
it looks like if I went with the worm gear approach, going with a larger gear would reduce positioning errors due to errors in the worm gear hobbing. Also, this probably points out that it could be a good idea to follow HSD's lead and have an encoder directly on the head itself instead of depending on the encoder on the servo motor.

So from the above, it looks like a planetary head could work fine, but the penalties would probably be cost and weight. A worm gear drive is looking better.

Btw, I know I'm mixing units horribly, but I find it easier to work in metric units for everything but positioning and speed. I still find it easiest to judge those results in English units.

[jsheerin]

Oops - misplaced a decimal point. Backlash of 5.0 arc*minutes (instead of 0.5) in a planetary gear head would give about 0.023" of cutter backlash. Not acceptable for what I'm shooting for.

[jsheerin]

What if I put two worm pinions next to each other on a shaft with a belleville washer in between? This would load the pinions apart to take out backlash, like a split pinion on a rack and pinion. The pinions would be keyed to the shaft. The disadvantage to this would be that less of each pinion would be engaged with the worm gear. I calculated the spring force required if I wanted to remove backlash up to my continuous input torque, and it would be around 1000 lbf. Belleville washers are available at that load rating. In practice I might want to go with a lower load.

Another way could be to stack two worm gears and pin them together. Then use two worm pinions, one engaged with each gear, that were spring loaded in opposite directions. The shafts the pinions are keyed to would be linked with a timing belt. This would allow full engagement of each pinion with the gear but increase cost (extra worm gear which is several times more than the pinion).

Thoughts?

[wildwestpat]

If you use a very large diameter worm wheel with a fine pitch you might get your idea of a split spring loaded worm to work. Guess you could use two worms linked with a belt drive where the belt acts as the spring but I still think your best bet is to use a worm wheel set with an adjustable mesh. This will require that the tooth is sufficiently deep that the worm tooth crest does not bottom in the wheel and the wheel crest does not bottom in the worm. Most commercial sets would be OK when new.

Regards

Pat

[jsheerin]

Hey Pat,
I like that idea from a simplicity stand point (adjustable mesh), but I'm afraid a lot of my wear would be concentrated in one spot on the gear and thus adjustment over time would be difficult. That's why I'm interested in something that has some ability to self adjust. I'm probably over-thinking it, but it's easier to do now than later. If I can't come up with something that self adjusts that I like, I will go with the adjustable mesh.

Maybe if I set up the head so it could be easily rotated some amount (60 degrees or something like that) and reattached to the shaft, that could be something I did every so many hours to distribute the wear more evenly around the gear. That would eliminate my main concern.

[wildwestpat]

Good thinking about long term maintenance. Yes fitting the worm wheel to the axis bearing / shaft with a precision bore and say six bolts to a suitable flange that is integral to the shaft. That way you could change the wear segment when the backlash becomes objectionable as you say. To get the ridgidity and accuracy as high as practical points to a large worm wheel and this lowers the tooth pressure hence wear. Also using a suitable lubricant with anti scuffing or molyslip additives would also help cope with the constant reversing. It is not too difficult to cut your own worm and wheel as you don't need to have it interchangeable for commercially available parts try searching for astroscope equitorial mount builds http://www.mini-lathe.org.uk/worm.shtml is but one to give you some more ideas.

Good luck with your design - Pat

[jsheerin]

Yeah, that's what I was thinking with the bolts and a flange. I was looking at a 10" diameter worm wheel. I'd guess that I could get a fairly long life out of this by rotating it and once it got too bad, a replacement would be in the $150 range. It's more that I'd like to design something that I don't have to service a lot. I think this will be a reasonable way to go though - low complexity and cost traded for a bit of occasional service.

Thanks,
John