Worm Gear to Reduce the Reverse Friction

I'd like to see some 'Electronic Cat Brain' come up with that answer!

But the design is used to drive the load as close as possible to the hard limit. The motor will be powered off after it hit the limit, however turning it back is the problem.

Thanks!

But the design is used to drive the load as close as possible to the hard limit..
Yeah, that's my point ...the design is wrong!
'As close as possible so that it still works' would be a better target.

The microswitch doesn't even have to stop the motor it could just limit the current to give you a soft stop.

If you insist on not solving the problem, you must try and obviate the symptoms, e.g find out exactly what is causing the jamming and do something to relieve it.

I always say that solving problems is easy...the hard thing is discovering what the problem is. Maybe a different gear form on the worm or drive gears would solve the problem, maybe rubber mounting the motor would help. How does it stop at the moment ? Does it just smash into the end stop and stall? or is there some sort of limit switch? if you have a limit switch just move the darned thing 1mm.. at a time until the problem goes away.
Del

The hit-direction current is set as low as it rotate the worm gear (plus some tolerance). I'm sure it is safe and that no component will be damaged (gear, bearing, ...). I checked the gear and related components after a lot of trials. Everything are Okay, I can not even find any hit-mark on the surface of gear, or hard stop.

The jam happens between the worm wheel and worm lead. After the gear is pushed on the hard stop, the friction on the worm gear mesh stops it turning back. I somehow compared this friction with free-rotating friction, it is 5-10 times bigger. I can increase the power 10 times for reverse direction, however it should be better if this friction can be somehow reduced. It shouldn't be that high(10 times).

I'm not an expert of worm gear, I'm wondering if there are ways to reduce this friction. for example: make the gear with some backlash, change the mesh angle, ...

Thanks

Try not to find a problem for every solution.

This sounds so much like my day, today.

Visiting Scientist: I want to be able to move my instrument with this 100 mm slide.

Redfred: Ok

VS: But I need to move it 2more millimeters to the right.

RF: So move the whole stage 5 millimeters to the right. That'll give you 4mm of adjustment.

VS: I tried that, I can't get the stage to move back to zero then.

RF: Well that depends on where you assign zero to be. Where is zero?

VS: All the way to the left.

RF: So you want the stage the 100mm stage to move 102mm. I can't help you. Why didn't you buy a 120mm or 110mm stage?

VS: I calculated that I'd need 99mm of travel. So I bought a 100mm stage.

RF: 99mm± what?

VS: What do you mean?

RF: Your calculation of the travel length included a measurement uncertainty didn't it?

VS: Oh, I haven't done uncertainty calculations since my undergraduate days. Do you think that if I do that, then maybe I go for a shorter stage.

RF:

RF: I can't help you. (Tries to leave.)

VS: Wait, I'm having trouble also with the vertical stage. My other experiments I can turn off the stepping motor drivers. When I do this all of the motors stay where they were. This vertical stage drops to the bottom of travel when the power gets turned OFF.

RF: Sounds fine to me, how much mass are you moving with this vertical stage?

VS: 15kg.

RF: I can't help you.

Slamming a worm drive is sort of like tightening a screw really hard with an impact wrench. In this case, you may also be squeezing out the oil film between the worm and the gear. Reversing the motor is now like trying to loosen a stuck screw. A gear oil with a higher pressure rating might help. (In the short run, that is; over time it sounds like accelerated wear.)

Yes, it likes to fasten and release a screw. However the fastening torque is controlled and the un-fastening torque is set a couple times bigger than the fastening torque.

However, in the design, the un-fastening torque is 5-10 times bigger than the controlled fastening torque. Are there any way to reduce this ratio ?

Thank you.

One way might be a higher-pressure lubricant, as previously mentioned. Another way might be a two-stage worm gear (1:5 x 1:6 = 1:30) The greater pitch angles would make it easier to be rotated in reverse.

Hi Tornado,

Thanks for the information, I think I can understand why the greater pitch angle helps. However, would you mind describe more about why does 2-stage gear help? Thank you again.

Do loop alert!

1. I have a problem.
2. Here's the solution(s)
3. No everything I have done is fine
4. goto 1.

Someone PM me if you ever escape?
Del

That won't happen. No one will admit to that much egg on the face.

I considered the possibility that the OP was driving a trip hammer (or maybe even some sort of Brinell or Rockwell hardness tester). But no cogent explanations were given as to the rationale for this ram-it-into-the-wall scenario. So far your car-in-garage idea seems to be holding up the best.

Reminds me of an old Groucho Marx bit:

Guy comes into the doctor's office saying "Doc! Doc! It hurts when I do this! (while flexing elbow)"

Groucho says "Well don't do that!"

Is it possible to reduce speed before yopu hit the hard limit so the 'hit' is not so hard.

If using DC control you can also introduce a current limit at this point but it sounds like momentum driving you in and reducing speed may make it softer and easier to back out of?

Already limit the DC motor 'hit' current; and power off the DC motor completely after 0.1 second of 'hit; the 'hit' speed is reduced also to an acceptable low value. I think the electronic side had done its job pretty much. And the problem is almost solved.

However, to make the design reliable, the reverse voltage need to be very high which is not acceptable. I'm looking forward for the tips on the mechanical side (especially the worm gear) so that I can use lower reverse voltage. Or in other words, the tips to reduce the 'release/fasten torque ratio' of worm gear is what I'm looking for.

Thank you very much.

Ah more information!
It sounds as though you have done all the right stuff, but not adjusted it correctly.
I'd suggest deliberately limiting the current, travel even more etc, until it stops just short of where you want, and then slowly adjsuting until it behaves exactly how you want.
It's a technique called bracketing...e.g. if you are aiming to stop at say 3mm (approaching from zero) adjust to stop at 4mmm (eg jammed solid which is where you are now), then adjust to stop at 2mm, this shows then adjust to get the mid point. And finally adjust to the correct point.

It seems to me that you have yet to demostrate that you can stop at the 2mm point, thus you are floundering around at the end stop.
Of course it could be that you are giving us insufficient information to be more helpful, or you are going to tease us by drip feeding us with it....
You see curiosity is in danger of keeping this cat stuck here.

I s'pose the real question is, do you need the 'position' of the drive or the 'force'?
If you only need the 'position' then please please please stop whacking it into the end stop.
Del

I'm sorry that I did not explain my design well. The design is to drive a target to stop at one fixed position for 0.1 second, within +/- 0.01mm every time, and then go away from that point until next time. That's where I introduced the hard stop to 'hit'.

Right now it works, except that some times it get stuck at that point even I increase the voltage 3 times for the reverse direction. There is nothing the electronic side can do unless I want to use a complex positioning control system, I think.

In the design, if I can somehow reduce the gear reverse friction, that will be fine.

Thank you for all your help.

I see three possible solutions

Decrease the reduction ratio of the worm gear and decrease at the same time the speed of your motor. A worm/wormwheel gear with a smaller reduction factor (let say i=10 in stead of 30) will have a higher efficiency and will also be reversible.

• i = 7 total reversible
• i = 10-20 statically reversible
• i= 15-60 variable static non reversing, quick return in case of vibrations, dynamically reversible
• i= 40-100 statically non reversing, slow mouvement return in case of vibrations, low dynamically reversing
• i= 70 - 100 stativally non reversing, no return, low dynamic reversing

Second solution, using two gears behind each other. The first angle gear with ratio i=1 and and second gear with ratio 30 (helical tooth wheels)

Third solution, using a bevel gear (like is in your grinder)

Is the 'reversible worm gear' you mean here the one that can be driven from the worm wheel? If yes, it is not the case in the design, the 'reverse direction' I described means rotating in the reverse direction after jamming.

Do you imply that the 'reversible worm gear' is easier to rotate back after jamming.

Thanks

There's your problem right there!! You have a motor spinning which has momentum and you are trying to stop it with a dead stop! You hit the dead stop and the momentum in the motor binds up in the worm and worm wheel! Could you not move the target with a piston or linear motor which would not be prone to binding up! If not, how about changing the DC motor to a controlled stepper motor! That way, you can control the exact position of the stop without having to smash into a dead stop! You can have a dead stop but you use it more as a fixed point of reference once you have set up the system to stop there!

Like your post!!

If he used proximity detectors (never unreliable switches with bounce problems) to "warn" him of nearing the ends, he could reduce speed so as not to run into the ends so hard, but still use full power to drive away from the ends again.

A further refinement would be to use current sensors to shut power off in the direction of the stop, they are available as chips with all the clever stuff built in.

If I was doing this I would use a PIC as master controller and just tell the PIC what I wanted it to do......

You can design the Worm with 2 leads instead of One Lead. This will give better power transmission...The worm wheel will be the same.

Also, if you keep the original 1 lead design, to reverse, maybe you can go forward/Reverse in one sequence to dislodge the grip and initiate the motion. This might need setting the stop just before the dead end if possible...?

Well this guest has agreed that several critical factors were not included in the original post. I applaud your honesty and clarifications that you have added to people's inquiries for further details here. Allow me to formally invite you to sign up and join the community. Now for my two cents on this topic.

While I have done many simple and complicated jobs with motion control, the mechanics of gear design is not my realm. So to your specific question on the gear design, I cannot contribute anything but a layman's perspective and certainly nothing more than what has already been contributed. But I and many other here think that how your system is being used is more likely the root of your problem, not the gear design. You've now indicated that this DC motor is intended to move an object between two locations. One position must be to a high precision and the object must be moved away from this position after only a brief stay. While the idea of moving one object into an immovable object to retain precision seems like a plausible idea, at ±10 micron precision this is not possible. At the speeds that you intend to move this object (at least at retraction) and with the unknown masses of all objects, I'm certain that some of the mechanical parts will deform greater than 10 microns over time. Without some feedback to your system you will never really know that this object is precisely at any location to within ±10 micron precision. There's a good reason that position controls at this level of precision can get complicated, it has to be.

Ah, now we know the application!
Put a spring between the target and it's carriage so that the target hits the stop and is held firmly against it by the spring, with the drive carriage compressing the spring but not actually hitting the end stop (or bottoming out the spring).
E.G. Let the taget hit the stop but not the drive, which will just compress the spring.
Of course there will be a slight delay when reversing as the spring travel is taken up but that is a timing issue and should be capable of being calibrated out.
The spring may actually help the reversing action!

(Sliding pins etc can be added to ensure the target doesn't wobble about too much)
Del

I think using a s sping to absorb the impact is a good idea, I will try that in the design. Thanks for your idea.

Thanks to all construibutors, too!

A few silly questions comes to me, how massive is the object that will rest for only 0.1 seconds at a location with 10 micron precision? How far and quickly does the object have to move to get out of this precise location?

The 0.1 second is the time when the target stopped. And the time doesn't need to be very accurate(0.1-0.2 seconds is acceptable). The deceleration and acceleration time can be 1 seconds which is not a problem; and the target is only 1kg max.

This certainly does not seem unreasonable now, just a little tricky. I still think that you'll need a position sensing switch like the link provides. Without having something that accurately tells you that the object is actually in position, you will neve actually know. I like several of Del's suggestions, specifically that you compress a spring at the end of travel to assist the slow down. I also like his initial suggestion of a near end of travel switch to slow the motion down prior to reaching the end of limit. I recommend one with a roller on the leaf. This way the stage can roll past the slowdown switch position and permit a precise stop at the precision stop switch position. Now how your motion control system handles this collection of contact signals is another story all together. For your application I believe this can all be easily done with a few relays, one of which will be a time delay relay that will handle your time at target position.

Thanks for the advice. Yes, I have these two kind of sensors in the design, one to sens the contact, and another two photo interrupters for slowing down on two sides.

For the contact, I use two pieces of grind-ed stainless steel right now. The positioning repeatability is 2um which I measured by the gauge. However, like you say, I must also consider the long term contact deformation. Fortunately, the design allow the user to do some alignment. Maybe I should use special material to minimize the contact deformation.

I expect, when I add the spring, I might sacrifices some repeatability but gain some long-term accuracy though. Fair!

I used to work on Robots, we never even used any form of limit switch, sensor or spring, BUT we did use good quality servo systems and motor current sensors. The Robot, when first powered up moved slowly and found the end stops for each movement, and remembered exactly where they were.

Thus it was bale to move at full speed to ANY position between the two end stops with full acceleration and deceleration. It could repeat the action time after time perfectly.......

If you need exact positioning, go the servo motor method, or even cheaper, the stepper motor method.....but the stepper motor runs without feedback generally.....but once you know where the end stops are and if you keep proper track, you have full repeatability as well......

Best of luck!

I think If this had been my project I would have gone a whole different path, first I would have used a DC servo with either an encoder or a resolver(the one you chose depends on if you need to sense position after a power failure or not) to sense position (probably Baldor simply because I've used them before and they are easy to use but there are lots of others out there). And I would have stopped the target under servo control well shy of the end stops.

Del has given you lots of good advice, as has redfred, Tornado, and others, I think the best advise you have been given (repeatedly) is don't try to use the hard stop as your positioning element. You have gone and painted yourself into a corner with this design and now you are trying to figure out how to knock a hole in the wall to get out. Sometimes it is best to step back and look at the design honestly and try to overcome your emotional investment in the work and realize that there might just be a better way of skinning the cat (sorry Del, you know what I mean...). Such as using the servo itself and dispensing with the worm drive.

But if you absolutely MUST do this. consider coating the worm with xylan.

I 200% agree that I'm doing something unusual. This 'version-II' design is mainly for cost down, and also for the consideration of reducing the mechanical size. Our 1st design used for many years has a fully digital servo closed loop with encoder feedback, it drive directly without using any gear to minimize the backlash. The motor is bigger, plus the length of encoder, space for wiring, precise ball screw, ... , Customer want it to be cheaper and smaller. (I don't agree that 'customer is always right'. But as an engineer if I don't try, I'm not right.)

Thanks for your coating suggestion.

I just made a similar post to yours, you were first though. Apologies.

No worries, great minds and all.....=b

Thanks.

Hi,

Is there any recommendation for the spring having good stability of force constant after a lot of times of deflection and extension?

Thank you!

After reading 36 posts, I did not see anyone suggest shorting the motor. The DC motor has inertia;it will continue to spin after power has been removed. Instead of just removing power, remove power AND short the two motor leads to each other. The motor now acts as a generator with its own windings as the only current limiter; it will stop almost instantly instead of coasting to jam the gears.

Its called "Dynamic Braking", is used in many areas like cranes and the like and sadly, few understand and believe just how well it works!!!

Well remembered/mentioned and thanks. I had quite forgotten.....

Hi,

Thanks for the suggestion, I did that. The tests shows that helps but not much. The reason might be the contact is happening before the motor is shorted, and the time to stop is too short for the motor to 'brake'.

The following just come up to my mind :

Electronic side: using a 'dynamic current limiting' control algorithm that will gives the motor just the enough current to rotate forward before contact, so the forward friction will be dynamically compensated. Because the gear friction is not really a constant every time, without dynamically compensation, the applied limited constant current should be big enough to make sure that the motor run at worst case. This safe current might cause problem when the friction condition is too 'good' just before the contact.

Mechanical side: using the belt instead of worm gear (ratio 30 ?)

Roller chain will be more efficient than a belt. but getting a 30-1 ratio in a single stage might be a trick either way.

Thanks for the information. Maybe use a gear motor and reduce the ratio as long as it meets the size requirement?

OK

No timing belt or chain-and-sprocket I'm aware of will give you the precision you are asking for.

Use the worm system as you are, and put the end-of travel sensor on a micrometer (fine-threaded screw) adjustment mount, and use trial-and-error to determine the correct placement.

Start with the sensor stopping too soon, and adjust until it just passes, then back.

I think the point was to drive the target to the hard stop through a chain or belt drive and hold it there under tension. the hard stop would provide the precision required just as the worm does now.

I have been thinking about this and I find that for the accuracy he wants, either a servo or stepper is the way to go, with a PIC to handle the fine details....via a worm drive.

That way accuracy is assured after calibration (which should be automatic each time power is restored by the way!).....

This "hard & mechanical" method is too "hit & miss" for me personally.....

I still don't think that it needs to be expensive either.....especially with a stepper system.....

The worm gear has certain backlash either. To achieve 0.01mm, is not only a servo or stepper system. It should drive direct, and a double nut (no backlash) and precision grade ball screw is required. A no backlash coupling between the motor and screw is required. The motor power is much bigger when driving direct.

There are in actual fact gearboxes with as good as zero backlash made by various companies.....

Depending upon the configuration of the gearbox, you can either have increased speed or increased torque......

If there are some long-term zero backlash gear box I can use, then this design can be much easier. I know one type of them, the harmonic gear. Really expansive though.

Could you give me some info if you know some other types, Thanks.

Search on the web, but none are really cheap. Use CNC as part of the search argument as well.....

I personally (as mentioned twice) would use steppers and a PIC as controller......no gearboxes...

Best of luck.

Given the brief intervals of travel/reciprocation, have you considered a cam mechanism with spring loaded follower to achieve your positioning?

You have not to be surprised that the torque ratios is between 5 and 10. It should be so in the way you tried to solve the problem. It is normal since when the worm works under normal conditions a lubricant layer builds up and friction will be in the range of 0.05...0.02, when you stop against a hard stop the whole kinetic energy of the motor will first squeeze the lubricant between the worm and gear flanks and then preload the whole system consisting of the worm its bearings and the gear + shaft + stop. The harder the stop the higher the peak force which will appear. The contact after stop will get in static friction conditions and reach static values of 0.1...0.14. You see that the ratios are 2 to 7 or even more. I see only a combined solution consisting of a drastic reduction of incoming speed at contact and an increase of stop compliance combined with the usage of a lubricant with a higher viscosity and additives so that its expulsion will be less easy and the traces of additives will maintain a lower "dry" coefficient of friction, such additives are MoS2 or graphite or Teflon. You cannot fight against tribology only try to avoid its negative aspects. The kinetic energy reduction can be obtained via a motor speed reduction controlled from a given distance to the stop as many comments already recommended and use a stop with reduced stiffness. If energy is small the stiffness reduction will not compromise the precision requirements. You can also avoid the mechanical stop with a servo-system thus eliminating the preloading of the mechanical chain. You can avoid the mechanical "clamping" using a drive without self-locking as the worm drive is but then you have a risk of "back -bouncing" which could compromise the position repeatability. To give you feeling about the self-locking threads have a ratio P/d about 1/10 and are self-locking for friction coefficients down to 0.05.

A further thought:-

If you need to go to a hard stop, but the motor/gearbox pushes with a stiff spring only the target, not to a spring, the hard stop will always be exactly in the same place, but the pressure will not be as high and maybe it will not lock on.....

Here is a (very) simple diagram:-

If the spring is fairly stiff (experimentation needed), I think that it will allow you to shut the motor off, but not hang too heavily on the end stop.....

I still prefer the stepper motor idea best of all though.......and it would be relatively cheap too....

I finally changed the motor so that I can get bigger reverse torque (15 times bigger than the forward torque), and the problem solved.

I'd like to thank everyone for providing all kinds of help here.

What a waste of brain power this one is.. Redfred got it right first time..Guest takes little or no notice and goes and does something completely different despite all of the input you give him..

Another option is to reduce the gear ratio and run the motor proportionally slower. The softer angle-of-attack on the gears might be sufficient, although one presumes that gear lubrication is all that it can be.