It is currently Mon Nov 18, 2019 3:29 am




Post new topic Reply to topic  [ 146 posts ]  Go to page 1, 2, 3, 4  Next
 Affordable low backlash worm gearing? 
Author Message

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Affordable low backlash worm gearing?
Hi there,

I've been scouring the net for the past few days looking for worm-gear reductions for my rig (why worms? because they provide the highest reduction ratios with the least amount of backlash and are self-locking, which eliminates the need for additional brakes).

Turns out I have two options so far:
- buy ready-made worm gearboxes - which are insanely expensive, starting with ~$200 for 30-40 arc-minutes of backlash and going upwards of $350-$400 for less than 8 arc-minutes
- source separate gears and worms - which are a bit more affordable, in the area of $80 for the pair - but this leaves me with the tricky option of having to build myself the whole gearing holding set.

The only decent alternative to worm gearing is to use timing pulleys and belts + a motor brake (not quite so elegant, plus it complicates the design quite a bit).

So, anybody out there with knowledge of some more down-to earth prices for worm gearings?
Thank you so much!


Sun Feb 24, 2013 3:09 pm
Profile
User avatar

Joined: Mon Oct 04, 2010 5:00 pm
Posts: 253
Post Re: Affordable low backlash worm gearing?
I don't know if any of them will fit your needs but servocity.com has some.


Sun Feb 24, 2013 9:46 pm
Profile

Joined: Tue Apr 05, 2011 6:16 am
Posts: 487
Post Re: Affordable low backlash worm gearing?
A bit of warning for rolling your own -- component gears should have an AGMA quality rating that will indicate the best possible backlash for that gearset, but then the problem becomes you need to machine the mounting to a very high level of tolerance (for example, 1 thou of deviation in center mounts for a gearset can add almost a thou of backlash).

What is a level of acceptable backlash for you? Is it a linear or rotary move? Your pricing seems on-par with quality gears in small numbers. If 30-40 arc minutes is acceptable, then you can DIY that with a mill and some time.

If you're doing linear motion, you know, backlash doesn't really matter as much, as you can (and nearly everyone does) compensate for it in the control system. If you're doing fine rotary work, the fall-forward effect is major problem just past the apex.

Why do you need a brake? Are you intending to use steppers and the cut power between moves? If not, then the holding torque should be spec'd to the requirement. DC motors can be put into brake as well.

!c


Mon Feb 25, 2013 6:54 am
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
Hi, shutterdrone,

First of all, I excluded the indivitual-worm-and-gear solution too - because reading further led me to the exact conclusion as you.

I need these reductions for precise rotary movement - basically, for tilt and roll. In pan and dolly backlash is irelevant, indeed, but in tilt and roll it's a real pain. I will probably use timing pulleys in the end, combined with a motor brake. Safest and cheapest bet. Although getting custom timing pulleys to my specs (min. 5mm bore for the small pulley, max. 120mm outside diameter of large pulley and 1:10 ratio) is not cheap nor easy, either...

As for stopping the motors - yes, I plan to take my jig out in the field - that's what I'm building it for - so every milliamp of saved battery counts...


Mon Feb 25, 2013 7:33 am
Profile

Joined: Tue Apr 05, 2011 6:16 am
Posts: 487
Post Re: Affordable low backlash worm gearing?
erosnicolau wrote:
As for stopping the motors - yes, I plan to take my jig out in the field - that's what I'm building it for - so every milliamp of saved battery counts...


I understand, but how much battery would you need to retain holding torque, and how much would those batteries weigh and how much cost would they add relative to the additional weight and cost of an external brake? =) Are you intending to microstep to add additional resolution? If so, you must remain powered on at all times any how*. We often under-appreciate one factor for another in system design. If you had a design that could push a 3lb camera using 40mA @ 12V, you could run that axis between 30-60 hours (depending on the efficiency of your driver) on a 2700mAh 12V Li-Ion battery pack, which weighs about 4-6 oz's in most cases.

* You could do some second-order validation to ensure that you've arrived at an exact 0.9' or 1.8' position on the motor shaft before cutting power, but once you add that, you could do SMS-style shooting without fear of any backlash -- you just use your encoders to verify position before firing the next shot, and fall-over becomes irrelevant.

!c


Mon Feb 25, 2013 8:09 am
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
Hi, again,

I understand, from you answer, that you're not familiar with those darn stepper motors... They consume, while holding a position, just as much as they consume under full load - which is the max. amps the motors are spec'ed for.

I know that disabling the stepper between shots defaults it to the nearest full step - so this will render micro-stepping useless. I guess our software will include 2 options in this regard:
- high power source: working with micro-steps (which permanently power the passive brakes, so they release the motors), while keeping the motors charged at all times. huge energy guzzler, but very accurate
- low power source: working only with full steps, and only powering the brakes, for release, while moving the motors. this will work best out in the field, when battery life IS a concern.

The controller for the system will be a Raspberry Pi connected to a 6-drivers special Arduino board, so I will have multiple consummers:
- the RPi (rated 0.7mA) + the Arduino (almost negligeable)
- the motors (6 motors rated 1A each) - but only running around 5-10% of the whole time-lapse time
- the brakes (4 x 24V/0.27A) - also only consuming while the motors are running

Tell me if this config can still run on 12V-2.7A for 20-30 hours, and you've made my day! As I was expecting to need at least 12V/16A for 12 hours.

P.S. As for the encoders: still not sure how to use/config those... Plus my understanding is they cost a lot as well... Plus my intuition tells me they should be placed on the output of the reduction, not on the input (motor). What do you reckon?


Mon Feb 25, 2013 8:21 am
Profile

Joined: Mon Nov 14, 2011 11:34 pm
Posts: 401
Post Re: Affordable low backlash worm gearing?
erosnicolau wrote:
I understand, from you answer, that you're not familiar with those darn stepper motors... They consume, while holding a position, just as much as they consume under full load - which is the max. amps the motors are spec'ed for.


Not necessarily ...... as an alternative you could use a microstepping controller that allows you to dynamically change and therefore limit the max current during braking ... Not all steppers use max power on hold .... Mine don't .... There are different ways to approach this ....

_________________
For further info re keyframing see >> http://keyframemoco.com

"tick-tock ....click-clock ...."


Mon Feb 25, 2013 8:43 am
Profile

Joined: Tue Apr 05, 2011 6:16 am
Posts: 487
Post Re: Affordable low backlash worm gearing?
erosnicolau wrote:
I understand, from you answer, that you're not familiar with those darn stepper motors...


You would be mistaken.

(Consider why would I have mentioned "push the load @ 40mA", and then extrapolate 40mA out to the entire runtime...)

You sound like you've made up your mind to use a non-standard pulley set with brakes. So, good luck to you!

Quote:
P.S. As for the encoders: still not sure how to use/config those... Plus my understanding is they cost a lot as well... Plus my intuition tells me they should be placed on the output of the reduction, not on the input (motor). What do you reckon?


They cost a lot less than those brakes you're thinking of, and don't consume any appreciable amount of power... Consider where you'd place the encoders and how you could combine encoders with planetary gear boxes (much denser ratio, much less expensive than worm boxes) and low-multiplier pulleys*. You might wonder why I talk about system issues so much, you're welcome to ask.

!c

EDIT: Clarification: * - a proper timing belt setup, e.g. HTD, will have no backlash, the planetary gearbox will, so, obviously, if you place it on the planetary output shaft leading to a 3x belt reduction, you'll backlash detection and a 3x resolution increase for your encoder.


Mon Feb 25, 2013 8:47 am
Profile
User avatar

Joined: Wed Nov 11, 2009 10:45 am
Posts: 1685
Location: Merritt Island, Florida, Estates Unitas
Post Re: Affordable low backlash worm gearing?
Shutterdrone is right about the tight tolerances for making worm gearboxes. Professional machinists with a lot more skills than I have, tried this three times and never got a low enough backlash for a decent P/T head.

It matters what axis you are moving, how much weight, is it balanced, and even wind. Trying to build one module that gets reproduced and used to move every axis becomes kind of wasteful. Backlash has very little effect on a dolly and can be compensated for in software. This axis usually needs the most torque. Backlash can be very bad in a P/T head for two reasons. An unbalanced load moving past its balance point will move within the backlash range (going from riding the front face of the gearteeth to riding the back side of the tooth) which makes a little klunk in the timelapse. Wind can also blow the head around within the backlash range of the gears, especially if there is a mattebox, hood, or other sail-like device attached to the camera. A real simple solution is to not make the bearings so good. A little bit of drag introduced into the drive train will resist the wind, yet be overwhelmed by motor torque when the move is done. Its easy to make a little friction brake which can be adjusted to produce as much or as little drag as necessary. It could even be adjusted for specific shots (more brake if shooting in a big wind).

Learning that cost a lot of money. For telling you, there is no extra charge.

Focus motors can be a lot smaller than the other axes. You need a lot of gear reduction to work DC gearmotors with the MX-2 controller, but steppers seem to need no reduction.

Check out this controller unless its the one you are already planning on using.

viewtopic.php?f=24&t=9863

Are you planning on having printed circuit boards made for connecting your stepper drivers to the arduino? You might save a lot of money by coordinating with John B. who is making the controller mentioned above. I think he was planning on building a driver board for the big easydrivers. You might also talk to Brian at eMotimo, I think he's developed a similar board for the TB-3 black edition. Buying a copy of an existing board will be cheaper than having a custom board made. An arduino mega has plenty of gpio for eight axes. No point in stopping at six. It would be great if you could make an eight axis driver board for John B's controller (unless he's already done it, in which case, you can just buy his.

Tell us more about what you are trying to build and we'll be able to help more.


Mon Feb 25, 2013 8:51 am
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
Quote:
You would be mistaken.

@shutterdrone I take it back, and then some! :)
Your news are too good to me to ignore! You're the first to tell me that steppers COULD in fact use less power to just hold torque. And since you've already worked with this kind of problems, I come to you with a plea: please, tell me how to achieve this low-consumption during stationary times.
Quote:
You sound like you've made up your mind to use a non-standard pulley set with brakes. So, good luck to you!

I haven't made my mind about anything yet - and those darn brakes are a real and honest pain in my back... so any solution to do without those brakes is more than welcome!
Quote:
They cost a lot less than those brakes you're thinking of

The brakes I found cost around 90$/piece - however I have managed to find Nema17+brake combos at around 55$/combo. Please guide me to a nice source of affordable encoders... I would definitely use those, should I only know how to. Can I count on you, for the next few days, to answer a couple of questions about encoders?
Quote:
EDIT: Clarification: * - a proper timing belt setup, e.g. HTD, will have no backlash, the planetary gearbox will, so, obviously, if you place it on the planetary output shaft leading to a 3x belt reduction, you'll backlash detection and a 3x resolution increase for your encoder.

Not sure I totally understand the setup you're telling me about. Could you please detail? Thank you so much!


Mon Feb 25, 2013 9:47 am
Profile
User avatar

Joined: Wed Mar 16, 2011 2:58 pm
Posts: 1348
Location: Denver, Colorado
Post Re: Affordable low backlash worm gearing?
shutterdrone wrote:
erosnicolau wrote:
I understand, from you answer, that you're not familiar with those darn stepper motors...


You would be mistaken.



Lol! Yep.

erosnicolau, I would not recommend on making any assumptions on the knowledge level of most people on here. ;) One of the reasons i like this site is you have a lot of builders on here. Very solid foundation of knowledge. I have learned quite a bit!

I also want to chime in on the difficulty of making a good low backlash gearbox. My build partner and I are no strangers to machining, and even with an accuracy level of 1/1000th of an inch it is still difficult to built a good gearbox. 3 prototypes down for a panning mechanism and we have a very low backlash gearbox, but I am still not satisfied.

The reality is backlash will be a fact of life with almost all drive trains. If you want to get rid of it entirely, Harmonic Drives. Be prepared to shell out a small fortune. Backlash in itself is not an unsurmountable problem though, there are many ways to skin a cat.

_________________
http://www.BioLapse.com
http://www.TheChronosProject.com


Mon Feb 25, 2013 10:53 am
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
sciencelookers wrote:
A real simple solution is to not make the bearings so good.

Nice one! It kind of goes on the same line as the split-disk worm gears, only more adjustable. Thank you for the tip! ;)
Quote:
Focus motors can be a lot smaller than the other axes

Yup, I figured out (with a rope and a bag and coke cans for want of a better precise weight measurement unit) that a zoom/focus ring on my 70-200mm (which is the max. for which I'm spec'ing my rig) needs around 1Kg of weight to move around. Divided by the ring's radius (~4cm) I got a minimum torque of around 250-300g-cm. So for these two rings I'm planning on using 600-900g-cm NEMA steppers.
Quote:
Check out this controller unless its the one you are already planning on using. viewtopic.php?f=24&t=9863

If you're referring to the Big Easy, then yes, it's one of the versions I'm planning to use.

Just to make it clear: I'm no electronics genie, no, Sir, not by far. I had to search a lot until I came up with this config:
- Raspberry Pi as a PLC
- Megatronics V2.0 http://reprapworld.com/?products_details&products_id=327&cPath=1591_1606 as an Arduino interface to 6 stepper drivers
- 6 stepper drivers - still undecided, still open to suggestions (@shutterdrone, please don't forget me with info on drivers that are capable of considerably reducing consumption while in holding position):
- - from Pololu: http://www.pololu.com/catalog/product/2133
- - from Trinamic http://www.motioncontrol-community.org/?page_id=353
- - from Schmaltzhaus: http://www.schmalzhaus.com/EasyDriver/ / http://www.schmalzhaus.com/BigEasyDriver/

Quote:
An arduino mega has plenty of gpio for eight axes. No point in stopping at six.

Agreed! For now, however, I have to keep an eye out on costs as well. And since I'm no electronics engineer, I'll have to make do with whatever comes in handier - the 6-driver Megatronics, for now. I sent out the purchase order for the Megatronics boards (3 of them, as we plan to initially build 3 parallel prototypes) but if any better ideas spring from here, I can halt the payment for those boards. As I said, I'm not "definitive" on anything except for the RPi-s, so far.

Speaking of ideas and coordination, I suggested the porting of the pretty smart chip from Trinamic to Pololu's micro format - and apparently the idea caught on ;)
http://forums.reprap.org/read.php?88,186139,186927#msg-186927 - you'll recognize some of what I just wrote here too ;)

What I'm trying to build is "simple": a 6-axis, modular, portable (how else?) time-lapse motion controller. The 6 axis are (you probably guessed), from the ground up: dolly, pan, tilt, roll, focus, zoom. I know that smarter cameras and lenses know how to "focus" remotely, but I figured this could come in handy anyway, for older lenses, for instance. For now I'm still figuring out various things:
- what drivers to use (able to keep up with 1-1.5A steppers, able to micro-step, able of some sort of stall-prevention or step loss prevention, able to make the stepper hold with minimal energy consumption etc.)
- how to provide position feedback - with encoders, of course, I just haven't figured out yet how they work, how to position them, where to find affordable, decent resolution ones. this could solve both step loss, stalling detection and backlash correction and control... if only I learned how ;)
- what is this whole caboodle going to drain in terms of Amps
- what portable power supply to rely on
etc.

Thank you for your feedback: please keep it coming, as I see that I have much to learn from here! :)


Mon Feb 25, 2013 11:15 am
Profile

Joined: Tue Apr 05, 2011 6:16 am
Posts: 487
Post Re: Affordable low backlash worm gearing?
erosnicolau wrote:
@shutterdrone I take it back, and then some! :)
Your news are too good to me to ignore! You're the first to tell me that steppers COULD in fact use less power to just hold torque. And since you've already worked with this kind of problems, I come to you with a plea: please, tell me how to achieve this low-consumption during stationary times.


Well, that was John that said that *grin* I, personally, don't much like those drivers (and they'll cost a lot more than the easydriver's you're thinking of!) You could, however, fabricate your own drivers. Take a look at our nanoMoCo as an open-source design you can start from: http://dynamicperception.com/products/n ... controller

My question to you: why are you needing 6x 1A motor axes? I can think of 6 axes as linear, pan, tilt, roll, focus, and zoom. What is the payload? Why do all axes need to run at 1A? Certainly focus and zoom can run on NEMA-8 motors, at 20-120mA, and won't need braking of any sort (even backlash should be less relevant in most cases). I can run 15lbs of payload linearly at 1500 full steps/second with NEMA-17 on a 19:1 planetary gearbox @ 12V/600mA. We can greatly exceed that with a smarter driver that does automatic resonance detection. We've done testing with a 100:1 rotary worm gearbox using a NEMA-11 motor at 200mA that was able to push a 5lb load at about 700 degrees per minute in full steps (that's ~ 1100 steps/second, and in certain occasions we got as high as 3,000 steps/second using damping/resonance control). With 10:1 boxes, we move back to NEMA-17 and around 600mA. Once you're into worm gear ranges, you're looking at lower speeds with much reduced torque requirements. You should calculate a 50-70% efficiency in commercial gearboxes, or about 35-50% efficiency in DIY worm gear boxes. Some people might get plastic gears and mash them together hard to reduce backlash, these will run at around 35% efficiency and have a shortened lifespan.

The only axes that would need to remain powered on, as you note, would be your tilt and roll (unless you're in heavy, heavy wind, so you'd probably want pan held too).

Remember that you only need to provide enough current to your steppers to meet the specific torque requirement (you can always under-torque them as needed), and as much voltage to meet your speed requirements.

If I were focusing on low power consumption in the field, I would ditch that Raspberry Pi in a heartbeat. Why do you need to power a GPU? Without knowing the details of your control logic, I'm wondering what an RPi does for you in this case that a few AVR chips can't? HDMI displays seem like a luxury in the field with power constraints =) If you just need more instructions per second and more memory, consider the Arduino Due, which would replace the RPi and the Arduino (likely Uno or Mega) you were planning to use, reducing the power consumption, software LoC, overall cost, and complexity of the system. TI also has some nice ARM boards at a low price.

Also note, running big easydrivers at 1A without active cooling is a crapshoot. Expect to need big heatsinks, and they just buy you time to build up the thermal load without a fan.

Quote:
The brakes I found cost around 90$/piece - however I have managed to find Nema17+brake combos at around 55$/combo. Please guide me to a nice source of affordable encoders... I would definitely use those, should I only know how to. Can I count on you, for the next few days, to answer a couple of questions about encoders?
Quote:
EDIT: Clarification: * - a proper timing belt setup, e.g. HTD, will have no backlash, the planetary gearbox will, so, obviously, if you place it on the planetary output shaft leading to a 3x belt reduction, you'll backlash detection and a 3x resolution increase for your encoder.

Not sure I totally understand the setup you're telling me about. Could you please detail? Thank you so much!


Since you're likely doing a 1-off, just buy used/surplus encoders. Right now there's a deal on eBay for 6x 2500 ppr hollow shaft encoders for $150. Otherwise, consider the following:

If I have a motor, with dual shafts, and the front shaft is in a mounted planetary gearbox of 19:1, I put a 200-count encoder on the rear shaft (non-geared), and one on the gear box output shaft. This means my front shaft pulses once per 19 steps, and my rear shaft once per step. If I measure 19 steps on the rear, I should measure one one the front - if I don't, that means I have backlash, and I keep stepping until the front makes one step. If my front shaft goes through the encoder, and then onto a pulley, that transfers via a belt, using a precision drive type belt (with rounded teeth), to a pulley 3x larger than the one on my shaft, I now have no backlash at the final 3x multiplier output. This means I have a total gear reduction of 3 * 19 = 57:1, with high accuracy at full-steps. The NEMA-17 motor, with gearbox, runs ~ $60-70 (Phidgets and others sell motors with planetary gears, but they tend to have only a single output shaft, you can get the same motor with dual shafts from China via Alibaba.) With that setup, your full-step accuracy would be 11,400 steps per revolution, or a minimum full-step of 0.03158 degrees per step. You want macro through, so you go to microstepping, which means you leave power on, but can now go to 0.01974/mS or even lower depending on the driver. If you want more accuracy for each of these steps, you then need to increase the resolution of the encoders. You of course, can use one, higher resolution encoder on the gearbox output shaft, and simply correct for whether or not it was a missed step or backlash.

What happens when you brake and power cycle the motor, even at full-steps? What holds the motor at an exact full step between power cycles? What are the factors that can cause a motor to not return to the same detent between power cycles, even in full stepping? Is your start-up cycle at the driver perfectly clean? If not, ghost steps can be had - if you release the brake before powering the coils of the motor, only detent torque will be present, causing movement from position anyhow. How much accuracy do you actually need, and is that accuracy measured over one frame, or dozens? I've noted a lot of cases of the easydrivers causing detent jump when re-activating the driver. Brian S. did a good job on those boards, but they're optimized for cost and he can't account for what you attach them to...

If your primary system constraints are cost and power consumption, then optimize for those first - figure out where you need torque and power, and hit those first.

!c


Mon Feb 25, 2013 11:17 am
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
Jack Ripper wrote:
shutterdrone wrote:
erosnicolau wrote:
I understand, from you answer, that you're not familiar with those darn stepper motors...


You would be mistaken.



Lol! Yep.

erosnicolau, I would not recommend on making any assumptions on the knowledge level of most people on here. ;) One of the reasons i like this site is you have a lot of builders on here. Very solid foundation of knowledge. I have learned quite a bit!


Again, I stand corrected. I didn't mean for a minute to imply @shutterdrone wouldn't know what he was talking about. On the contrary, I assumed he chose the higher, finer, more expensive and reliable path of servos...

As for what can be found on this forum... you can't imagine how I'm stalking my mail lately for notifications on your replies, guys! ;)


Mon Feb 25, 2013 11:20 am
Profile

Joined: Tue Apr 05, 2011 6:16 am
Posts: 487
Post Re: Affordable low backlash worm gearing?
erosnicolau wrote:
Yup, I figured out (with a rope and a bag and coke cans for want of a better precise weight measurement unit) that a zoom/focus ring on my 70-200mm (which is the max. for which I'm spec'ing my rig) needs around 1Kg of weight to move around. Divided by the ring's radius (~4cm) I got a minimum torque of around 250-300g-cm. So for these two rings I'm planning on using 600-900g-cm NEMA steppers.


Also remember that if you use a large flexible toothed gear for zoom/focus, and a smaller toothed gear on the motor shaft, you are doing a gear reduction, and reducing the torque requirement of the motor.

!c


Mon Feb 25, 2013 11:26 am
Profile
User avatar

Joined: Wed Mar 16, 2011 2:58 pm
Posts: 1348
Location: Denver, Colorado
Post Re: Affordable low backlash worm gearing?
First off, i really commend you for your intentions here. I'm the kind of guy who swaps a clutch before learning how to replace the oil. So I have a very strong appreciation for ambitious DIY projects and jumping in head first.

The task you are undertaking is pretty big.

I am sure you have heard that you need to learn how to walk before you can dance the Tango.

if you are building a one-off, i really recommend looking at shutterdrones mocobus jobbers. you can find them on Dynamic Perceptions website. they will probably make your life a LOT easier. From that point all the software and control are really figured out for you, leaving you to worry about the mechanics.

That also gives you the ability to focus on one part at a time. First get the linear motion down and working how you want. Then look into the Pan, then tilt, then roll, then lens, etc. etc.

Trying to do it all at once is going to make your head pop.

_________________
http://www.BioLapse.com
http://www.TheChronosProject.com


Mon Feb 25, 2013 11:36 am
Profile
User avatar

Joined: Tue Nov 24, 2009 7:55 am
Posts: 951
Location: UK
Post Re: Affordable low backlash worm gearing?
Quote:
Certainly focus and zoom can run on NEMA-8 motors, at 20-120mA,


I ran Focus and Zoom using NEMA-8 motors on a Canon 24-105 but have since swapped to NMEA-11. I find they run cold in my application using BED's and have the advantage of being compact enough lengthwise to fit the Focus motor between the Focus / Zoom ring gears instead of protruding to the edge of the lens with the 8. They are larger in width so need a larger gear to stop them fouling the lens body.

_________________
Smile, it gives your face something to do!

http://www.vimeo.com/user1214873/videos
http://www.youtube.com/user/MrReggub


Mon Feb 25, 2013 11:49 am
Profile
User avatar

Joined: Mon Nov 22, 2010 1:18 pm
Posts: 1294
Post Re: Affordable low backlash worm gearing?
Think we been here before actually we here before a few times. These post may help
viewtopic.php?f=39&t=8192
viewtopic.php?f=43&t=9236
viewtopic.php?f=24&t=8416
viewtopic.php?f=18&t=8209
Ps I don’t have backlash and zoom and focus is not a axis :D


Mon Feb 25, 2013 11:50 am
Profile
User avatar

Joined: Wed May 25, 2011 11:07 pm
Posts: 205
Location: Portland OR
Post Re: Affordable low backlash worm gearing?
Thanks for the links and all the info guys, I have nothing yet to contribute here but am planning to add focus and zoom drives to my rig. Subbed to see where this goes.


Mon Feb 25, 2013 12:04 pm
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
@shutterdrone hehe I see we were writing stuff at the same time.

Part of your questions are already answered in my previous post. To detail:

- I don't plan to use the same config 6 times. On the contrary, I have calculated the torque loads for all 6 motors (and came up with NEMA 17 motors for dolly, pan and roll, a NEMA 23 motor for tilt and NEMA 11 motors for focus and zoom). Of all these, only Tilt and Roll actually need precise, low-backlash motion control (hence encoders) - although I think Pan could very well benefit too from the same feature. Focus and zoom will simply have 1cm-diameter pulleys and that'll be that. For Pan, I think of using some sort of reduction, for the sake of finer/smooth stepping, but backlash won't be a problem there.

- The reason I said the drivers should be able to handle 1-1.5A is that most free, low footprint drivers, can't usually handle more than 2-2.5A without active heating - which i'm trying to avoid. Starting from this consideration, over-spec'ing the drivers, or under-spec'ing the motors helps, imho, with the drivers heating problem. I do plan to use heat-sinks anyway, just not fans. Why the same driver for all motors, when the focus motors need much less? Simply for the sake of uniformity, for not having to code different code for them.
Note: in here comes very handy a good question: do you control all Arduino-compatible drivers in the same way, with the same code?

- As soon as you guys bring me up to date on encoders and position correction (and I hope for "real soon!"), then I'm 100% on board with the phidgets (or whatever) gearboxes, backlash or not. This will enable me to use smaller motors, hence less amps. One of the 3 partners involved in making this dream come true wants to be able to use this rig at higher speeds in a studio - so, taking this into account, we should be able to switch at any time with bigger motors and lesser reduction ratios. (we're 3 guys: I'm doing the general design, research, sourcing; there's a good friend handling all the programming - you'll see what programming - and there's the third friend with mechanical engineering skills and tools, that comes to complete the team. we definitely needed an electronics guy too, but we came short in supply of that ;)). Bigger motors for higher speed will also mean stronger drivers (able to withstand more amps)

- BTW, I'm having a hard time finding planetary gearboxes able to exert more than 30kg-cm on the output... any suggestions here?

- RPi will be indeed a power hog. However, our plan is to interface and control the whole thing via a very user-friendly php/html GUI, hosted on the RPi server, accessible via mobile phone/tablet. This kind of "click and drag" interfacing is very important to all 3 of us partners in this project.

- If EasyDrivers start failing under 1A, then I'll probably settle for something stronger, from the other options

- For the rest of your message - the part about the encoders - I will stop for now: to allow me time to read through very carefully and rummage things through ;)


Mon Feb 25, 2013 12:08 pm
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
shutterdrone wrote:
erosnicolau wrote:
Yup, I figured out (with a rope and a bag and coke cans for want of a better precise weight measurement unit) that a zoom/focus ring on my 70-200mm (which is the max. for which I'm spec'ing my rig) needs around 1Kg of weight to move around. Divided by the ring's radius (~4cm) I got a minimum torque of around 250-300g-cm. So for these two rings I'm planning on using 600-900g-cm NEMA steppers.


Also remember that if you use a large flexible toothed gear for zoom/focus, and a smaller toothed gear on the motor shaft, you are doing a gear reduction, and reducing the torque requirement of the motor.

!c


Mmm I was rather thinking of using a fixed length round belt, driven around the focus/zoom ring and the around the motor pulley, and the stretch/pre-tension that belt by adjusting the angle of the arm holding the small motor. Is this a bad idea? In doing so, using a 10mm diameter pulley gives me a 5mm/0.5cm lever arm, which basically doubles my motor's strength.


Mon Feb 25, 2013 12:13 pm
Profile
User avatar

Joined: Wed May 25, 2011 11:07 pm
Posts: 205
Location: Portland OR
Post Re: Affordable low backlash worm gearing?
erosnicolau wrote:
shutterdrone wrote:
erosnicolau wrote:
Yup, I figured out (with a rope and a bag and coke cans for want of a better precise weight measurement unit) that a zoom/focus ring on my 70-200mm (which is the max. for which I'm spec'ing my rig) needs around 1Kg of weight to move around. Divided by the ring's radius (~4cm) I got a minimum torque of around 250-300g-cm. So for these two rings I'm planning on using 600-900g-cm NEMA steppers.


Also remember that if you use a large flexible toothed gear for zoom/focus, and a smaller toothed gear on the motor shaft, you are doing a gear reduction, and reducing the torque requirement of the motor.

!c


Mmm I was rather thinking of using a fixed length round belt, driven around the focus/zoom ring and the around the motor pulley, and the stretch/pre-tension that belt by adjusting the angle of the arm holding the small motor. Is this a bad idea? In doing so, using a 10mm diameter pulley gives me a 5mm/0.5cm lever arm, which basically doubles my motor's strength.



I was thinking along these lines for my focus/zoom. I was thinking of having the cam on rails and mounting the motors on a bracket like the one linked below, so I can just loop the belts around the motor pulley, and adjust the tension by changing the angle with the quick release clamp.

http://www.proaimshop.com/pas/-CT-15P-1 ... PLATE.html


Mon Feb 25, 2013 12:19 pm
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
@MikeA - I'm guessing it all depends on how you set up the zoom-focus motors relative to the lens...

@DISPLACEMENT 1 - thank you for the links - I'm on them! I'm sure I'm not raising any new questions here - the only thing new here is me ;)
P.S. You're a lucky one not to have backlash :p and focus and zoom sure aren't axis, but sure can use motors with axis ;)

@lifeskills - with another hinge idea, this is exactly how I plan to drive my focus/zoom, too! :)


Mon Feb 25, 2013 12:33 pm
Profile

Joined: Tue Apr 05, 2011 6:16 am
Posts: 487
Post Re: Affordable low backlash worm gearing?
erosnicolau wrote:
@shutterdrone hehe I see we were writing stuff at the same time.

- The reason I said the drivers should be able to handle 1-1.5A is that most free, low footprint drivers, can't usually handle more than 2-2.5A without active heating - which i'm trying to avoid. Starting from this consideration, over-spec'ing the drivers, or under-spec'ing the motors helps, imho, with the drivers heating problem. I do plan to use heat-sinks anyway, just not fans. Why the same driver for all motors, when the focus motors need much less? Simply for the sake of uniformity, for not having to code different code for them.
Note: in here comes very handy a good question: do you control all Arduino-compatible drivers in the same way, with the same code?


BigED, nM, Pololu stepper board, all use the same driver, have similar amounts of copper, and won't be reliable north of 1A. If you need to be reliable north of 1A for more than a few seconds, you'll need to upgrade to proper drivers. Any driver you deal with is going to be a step/dir style driver, so don't worry about the code. Spend money only where its needed.

Quote:
- BTW, I'm having a hard time finding planetary gearboxes able to exert more than 30kg-cm on the output... any suggestions here?


30KG/cm? That's ~ 2700 oz/in. You're going to be looking at a motor/gear combo that's going to weight almost 2lbs alone. You're also looking at around NEMA-34 for a non-geared motor. Are you sure you calculated that correctly? I have two triple-stack NEMA-34's rated at 1700 oz/in's that weight about 15lbs each. But, if you really need that - Anaheim Automation has what you need, the gear motor will be north of $200.

Quote:
- RPi will be indeed a power hog. However, our plan is to interface and control the whole thing via a very user-friendly php/html GUI, hosted on the RPi server, accessible via mobile phone/tablet. This kind of "click and drag" interfacing is very important to all 3 of us partners in this project.


That's one way to do it... =)

!c


Mon Feb 25, 2013 12:34 pm
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
shutterdrone wrote:
Right now there's a deal on eBay for 6x 2500 ppr hollow shaft encoders for $150

Couldn't find it... Do you still have the link?


Mon Feb 25, 2013 12:36 pm
Profile

Joined: Tue Apr 05, 2011 6:16 am
Posts: 487
Post Re: Affordable low backlash worm gearing?
erosnicolau wrote:
shutterdrone wrote:
Right now there's a deal on eBay for 6x 2500 ppr hollow shaft encoders for $150

Couldn't find it... Do you still have the link?


Here ya go: http://www.ebay.com/itm/6-Stegmann-HS20 ... 3f221e89ef

!c


Mon Feb 25, 2013 12:45 pm
Profile
User avatar

Joined: Sat May 07, 2011 10:24 am
Posts: 316
Location: Netherlands
Post Re: Affordable low backlash worm gearing?
[quote="erosnicolau"
- RPi will be indeed a power hog. However, our plan is to interface and control the whole thing via a very user-friendly php/html GUI, hosted on the RPi server, accessible via mobile phone/tablet. This kind of "click and drag" interfacing is very important to all 3 of us partners in this project.
[/quote]

That is only true when you are using model B of the RPI. The model A uses a lot less juice. But then you are loosing the network connector and one USB port. But that still makes it very possible to use a sick interface.

_________________
eMotion Control Store - http://www.elysiavisuals.com
Exposure ramping with Nikon and Canon - http://www.elysiavisuals.com/catalog


Mon Feb 25, 2013 12:53 pm
Profile

Joined: Mon Nov 14, 2011 11:34 pm
Posts: 401
Post Re: Affordable low backlash worm gearing?
:D I must admit ... This is the best thread for a while :)

_________________
For further info re keyframing see >> http://keyframemoco.com

"tick-tock ....click-clock ...."


Mon Feb 25, 2013 1:07 pm
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
@cronix well, thanks, mate, for telling me this now! ;)
the RPi's are already ordered and on their way... version B, to be sure...

@JohnB thank you, mate, I sure am happy to be part of it and be learning all these new things!

@shutterdrone any reading on how to use encoders, for me? ;)


Mon Feb 25, 2013 1:20 pm
Profile

Joined: Tue Apr 05, 2011 6:16 am
Posts: 487
Post Re: Affordable low backlash worm gearing?
erosnicolau wrote:
@shutterdrone any reading on how to use encoders, for me? ;)


Well, erm, there are several ways to go - but since you're talking about Arduino, start with the Arduino Playground article on the subject: http://playground.arduino.cc/Main/RotaryEncoders

!c


Mon Feb 25, 2013 1:46 pm
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
shutterdrone wrote:
erosnicolau wrote:
@shutterdrone any reading on how to use encoders, for me? ;)


Well, erm, there are several ways to go - but since you're talking about Arduino, start with the Arduino Playground article on the subject: http://playground.arduino.cc/Main/RotaryEncoders

!c


Man, are you resourceful today, or what! Thank you so much!!


Mon Feb 25, 2013 1:55 pm
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
DISPLACEMENT 1 wrote:
Think we been here before actually we here before a few times. These post may help
viewtopic.php?f=39&t=8192
viewtopic.php?f=43&t=9236
viewtopic.php?f=24&t=8416
viewtopic.php?f=18&t=8209
Ps I don’t have backlash and zoom and focus is not a axis :D


...The first link is special to me: it's how I first discovered your forum :D


Mon Feb 25, 2013 2:48 pm
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
shutterdrone wrote:
If I have a motor, with dual shafts, and the front shaft is in a mounted planetary gearbox of 19:1, I put a 200-count encoder on the rear shaft (non-geared), and one on the gear box output shaft. This means my front shaft pulses once per 19 steps, and my rear shaft once per step. If I measure 19 steps on the rear, I should measure one one the front - if I don't, that means I have backlash, and I keep stepping until the front makes one step. If my front shaft goes through the encoder, and then onto a pulley, that transfers via a belt, using a precision drive type belt (with rounded teeth), to a pulley 3x larger than the one on my shaft, I now have no backlash at the final 3x multiplier output. This means I have a total gear reduction of 3 * 19 = 57:1, with high accuracy at full-steps. The NEMA-17 motor, with gearbox, runs ~ $60-70 (Phidgets and others sell motors with planetary gears, but they tend to have only a single output shaft, you can get the same motor with dual shafts from China via Alibaba.) With that setup, your full-step accuracy would be 11,400 steps per revolution, or a minimum full-step of 0.03158 degrees per step. You want macro through, so you go to microstepping, which means you leave power on, but can now go to 0.01974/mS or even lower depending on the driver. If you want more accuracy for each of these steps, you then need to increase the resolution of the encoders. You of course, can use one, higher resolution encoder on the gearbox output shaft, and simply correct for whether or not it was a missed step or backlash.

Very clever solution! I can sense that this solution (comparing the 2 encoders and waiting for confirmation from the gearbox encoder) only works for dolly/pan, when shifting direction of movement (when the camera moves LESS than it's supposed to)
How do you see this solved for tilt/roll, where gravity tends to "steal" the head, and the camera moves MORE than it's supposed to?
shutterdrone wrote:
What happens when you brake and power cycle the motor, even at full-steps?

From what I found out so far (but again, I might be mistaken), any stepper motor defaults/moves to the closest full-step, when disabled. In theory, I see two possible work scenarios:
Scenario 1: (hoping this is the right one in real application)
Braking:
- I reach the desired micro-step position and I hold that position until I apply the brakes, then I disable the motor
Resuming:
- With the brakes on, I re-enable the motor and tell it to go to the current position and hold still. It won't budge, but it'll think it did move, so it'll stay put, on the correct (previously set) position
- Once I gave it that "ghost" command and time to think it executed the move, I release the brakes and then tell the motor to go to the new position
Scenario 2:
Braking:
- I reach the desired micro-step position and I hold that position until I apply the brakes, then I disable the motor
Resuming:
- With the brakes on, I re-enable the motor and tell it to go to the current position and hold still. It won't budge, being braked
- I release the brakes, and the motor may try to go to the closest full step. If this (second scenario) happens, then I should be able to estimate what that closest full step is, and compensate.

Without working with micro-stepping, the problem tends to get simpler, as I would always have only full steps, to which the motor defaults anyway when powering up again - the drivers make sure that the motor powers up in the exactly same full step.
shutterdrone wrote:
What holds the motor at an exact full step between power cycles? What are the factors that can cause a motor to not return to the same detent between power cycles, even in full stepping? Is your start-up cycle at the driver perfectly clean? If not, ghost steps can be had - if you release the brake before powering the coils of the motor, only detent torque will be present, causing movement from position anyhow. How much accuracy do you actually need, and is that accuracy measured over one frame, or dozens? I've noted a lot of cases of the easydrivers causing detent jump when re-activating the driver. Brian S. did a good job on those boards, but they're optimized for cost and he can't account for what you attach them to...

- The brake should hold the motor still between power cycles.
- About error factors - those are to be discovered, hopefully not the case, but hope is not enough. I don't know.
- Accuracy calculation is in progress. I want to be able to get smooth frame transitions when using up to 200mm lenses - although most of the cases I'll be using wider angles. By the way, this is not meant to be used for macro, so the wild precision requirements of macro are not so necessary here.


Tue Feb 26, 2013 4:37 am
Profile

Joined: Tue Apr 05, 2011 6:16 am
Posts: 487
Post Re: Affordable low backlash worm gearing?
erosnicolau wrote:
Very clever solution! I can sense that this solution (comparing the 2 encoders and waiting for confirmation from the gearbox encoder) only works for dolly/pan, when shifting direction of movement (when the camera moves LESS than it's supposed to)
How do you see this solved for tilt/roll, where gravity tends to "steal" the head, and the camera moves MORE than it's supposed to?


How would it move without rotating the motor shaft? If it rotates the motor shaft, then the encoder could detect it, presuming the movement is greater than the minimum encoder detection. Besides, you'll find that transitioning to a belt output will often inhibit much of any fall-ahead effect. (Consider what happens to torque elicited upon the camera mount, and how it would travel back to the planetary gear box.)

erosnicolau wrote:
- Accuracy calculation is in progress. I want to be able to get smooth frame transitions when using up to 200mm lenses - although most of the cases I'll be using wider angles. By the way, this is not meant to be used for macro, so the wild precision requirements of macro are not so necessary here.


Encoders and brakes are probably overkill. Why not try something cheap, like a belt transition from a planetary geared stepper, and see where that takes you? (Remember, you can always tune out the directional backlash via software.) Heck, replacing that RPi B with another device that has lower power consumption will open up enough power to keep a moderate stepper turned on the entire time =)

!c


Tue Feb 26, 2013 6:26 am
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
Quote:
How would it move without rotating the motor shaft? If it rotates the motor shaft, then the encoder could detect it, presuming the movement is greater than the minimum encoder detection. Besides, you'll find that transitioning to a belt output will often inhibit much of any fall-ahead effect. (Consider what happens to torque elicited upon the camera mount, and how it would travel back to the planetary gear box.)

Mmmm... I thought THAT's gearbox backlash, at the end of the day: the amount by which the output shaft can move independently from the input shaft. That little jiggle...

Quote:
Encoders and brakes are probably overkill. Why not try something cheap, like a belt transition from a planetary geared stepper, and see where that takes you? (Remember, you can always tune out the directional backlash via software.) Heck, replacing that RPi B with another device that has lower power consumption will open up enough power to keep a moderate stepper turned on the entire time =)

Encoders and brakes are a pain, I know... but for lack of a better one...
As for the belt output: do you refer to that 1(gearbox output):3(arm joint) ratio you were giving as an example previously? That's similar to the nut adjustable brake, is it not? Because if that's the ratio you're talking about, then the arm joint has 3:1 leverage on the gearbox shaft. I may be on the wrong page here, please detail :)
Again, I agree that direction-shift backlash can be accounted for and compensated with known values in software, but for gravitation-independent joints.
The RPi's are already paid for and on their way... If we'll use a solution with small motors with planetaries, then I guess the motors would consume less current in hold... we'll see :)


Tue Feb 26, 2013 7:07 am
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
Just thought I might update you guys on my further progress with the low-backlash quest.

I obtained some painfully expensive offers from Harmonic Drive (around 630-650eur / 1:100 reduction), so I decided to take matters into my own hands and go ahead and start experimenting with building a hypocycloidal 1:100 reduction.

So far I have a preliminary design which is under work on the lathe+CNC, but have managed to slap together a small, plexiglass prototype, full of errors (I had to manually adjust the lobes and pins, a total mess) ans STILL managed to get a 1.6* backlash out of it - less than the 1.8* of the Chinese 1:100 planetaries.

My estimate for the well done prototype (which I'm hoping to have ready in a matter of days) is of less than 0.1* of backlash - hopefully even more precise.

In the meanwhile, I also started experimenting with plans and designs for a timing belt harmonic drive ;)) let's see how this turns out, later.

Until then, here's my extremely crude 1:100, 1.6* backlash reduction made of 3 pieces of plastic. (see the 2 black marker lines on the sides of the disks)



Fri Mar 08, 2013 4:48 pm
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
Status update 2: the second prototype is finished. Body made of Duramid, inner wheels made of brass. Due to a CNC... backlash :)), there was a 0.2mm error in the X axis, which resulted in some harder contact spots and a somewhat irregular running of the reduction. Even so, the backlash resulted from this second, half-plastic version, amounts to around 15 degrees in the input, equal to 0.15 degrees in the output, or 9 arc-minutes.

We're already well under way with version 3, a more compact design where the outer pins will be replaced by 5mm bearings in pin cages, and where, having eliminated much of the error in the CNC, we're hoping for much better milling accuracy.

Best news of all: the system is self-braking ;) so no more need for encoders and/or brakes!

The belt-driven harmonic drive project is put on hold for now.


Attachments:
2013-03-11 14.50.jpg
2013-03-11 14.50.jpg [ 101.64 KiB | Viewed 21974 times ]
2013-03-11 17.40.jpg
2013-03-11 17.40.jpg [ 115.16 KiB | Viewed 21974 times ]
2013-03-11 17.48.jpg
2013-03-11 17.48.jpg [ 116.86 KiB | Viewed 21974 times ]
Tue Mar 19, 2013 7:19 am
Profile
User avatar

Joined: Wed Nov 11, 2009 10:45 am
Posts: 1685
Location: Merritt Island, Florida, Estates Unitas
Post Re: Affordable low backlash worm gearing?
Very impressive progress. Congratulations on tackling the hard bit.

What are you planning on doing with these drives once you're done? Are these one offs or do you plan to sell gear sets, complete gearboxes, or complete P/T heads and other stuff built around them? Is there any way I can buy several sets of your CNC'd parts and a list of the bearings and other parts that go into this? I'd like to experiment with these gear reducers for use in some of my projects. I'd also buy complete, assembled gear reducers if you prefer.


Wed Mar 20, 2013 11:33 am
Profile

Joined: Fri Feb 22, 2013 9:31 am
Posts: 70
Post Re: Affordable low backlash worm gearing?
Thanks, man! :)

We started developing this hypo-cycloidal 1:100 reduction from the need of a zero or at least very low backlash reduction for our time-lapse robot project. Since we're on a very tight budget and need a total of 12 reductions for our 3 robots, we can't afford the existing commercial reduction options:
- worm gears: self-braking, but with backlash a great deal of wear over time - starting with ~300$ a piece
- planetary gears: reversible (not self-braking, which is bad), with similar amounts of backlash and starting at ~200-250$ a piece
- harmonic drives: self-braking, zero backlash, perfect - but starting at ~850$ a piece

As for your question/request: just give us some more time (a couple of weeks, for piecing together the third, most precise prototype of our reduction) and then we could come up with some tested specs (torque, backlash, size, weight) and why not, we can try and see how much we could ask for some complete reductions.

Our plan is to research more affordable options for the time-lapse fan in me - and in everybody else on this forum for that matter - and, based on our findings, to come up with some more decent-priced solutions, ranging from reductions to omni-heads to full-systems. So... stay tuned :)


Thu Mar 21, 2013 2:09 am
Profile
User avatar

Joined: Wed Nov 11, 2009 10:45 am
Posts: 1685
Location: Merritt Island, Florida, Estates Unitas
Post Re: Affordable low backlash worm gearing?
Awesome! Can't wait to see what you come up with. It would be nice to see something other than the standard dolly-pan-tilt. Send me a PM when you have some reducers you can part with.


Sat Mar 23, 2013 7:34 pm
Profile
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 146 posts ]  Go to page 1, 2, 3, 4  Next


Who is online

Users browsing this forum: Majestic-12 [Bot] and 8 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Powered by phpBB © phpBB Group.
Designed by Vjacheslav Trushkin for Free Forums/DivisionCore. pozycjonowanie