Servoless leg design

General robot discussions

Re: Servoless leg design

Postby circuit » Fri Jun 13, 2008 10:51 pm

Well, all that motor positioning electronics is not very tricky. As you can see in this drawing, there are blue "flowers" on each motor. I'll mount miniature optical encoders there.
Image
I'll use separate microcontroller for each leg. H-bridges as well. All that microcontroller programming and electronics stuff is not a problem to me, I live from it :) The only concern is about those motors.
Those long red things in my drawing are warm gears:
- Step: 0.4mm;
- Length: 35mm;
- Width: 2mm.

These motors are running at ~6000 RPM, so this gear should reposition in about 0.9 seconds. This is not very fast, so reduction rate is kinda high.
These motors have no ball bearings, but they do not need ones - all pressure lies on warm's bearing.

To reduce overall wight I'll make hollow legs. That's made simple with laser :)

Total cost is not very high: I've bought those motors for $1.5 each, ball bearings for $0.40 each. Laser-cutting should cost somewhere about $30. Of cause electronics will add some $$, but it won't be more than $20, as I am designing everything on my own.
circuit
 
Posts: 6
Joined: Thu Jun 12, 2008 4:50 pm

Re: Servoless leg design

Postby paulp » Fri Jun 13, 2008 11:50 pm

I've done a bit of work with worm drives myself. They give you quite a bit of accuracy but at those speeds they will need to be watch precision to prevent vibration. Also, the further they are wound through, i.e. the nearer the nut gets to the motor, the greater the vibration thats experienced. Having said that at lower speeds you could achieve a very smooth, realistic movement.

You may also suffer some of the problems that many stepper motor designs have and thats initial positioning at power on. Its fine counting pulses from a known start point but if you are no where near the expected start then its a lot of turning to get there. End of range sensors would help there but the sensor count and therefore the cost goes up.

Its an interesting proposition to say the least.

You mentioned no bearings, even if the motor isnt supporting the main load, they are designed to be virtually free running in a phone. When under load you will get torque twisting of the armature which will tend to cause a side-ways thrust at each end and wear whatever bearing material is there. Usually plastic.
paulp
 
Posts: 279
Joined: Wed May 21, 2008 8:30 pm
Location: Midlands U.K.

Re: Servoless leg design

Postby Matt Denton » Sat Jun 14, 2008 9:20 am

MetalSkullRobotics wrote:By the way, the servo's that paulp mentioned are probably the ones I use. 3 dollar each, shipped in from China. I ordered 60 of these and when these arrive i will certainly post my findings about them. Like paulp said the quality will be very low, and the power that they should produce (3kg/cm) won’t be precise. But like I said elsewhere, I'm just using them to test the simulation of the walking movement.

For around 10dollars each you can buy nice 12kg servo's. Have a look on http://www.bidproduct.com or http://www.unitedhobbies.com if you are looking for some affordable servo's.


The biggest problem with cheap servos, is their linearity. Essentially this means they all move different amounts when applied with the same signal. This can be overcome in software, but only by calibrating each servo with at least three calibration points. I tried to use some cheap servos once, complete headache! My motto: Buy cheap.. buy twice!
Matt Denton
AKA: Winchy_Matt

micromagic systems ltd
Matt Denton
Site Admin
 
Posts: 1622
Joined: Tue May 20, 2008 9:15 pm
Location: Winchester UK

Re: Servoless leg design

Postby MetalSkullRobotics » Sat Jun 14, 2008 10:21 am

Matt Denton wrote:My motto: Buy cheap.. buy twice!


Like i said, i know they aren't useful for the "real" hexapod. But I rather pay 80dollars extra and know it will work then pay 300dollars and find out it wont be working at all.
MetalSkullRobotics
 
Posts: 68
Joined: Tue Jun 03, 2008 1:19 pm
Location: Zoetermeer, the Netherlands

Re: Servoless leg design

Postby Matt Denton » Sat Jun 14, 2008 6:07 pm

I'm with Paul on this one, I think your a bit shy on bearings. I can see there is a bearing at the front of the motor housing, but only having one will not stop angular displacement of the worm shaft. Also with a linear drive system you need thrust bearings, otherwise the thrust loads are transferred directly onto the shaft of the motor. Even a motor that has good radial bearings doesn't come with thrust bearings. If you take a look at the router I built form my CNC hexapod, I added a thrust bearing that sits between the dremmel collet adaptor and the faceplate of the motor to take thrust loads on the end of the tool. With your design you have to take both pulling and pushing loads!

I have used linear drives within animatronics, and they work very well as a drive system for robots, I started to build a hexapod about 1200mm diameter using custom made linear drives, I will post a photo of the leg design when I find one. The one disadvantage of a linear drive system is not being able to back drive the joint.
Matt Denton
AKA: Winchy_Matt

micromagic systems ltd
Matt Denton
Site Admin
 
Posts: 1622
Joined: Tue May 20, 2008 9:15 pm
Location: Winchester UK

Re: Servoless leg design

Postby Matt Denton » Sat Jun 14, 2008 6:15 pm

I can't find the linear drive pictures, but this is the test leg we built and converted to linear drives. In these images the leg is using 32mm maxon motors, planetary gearbox's with right angle drives and my own PID motor drivers. We changed to femur and tibia joint to use linear actuators which worked much better with less backlash in this application.

Image

Image
Matt Denton
AKA: Winchy_Matt

micromagic systems ltd
Matt Denton
Site Admin
 
Posts: 1622
Joined: Tue May 20, 2008 9:15 pm
Location: Winchester UK

Re: Servoless leg design

Postby circuit » Mon Jun 16, 2008 1:30 pm

That leg is really huge! Is it going to be a hexapod?


Disregarding motor power, I am continuing working on CAD drawings. So here it is:
Image

I will do some more cutouts to lose weight.
circuit
 
Posts: 6
Joined: Thu Jun 12, 2008 4:50 pm

Re: Servoless leg design

Postby Matt Denton » Mon Jun 16, 2008 1:51 pm

circuit wrote:That leg is really huge! Is it going to be a hexapod?


It was a test leg built while I was working on Harry Potter 4. It won't be going into production :)
Matt Denton
AKA: Winchy_Matt

micromagic systems ltd
Matt Denton
Site Admin
 
Posts: 1622
Joined: Tue May 20, 2008 9:15 pm
Location: Winchester UK

Re: Servoless leg design

Postby blegas78 » Sun Jan 17, 2010 4:29 am

I just saw your work here, great progress! I look forward to seeing the completed hexapod.


blegas78
 
Posts: 110
Joined: Mon Feb 09, 2009 5:34 pm
Location: Tucson, Arizona

Re: Servoless leg design

Postby Matt Denton » Sun Jan 17, 2010 6:05 pm

Nice work!! are those small stepper motors with linear ball drives? if so, I presume you need to calibrate them after a power up?
Matt Denton
AKA: Winchy_Matt

micromagic systems ltd
Matt Denton
Site Admin
 
Posts: 1622
Joined: Tue May 20, 2008 9:15 pm
Location: Winchester UK

Previous

Return to Other

Who is online

Users browsing this forum: No registered users and 1 guest

cron