Rover suspension ideas needed

[jwatte]

I'm building a rover for light outdoors navigation (Robomagellan.)

I'm an accomplished software engineer, I've had electronics as a hobby since I was 10, but I was a clueless newbie when it comes to mechanics. I've taken some classes in metalworking and CNC milling, and have paid my dues in broken bits to the point where I can mill parts out of stock pretty reasonably, using 3-axis machining. Centering, alignment, and workholding for 4-axis is so far a big mystery to me, but I'll probably take a class and start figuring it out at some point.

My ambitious goal is to place in top 3 at Robomagellan at Robogames in San Mateo 2013.

I started with an RC truck and got all the main concepts down. Microcontrollers -- easy. Software -- check. Sensors -- all integrated over I2C with little slave AVR chips driving them. Motor control -- I built my own from N-channel MOSFETs. It all worked pretty well, but the RC truck chassis (a 10 year old plastic toy from Nikko) started to turn to mush, even though I put a frame on top of it built of 3/4" L-angle aluminum. Also, I found that I really need encoders on the wheels to know how I'm going, and the RC truck just ain't gonna make that happen.

It's time to build a more suitable chassis.

The mechanical components I have, and likely won't change ('cause I'm all bought out at this time :-):
- Pololu 37D gear motors with encoders, 4 of them + 1 spare
- RC tires, 7" diameter, 14 mm hex, 4 of them
- Pololu 18V25A CS motor controllers, 2 of them (one for each side, or one front/one back) + 1 spare
- Strong hobby servos for steering (Solar D772,) 4 of them + 1 spare
- Assorted thrust, needle, and ball bearings
- Aluminum blanks, steel round bar blanks
- Assorted springs, nuts, washers, bolts, etc

I'd like all four wheels to be able to individually steer and drive. I'd like the wheels to be sprung/suspended somehow. And that's where I run into trouble. I have to come up with a linkage where a direct-drive motor is mounted to an up/down (or torsional) suspension, that then turns clockwise/counterclockwise for steering.

The best I've come up with so far looks like this:






it's hard to see from the perspective -- there's a thrust bearing at the bottom of the chassis; there's a hole where a tall needle bearing holds an axle for steering through the bottom plate. All the hooks/springs are underneath the plate. There's an axle for suspension threaded through the steering shaft, that all the hooks are mounted on. The top hooks press against the thrust bearing and don't rotate much up/down; the bottom hooks, that the motor bracket in turn is mounted to, is the main moving part for suspension, with springs attached at the back.

I know how to build this and assemble it. However, it is weak in that there's really not much preventing the entire assembly from falling out the bottom if I pick up the robot -- I need some kind of collar around the vertical steering shaft and/or the needle bearing. (The idea behind the needle bearing being so tall is that it takes most of the sideways twist, to make the axle mostly turn in the horizontal plane.)

Also, it looks... not like something a real mechanical engineer would build :-)

Another idea was to just mount the motor to the bottom of the shaft, and use a spring around the shaft as suspension (up/down instead of twisting around a secondary axle) and keep the upper part as-is. However, it seems even harder to come up with a good fastener for the axle on the topside, that allows turning while engaged to the top plate (full suspension extension) while still staying on while the suspension is fully compressed. Also, because the top moves, some kind of linkage would be needed for the servo steering -- preferrably something that doesn't cause turning of the wheel as the axle moves up and down. I'm totally at a loss for what that would look like.


I looked at getting RC chassis/axles that are pre-made to work with this, but they have at least two problems:

1) it'll easily cost me $500 or more for two axles and a bare bones skeleton in the size I'm going for
2) all the axles use differentials, so one motor per axle, not four independent motors, which makes  the motor encoders not know which wheel is slipping

So, I'm looking for any kind of reference, suggestion, idea, part, or epiphany to get to the next step. Thanks for reading this far :-)

[Soeren]

Hi,

2) all the axles use differentials, so one motor per axle, not four independent motors, which makes  the motor encoders not know which wheel is slipping

If that's your only worries about diffs, just use wheel encoders

[jwatte]

Thanks for your answer -- it's certainly a good idea to check entirely different solutions, too. No need over-complexificating things more than necessary.

If that's your only worries about diffs, just use wheel encoders

I looked into that, but it has several draw-backs that don't work for me.

1) A differential means that if one wheel slips, the axle doesn't drive. If the robot has the left side on a slippy thing (or hanging in the air while the bottom is grounded,) then the right side won't drive at all. Not good.

2) I have not found any wheel encoders that are sealed and would fit in the kinds of wheels I am using. I have found some very expensive hardware, and I have found some affordable, but non-sealed and non-robust hardware. This robot will run on soggy, dirty, lawns and probably romp over flower beds (entirely unintentionally of course) so, for example, the Parallax wheel encoders aren't quite enough.

So, given that I have some parts already, and would like to learn the mechanics of suspensions a little better than "I know next to nothing," are there any mechanics-like ideas that come to mind?

[Soeren]

Hi,

I looked into that, but it has several draw-backs that don't work for me.

1) A differential means that if one wheel slips, the axle doesn't drive. If the robot has the left side on a slippy thing (or hanging in the air while the bottom is grounded,) then the right side won't drive at all. Not good.

That's only valid for an "open" differential. with 50..80% lock it would still apply some power to the ground contact wheel.
Ball differentials, as used in R/C cars, can be adjusted for the amount of lock.

2) I have not found any wheel encoders that are sealed and would fit in the kinds of wheels I am using. I have found some very expensive hardware, and I have found some affordable, but non-sealed and non-robust hardware. This robot will run on soggy, dirty, lawns and probably romp over flower beds (entirely unintentionally of course) so, for example, the Parallax wheel encoders aren't quite enough.

There's more than one way to shave that goat.
One very rugged method is, to use a (steel) spur wheel and a Hall switch and the only way that would give is, if you were running it over endless amounts of iron dust.

So, given that I have some parts already, and would like to learn the mechanics of suspensions a little better than "I know next to nothing," are there any mechanics-like ideas that come to mind?

Use two bearings a distance apart for rigidity on the turning axle and use a parallelogram-like suspension. to keep the axles from falling out, in case the bearing/axle fit is wrong, a washer and a pin through a hole, or lock rings set into groes in the axle will hold it in place.
If you can bend the turning axle to pivot right over the center of the wheel, you'll get better steering.

Study the wheel suspension on older race cars where the wheels are a bit outside the main body, hanging in two "A-frames" for ideas.


You should be aware that the kind of steering you want has been abandoned by many experienced folks, as it is a time consuming pain in the sitter to get independently controlled wheels to behave well - and if they don't, you loose development time as well as traction.
It's not uncommon for beginners to have a (too?) high level of ambition paired with ideas of making exotic steering geometries, but the ones that actually win any races (seldom for rank beginners though, as they're usually up against more experienced people), are the ones that go with proven methods and apply them well. Simple works

[jwatte]

One very rugged method is, to use a (steel) spur wheel and a Hall switch and the only way that would give is, if you were running it over endless amounts of iron dust.

That actually sounds pretty good! I wish I could have fit that on the RC truck, and it wasn't made out of old, withering plastic...

A locking differential RC chassis might be a good backup for when my individual steering fails miserably :-) But, most things I've learned in life, I've learned through failure, so head-first into uncharted waters :-)

Study the wheel suspension on older race cars where the wheels are a bit outside the main body, hanging in two "A-frames" for ideas.

That's great advice! Thank you!

For reference: That brought me to the Wikipedia article about "double wishbone suspension," with a side link to "McPhearson struts." I'm sure car fanatics are groaning at my discovery of things invented even before I was born (which was a long time ago,) but my life so far has been on a different path.

Also, not sure I could make a suitable ball joint for a "true" double wishbone, and I prefer making my own over buying not because of quality, but because of learning.

Anyway, I got what I asked for: solid advice and ideas for different directions to investigate, so I'm a happy camper. Thanks again!

[jwatte]

Thanks to the advice in this thread, I will only steer on two wheels. I am also looking at a-frame or double wishbone suspensions. I want to build every part if I can,  but the a frame design simplifies steering decoupling and hub mounted motor.
Turns out, ball joints with quarter inch fittings are very cheap - less than $25 for ten!

Next question is one of axles and bearing fit.

I ordered some .125 steel rod and some 1/8" inner diameter flange ball bearings. However, the bearings don't fit on the rod. Measuring with calipers, I get the rod at .125 spot on but the bearings at .115" inner diameter. (Calipers are not great for this measurement, but this was as good as i could get)Is this an expected result?

How can I make sure to get bearings and an axle that will fit?

[Th232]

Not sure where you bought the bearings, but I'd be sending an email back to the vendor.  Even if there's supposed to be an interference fit for the bearings, 0.115 vs 0.125 is way off.

One way to make sure things will fit is to turn the axle yourself, but that's dependent on access to a lathe.  You did say you wanted to build every part if you could, right?

[jwatte]

One way to make sure things will fit is to turn the axle yourself, but that's dependent on access to a lathe.  You did say you wanted to build every part if you could, right?

I could use a manual lathe at Tech Shop if I took their class on it... Maybe now's the time :-)

(Also, there's a fourth axis/lathe for the Tormach CNC that I already use there; there might be a turn/mill option for me, too)

[Soeren]

Hi,

I ordered some .125 steel rod and some 1/8" inner diameter flange ball bearings. However, the bearings don't fit on the rod. Measuring with calipers, I get the rod at .125 spot on but the bearings at .115" inner diameter. (Calipers are not great for this measurement, but this was as good as i could get)Is this an expected result?

You can check the inner diameter by grinding down a plastic axle/stick until it fits snuggly, then measure the axle
Put the axle in a drill and let it run, while pressing a piece of emery paper on a flat stick (very lightly) against the axle - check the size frequently.

If both axle and bearing is 1/8" AND they're both perfect down to molecule level, you'll still need to press fit the bearing onto the axle.
Given a little roughness, you may need to cool the axle a couple of hours in a freezer, while heating the bearinga bit - say to ~45°C - don't go much higher, or you'll damage the lubricant.
To press it on, remember to use a method that only presses on the inner ring of the bearing (or you will hurt it).

[jwatte]

Given a little roughness, you may need to cool the axle a couple of hours in a freezer, while heating the bearinga bit - say to ~45°C - don't go much higher, or you'll damage the lubricant.

I would never have thought of that! Thanks a lot for the practical advise!