More Mule then Rover?

[jdolecki]

Tommy what are you running for a motor controller?

Have you gotton any reading on the amps the motors draw?

My whole goal right now is to develop the programming for the AI

putting a payload or mower on it is easy

thanks john

I will post pics this weekend.

[zoomkat]

I was mostly referring to your posting style in general, not this thread in particular. That is, that you tend to lean on the negative side of things in most of your posts, yet have failed to show any real work of your own that would validate your largely negative and opinionated stance.

I think I would be providing a disservice to the poster if I kept quiet about potential issues and solutions I think might be important for the project. I guess some might consider "wear a lifejacket" as negative if they really don't to wear one. I often see projects that have critical issues that are ignored/overlooked and the projects usually fail. As to my fun projects/tinkering, I generally only do things to verify an idea I have is valid, using the minimum parts and $$$ required to get it done. I usually do my own car/truck/motorcycle repairs along with fixing the usual household problems, so actual bot building is currently down the elective list.

[Tommy_T]

Tommy what are you running for a motor controller?

jdolecki
I have 2 drives I'm using for setting up the control software, a
Sabertooth 2X25 (would like to have a 2X50 but can't find one)
and a P&G dual 70Amp drive that came with the first testbed
I now also have a MK5 drive from the new testbed but have not
looked at interfacing it yet(picked up last night), I'll look at that this weekend.

Have you gotton any reading on the amps the motors draw?

Everything tells me these motor have a stall current of @40Amp.
I have not taken any direct reading just what I'v found on the net.

Tommy

[jdolecki]

Here are 2 pics
First bare chassis no batteries or motor controller
Second is placementof the Zotac Mother Board

[Tommy_T]

jdolecki
Nice looking platform you have there, them motors look sweet, I have a set
of them foam filled wheels also. I find I do have to replace them due to slipping
on the grass while pushing a 20" mowing deck on my lawn with 20+ deg. slops.

My footprint is not much different then yours, that includes weight distribution,
but I found that if it has to stop fast it would nose over, in my case the mowing
deck would bottom out on the lawn, digging dirt at times. right now I using the
battery that came with the mowing deck which is much lighter then the 2 car
batteries that are going to be installed soon, that should help with traction and
help keep it from nosing over. if not 2 small casters should do the trick.

One of the issues I have to answer is the lost motion in the gearbox. both sets
of motors I have(each about 6 years old) have about 1 deg of Backlash. the fact
that I have 2 casters in the back and that my lawn is uneven them 2 caster cause
uneven forces on the frame(only one caster on the ground,the forces want to turn it that way,
Backlash lets it turn that way). the testbed I just picked up has a gearbox connected
to a 8" ball screw for seat tilting, looks like the thing can lift 300lb. I'm going to use that
to raise and lower a single wheel between the caster for when it is moving forward,
when turning it would be raised to let the caster take the load. this should help with
gear backlash and let me use the P&G drive because smaller corrections on uneven surfaces
would be needed(I think). if not I could try replacing the casters with fixed wheels and raise
and lower a caster for turning.

This is a trick dump trucks use I believe.

Tommy

I gotta think if there was one place where all the information learned doing
these projects could be found, I'm sure this nut could be cracked much faster.

[Adrenalynn]

Have you torn the motor controllers apart and looked at the boards? A lot of these controllers are actually OEM Curtis. And documentation/pinouts are fairly easy to come by.

High-current/high-voltage motor controllers are actually pretty available, as they're common in combat robotics.

I've used RobotPower / Sidewinder's that are 80A or even 160A / channel. I have other controllers that are up to double that.

[Tommy_T]

Have you torn the motor controllers apart and looked at the boards?

Adrenalynn
Yes, while putting a scope on it to find a way to interface it to my main controller(netbook PC)
I did get at look at it, it seems it's a Pic based motor speed controller, the OEM has a serial
programmer to set speed profile which is also available to Dealer, but not end users(for liability reasons they say).

High-current/high-voltage motor controllers are actually pretty available

That is true, and I had planed to use the http://www.roboteq.com/brushed-dc-mo...ontroller.html
but after interfacing the P&G drive to the main controller(netbook PC) and using it, the more I liked
it(for safety reasons), also the $600 cost could then be moved to tracking and navigation senors or
a set of tracks(not sure about them yet).

After playing with this project all week I'm beginning to think the problem is not the Acc/Dec loops.
In my head I figured once I get the feedback loops connected and programed the Sabertooth drive
could do the job under light loading(only 25Amps so could not stall the motor without frying the drive)
, but it had similar issues, much better, but not perfect.

possible problems.
1) Lost motion in the drive train. (a CNC with linear feedback and .010" lost motions would not be sable)
I currently have 155 encoder counts of lost motion, in the CNC machine world the only fix would
be to remove the backlash, or put the encoder on the motor for semi-closed loop control and add
backlash comp. I'm looking at adding a way to push the lost motion always in the same direction (see last post about adding third wheel)

2) A two wheel drive mixed platform is not sable enough for uneven terrain. this can be seen when
running open loop(no feedback), if one wheel hits a curb the platform will turn that direction.

still wondering if three wheeled would be better then four?(see last post about adding third wheel)

3) Programming, it's possible my "Tim the toolman" thinking is wrong. throwing a few hundred Amps
at the problem is not going to help. in my case I'v been increasing the opposing wheel's viscosity to force it
back on course. with the P&G drive as OEM programed the Dec. loop is ten times shorter then the
Acc. loop, maybe I should program it to decrease the opposing wheel's viscosity to let it catch up,
which should reduce the overshoot errors. the problem I see with that is the thing would get stuck
often. It's also possible to use the feedback devises to track it's location(dead reckoning) then use
the software to make correction in an absolute mode.(line follower where the line is in software)
The problem I see here is related to the only abs feedback I have is a Phigets 3/3/3 but
that can only give me a few degs. accuracy(ouch!).

Feel free to add to the list.


Tommy

[Adrenalynn]

>> the more I liked it(for safety reasons)

I have to ask what those "safety reasons" versus a combat robot motor controller would be.

>> also the $600 cost

That's about double the cost of a good 160A 2-channel motor controller, or half again the cost of an outstanding multi-channel.

>> a CNC with linear feedback and .010" lost motions

Fortunately, you're not building a mobile CNC, you're building a lawn mower... Now, if'n you're wanting to machine your yard, that's probably a whole different topic. Your CNC table [had better be] a darned site more level and your reference frame a whole lot more fixed, not to mention that the tables tend to be a pretty good bit smaller.

Robots, with hobby technology, oscillate a bit. What you're needing is a PID loop and maybe some good Kalman filtering to tame it down as much as you can.

[Tommy_T]

jdolecki
If you can man-up and post pictures I should be able to swallow my pride and post some also

the last two pictures are of the new(to me) testbed I got this week, it has very strong 4 pole
motors, and wheels that should give good traction(they are the same size as my current tires).
seems like a very stable platform, but only has 2.5" ground clearance, if I'm going to use it, that
will have to change.

first picture is my current testbed with the mower deck attachment(snow attachment not made yet),
I have not installed the payload dump box yet.

Tommy

[Tommy_T]

I have to ask what those "safety reasons" versus a combat robot motor controller would be.

Adrenalynn
Keep in mind I understand all the safety features built in the the P&G drive can be mimicked with a
microprocessor and a few solid-state relays, but when your dealing in the higher amps solid-state
relays get kind of pricey.

Safety features in the P&G drive
1) differential analog voltage from the joystick(which is where I interface to the Main controller(netbook PC)
to the P&G drive. this arrangement is less susceptible to EMF noise, the P&G hall-effect joystick has
5 analog outputs, if any one of these are missing or out of range the drive will shut down and flag a
fault.

2) If the main controller(netbook PC) shuts down for any reason, or the analog output unit fails the
P&G drive will shut down and flag a fault. all battle bot drives will take off full reverse. a good
example of this can be seen when powering up the drive with the main controller turned off, the
P&G drive will shut down and flag a fault, battle bot drive will take off.

3) built in current limiting, built in temp. monitoring, built in battery level monitoring.

4) built in motor brake control (24vdc)

What you're needing is a PID loop and maybe some good Kalman filtering to tame it down as much as you can.

Granted I have only spent maybe 20 hours trying to tune the PID loop(without the results I want), but I see the days getting shorter
and the nights getting cooler and I feel compelled to find an answer soon.

Tommy