motor load sharing

4pyros said:

I am finding bits and peaces with google; "differential drive motor controller" or try "dual axis motor controler"

Click to expand...

Those are certainly dual-axis motor controllers and such - but that's not the same as an "electronic differential"; I am not absolutely certain how an electronic differential works, but I would imagine that by knowing the axle length, the wheelbase distance (from front to back axle centers), the wheel diameter, and a few other parameters - by knowing the angle the front wheels are turned, the controller (which would certainly include drivers for each motor to control speed via PWM or such) would then be able to apply the proper speed to each motor so that the inner-wheel motor turns slower than the outer-wheel motor, proportionate to the circles and such each are tracking inside based on the steering wheel angle, etc. Probably would also have to monitor the current and/or some kind of encoder to account for slippage and such (and accelerate or brake the motors as appropriate).

I have not seen any controller that can do this, though I do know they exist:

https://en.wikipedia.org/wiki/Electronic_differential

I suspect that any controller you could buy probably would be designed for some regular size or larger electric vehicle (for instance, they show a mining truck in the article above that uses such a controller!) - and probably wouldn't be cheap (if one could be purchased at all).

I was planning on trying to implement an electronic differential for my platform, using a microcontroller (likely an ATMega168 or 328 programmed via the Arduino IDE - but as a standalone controller) - if someone doesn't get there before I do (at the rate I'm going, someone is likely to - no biggie, though), I plan on releasing the code on the Arduino forum, or put it in the playground, or on my site - somewhere; so that there is an open-source e-diff codebase available for others to use.

You're correct. Most of these controllers are made for somewhat larger vehicles. There is a small but growing industry around small EV's and this type of drive technique is catching on. Wether not not one can purchase such a drive is questionable, but my point was the issues brought up about it are not insurmountable. Here are a couple good sites I discovered that discuss some issues involved:

https://www.4qd.co.uk/evs/EDDS/Electronic_Differential_Motor_Controller.pdf
https://hal.archives-ouvertes.fr/docs/00/56/47/28/PDF/IEEE_TVT_2011_TABBACHE.pdf

It appears to be a pretty simple simple problem of solving the geometry. I consider this one of the most important applications for anyone with engineernig skills. I'd be happy to collaborate with anyone who seriously wants to work on this.

There is yet another way to drive wheels with seperate motors ( I just thought of it ) that doesn't require a electronic differential. It uses a mechanical coupling that can transmit power in only one rotational direction. When turning, the outside wheel sort of spins freely. This technique is used on some self-propelled lawn mowers. Not an ideal solution, but very simple.

BrownOut said:

You're correct. Most of these controllers are made for somewhat larger vehicles. There is a small but growing industry around small EV's and this type of drive technique is catching on. Wether not not one can purchase such a drive is questionable, but my point was the issues brought up about it are not insurmountable. Here are a couple good sites I discovered that discuss some issues involved:

Click to expand...

Hey - thanks for the links; I'll be sure to check them out. I figured that the EV industry (particularly the DIY segment) would be pursuing this; with a four-wheel drive system, you could even have 2 or 4 wheel steering handled electronically via such a differential system (and as the wikipedia link pointed out, you could do traction control and other fancy things too).

BrownOut said:

It appears to be a pretty simple simple problem of solving the geometry. I consider this one of the most important applications for anyone with engineernig skills. I'd be happy to collaborate with anyone who seriously wants to work on this.

Click to expand...

I think if it were as simple as it seems on the surface, we'd see more instances of it, and demos, etc in the hobbyist or DIY world - but if any exist, they're few and far between, which leads me to believe that there are some hard problems to solve looking under the surface.

I'm just a robotics hobbyist, and as I noted I intend to implement something like this for when I convert my PowerWheels H2 platform to pneumatic tires; until then, I can deal with the standard plastic ones. Right now, my biggest stumbling block has been getting the steering system set up. I recently purchased a Linak linear actuator for use on it (6 inch travel, 4000N force both ways), but at 12 volts (its a 24 volt actuator) it runs waaaay too slow for me (takes about 10-15 seconds to travel 1.5 inches; I think even at 24 volts it would be too slow). So, I am looking into other options to gain a faster, but powerful actuator (I have a B&D 6V cordless drill that I am now planning on making into a linear actuator using a threaded rod).

Once I get that in place, then its just a matter of building the drive controllers; initially, I'll probably use relays for everything (although the steering drive might be an electronic h-bridge, depending on the current requirements - so far everything has been under 4 amps at stall, even the Linak - but I am not sure yet what the current draw of the drill motor will be), but I do have an older Roboteq dual h-bridge (assuming it works - I got it surplus, and all the wires were chopped - that should be fun to fix) that I might put into place (for the main drive motors) - or, I'll purchase one of those other dual motor controllers like has already been posted.

At that point, then - I can start thinking more about the whole e-diff situation; I'll definitely keep your offer in mind as I approach that milestone (I'm sure I'll need all the help I can get - there has to be some kind of kink in the plans for an e-diff; it looks like it should be deceptively simple).

cr0sh said:

I think if it were as simple as it seems on the surface, we'd see more instances of it, and demos, etc in the hobbyist or DIY world

Click to expand...

If you look at the math involved, it's actually very simple. The reason more hobbists aren't using it is because there is really no particular advantage to in in most cases. Robot hobbists have been happy with dual differential drives forever ( sounds the same as electronic differential, but is actually different ), and with pretty good reason. Most of the drive comes from the EV people. But, EV hasn't really become commplace, and DIY EV is a very, very small world. There are some PHD papers that discuss drives that use all kinds of advanced mathematical models, but I think that's just fodder for demonstrate the capabilities of the authors rather than anything practical. After all you ony talking about replacing a simple mechanical system. How hard can that be?

As the speed increases so does the difficulty. Thats why robots dont need them but EVs do.

Robots usually have a much smaller "track" (distance between right and left wheel), and also the consequence of a robot wheel scrubbing or spinning is usually quite small but that would have a large consequence (safety, tyre wear, efficiency) on an EV.