M-stick chassis

[mike_o]

My M-stick project has progressed a bit, as I have - hopefully - solved the problem with "rear diff" stability by adding a centre bearing. I haven't tried assembling the thingy yet, but it may prove impossible :-)

For those of you who haven't read the previous posts on this, it's a two-motor, electronic diff. M-chassis size RWD car intended for carpet fun-class. Each rear wheel is driven by separate F-1 style direct drive BL motors, balancing power as a function of steering input. It will probably not work, but I just have to try it out :-)

I have also made a set of 34mm dogbones using my newly purchased J-B Weld. I'm curious to see how it holds up.

[kontemax]

That's great! Please more picts and specs.

Max

My M-stick project has progressed a bit, as I have - hopefully - solved the problem with "rear diff" stability by adding a centre bearing. I haven't tried assembling the thingy yet, but it may prove impossible :-)

For those of you who haven't read the previous posts on this, it's a two-motor, electronic diff. M-chassis size RWD car intended for carpet fun-class. Each rear wheel is driven by separate F-1 style direct drive BL motors, balancing power as a function of steering input. It will probably not work, but I just have to try it out :-)

I have also made a set of 34mm dogbones using my newly purchased J-B Weld. I'm curious to see how it holds up.

[mike_o]

Max,

thank you. A few pics:

Chassis with wheels

Chassis with body vs Swift body

Detail of spur gear, drive cups, and rear suspension

- Motors are 28mm diameter x 22mm length 120watt BL fitted to the sides of the "U"-profile.

- Battery is a 3S 1300mAh LiPo (same energy capacity as a 2000mAh 7.2V NiMh) which fits inside the "U"-profile.

- Two 30A ESC, one on each side of profile, possibly using profile as heatsink.

[actualday]

I have also made a set of 34mm dogbones using my newly purchased J-B Weld. I'm curious to see how it holds up.

How did you make your dogbones??

[mike_o]

Actualday,

Bought a set of 45mm bones, cut a piece of the mid section out and J-B'ed them together again inside a suitable (20mm x ø3, Ø5) metal tube

[kontemax]

Great project, can't wait to see it complete with motors.

Max

[qatmix]

Great project, can't wait to see it complete with motors.

Max

agreed, cant wait to see what happens

[Mad Ax]

Cool! I have a suspicion that the diff won't work as planned but only a full test-run will show how well (or otherwise) it works. Great work so far and keep us updated with progress!

[mike_o]

...I have a suspicion that the diff won't work as planned...

So do I :-)

But at least I have solved the drivetrain problem and managed to assemble the parts:

I even did a small test of the set-up. VIDEO HERE. Ran it with a servo tester off a 6 volt (5 cell) NiCd battery. Will run 11.1 / 3S LiPo eventually.

It runs nice and smoothly, but the tested 14:80 (1:5,7) gearing will have to go, unless I use higher kV motors. I also have a 18:70 (1:3,9:) option at hand.

[Leethal Driver]

What a cool project, nice work. Love seeing stuff like this, makes me wish I had the brains to come up with it

[speedy_w_beans]

This is a really neat project; I like the idea very much. You need a little Microchip PIC processor in between the radio receiver, motor controllers, and steering servo to do some basic math on the PWM (pulse width modulated) signals coming from the receiver and then regenerate new PWM signals to the motors and servo. That way you can bias the motors appropriately during turns so the inner tire doesn't unload and spin out of control; you can optimize torque applied based on how each tire is loaded. A few simple linear equations running in a continuous loop in the PIC should work well:

Throttle_left = (throttle_received * (1 - scaling factor)) + (steering_received * scaling factor)

Throtle_right = (throttle_received * (1 - scaling factor)) - (steering_received * scaling factor)

Throttle_received is -100% to 100%, steering_received is -100% to +100%, and scaling factor is likely a very small percentage to compensate for tire speed differences in curves.

The equations are set up so that the result for each throttle is still from -100% to +100%.

You can swap the additions and subtractions between throttle and steering if steering is reversed, or even reverse the steering in the PIC and keep the radio switch on normal.

This is so much nicer than a regular ball diff or gear diff. It's like the ultimate limited slip differential; there's also opportunity to dial in a little extra bias and help with steering response up front.

Want to take it to another level? How about two steering servos as well? If you had two steering servos running from the same Microchip PIC, there would be an opportunity to adjust steering center, toe in/out for each wheel independently, and custom Ackerman profiles.

Steering_left = steering_received + centering_offset + left_toe_angle + Ackerman_compensation_left(steering_received)

Steering_right = steering_received + centering_offset + right_toe_angle + Ackerman_compensation_right(steering_received)

Steering_received is -100% to +100%, centering_offset is -10% to +10%, left_to_angle is -10% to +10%, right_toe_angle is -10% to +10%, and Ackerman_compensation is a function that returns an interpolated result from a 2D lookup table capturing the nuances of your steering geometry.

There would be no compromise in each wheel angle for each desired steering angle.

So, for racing I really like the electronic differential idea. It would be very nice to also have independent electronic steering to get the angles just right. You have the potential to use each tire to its maximum by eliminating open diff effects and scrub from nonoptimal steering angles.

One last thought -- you could have two setup profiles for this, too. One profile would be for regular grip racing, the other profile could be for drifting. The grip profile is pretty self-explanatory from the discussion above. The drift profile could bias the rear motors differently to help maintain drift angles, and the Ackerman profile could be set up to generate an artificially parallel, non-Ackerman steering profile. So, maybe having two sets of equations would make sense for the same vehicle depending on what tires are mounted at the time.

There are plenty of USB-programmable microcontroller evaluation boards out there that could work for this project. It would be very cool to see it in action.

-Paul

[mike_o]

Paul,

thanks for your kind and elaborate input!

I'm thinking in the lines of programmable u-device, but since that is a project in it self, I have acquired a small 64-step V-tail mixer, that essentially does just that. Problem is, that it's a fixed program/algorithm, and the balancing might not suit my application. The only control I have have over system gain (throttle_right - a*steering, f.i.) is the mechanical gain/ratio of the steering servo. Also, the steps/increments are quite big compared to the normal fluid ESC response.

Alternatively, I am looking into a progammable transmitter, using three channels, and do the mixing in the transmitter. I'm checking whether the Futaba 4PK (or is it PL?) could do that. If not, I may have to use my Graupner MX16C 8-channel airplane tranmitter, interfaced to a RF-passive pistol transmitter. This would also become a project in itself....

Concerning the steering, I have considered sped dependent rear-wheel steering; At high speeds, the rear wheels turns in same direction as the front wheels (kinda parallel shifting the car rather than turning it) and at low speeds, counter turning rear wheels providing more narrow turns without the front wheels turning excessively. But since that has been a succes in 1:1 cars, it probably would not work on 1:10 either

Conclusively, I'll try to get the car running with the V-tail mixer and see how the intrinsic performance is. If it is really convincing, I may want to improve on it, but - as I'm not racing seriously - chances are that I'll loose interest in it and go for something else. An RC ekranoplane, perhaps....

Still, I appreciate all TC inputs and comments very much!

[MadInventor]

It's a shame this scale of car has such a narrow chassis area for the spur gear drives, or it would have been possible to go 4wd by running 2 belts to the front wheels. This would give a very efficient 4wd system, just an axle with a spur gear and belt pulley on each side at the rear, and a belt pulley at the front (Front bulkheads could have cammed bearing housings a-la schumacher CAT to eliminate the need for belt tensioners. You could even have one motor on one side of the chassis at the front, and another motor on the other side at the rear for ultimate weight distribution.

Perhaps a similar design in a 1/10 buggy for your next project if this one works well.

PS:

Before worrying too much about mixing the speedos with the steering servo input, I would just drive it and see what it handles like with 2 ESCs running from the same channel, you may well find that the outer motor just runs a bit quicker as the load is taken off it slightly during cornering. Slowing the inner motor down may cause very tight turning, like standing on the brakes on one side of a tractors braking system (2 brake pedals, one for each rear wheel).

[mike_o]

In it´s most evolved form, the car would have one sensored brushless motor per wheel, using the hall sensor to feed back information to a power management unit providing ABS and traction control, differential/power distribution function, transformation of brake energy to battery charging, and also some degree of stability enhancement. You could probably use the algorithms designed for 1:1 car systems, if the time constants were changed a bit....

[mike_o]

Now both motors in place. The first motor is easy to mount via the opposite holes, but the second motor is a true pain...

This is with the 18:70-pinion:spur gears. Now the wheels are spinning wildly at 11.1V

Rear shocks and LiPo in place:

Need to work on the springs, though. They have to be stiffer.

[mike_o]

I just did a test. Diff steering is strong - even without front wheels

VIDEO HERE

My concern about too much gain in the system seems to be valid. I may have to give up on the small V-tail mixer solution.

[speedy_w_beans]

For motor installation, what if you drill some extra holes in the chassis and use a hex driver with a ball end? You could then tighten or loosen the fasteners at an angle...

I watched the video and it's neat to see the car steering with just the rear wheels. I was surprised how matched the motors seem to be; you could drive in a straight line or steer in either direction pretty well.

Reading about standard servos and signaling methods, it looks like typical receivers put out a 1 ms to 2 ms pulse about 50 to 60 times a second for each channel. 0% is 1.5 ms, -100% is 1 ms, and +100% is 2 ms. That says the range of the PWM signal for each throttle is about 500 us in each direction. A lot of small microcontrollers can capture the rising and falling edges of these signals to within 1 us or 500 ns pretty easily. So, it should be possible to compute the mixing with 9 or 10 bit resolution (512 or 1024 steps), maybe even more depending on the specific part used.

-Paul

[MadInventor]

I just did a test. Diff steering is strong - even without front wheels

VIDEO HERE

My concern about too much gain in the system seems to be valid. I may have to give up on the small V-tail mixer solution.

I'd just try putting some front wheels on and running both speedos from the same channel. Your current solution is exactly what I used to steer my 10.5 kilo tank (Before the gearbox went shazbat). Have you noticed yet that you can do spin turns by not touching the throttle and then moving just the steering stick ?

[mike_o]

@Paul: I have shortened a hex-drive, and that does the trick, albeit slowly.

@Madinventor: Yes - I noticed the car spinning round itself when just turning the steering wheel Actually it's difficult to avoid because the V-tail mixer is quite agressive.

The problem is that the V-tail mixer controlling the ESCs needs less of the steering signal than the steering servo does. Unless I mechanically compensate for it, but that would weaken the steering and introduce too much slack. So I'm faced with 3 options:

1) Find or design a 3-channel mixing, pistol-type transmitter (the Futaba 4pk?, perhaps) and use my standard receiver without the V-tail mixer

2) Design a less steering sensitive (receiver-end) mixer myself, or try motivating somebody to do it

3) Moderate the steering input signal for the existing V-tail mixer by design of another, less complex, active device

Right now option #1 seems most feasible. If the Futaba won't do the trick, then possibly by piggy-backing a pistol transmitter to my 8-ch computerized flight transmitter... There goes the simplicity of my lightweight M-stick project...

[speedy_w_beans]

I dug around Futaba's site and I think the 4PL will do what you want at a price point that is lower than the 4PKS.

The manual for the 4PL can be found at http://manuals.hobbico.com/fut/4pl-2_4ghz-manual.pdf .

Page 66 of the manual talks about the programmable mixing function which is available for all channels. It looks like you can set up channel 3 and channel 4 with the throttle as the master and steering as the slave. Then you can adjust the amount of mixing and even make one mix inverted relative to the other, ala throttle_left = throttle+steering*gain, throttle_right = throttle-steering*gain.

I would send an e-mail to Futaba to double check this; the instructions imply it is possible, but I have questions as to whether the mixing amount is a gain factor applied during the mixing, or a limiting factor applied after the mixing. I also question if the mixing amount is applied to the master channel or the slave channel. Ideally the mixing amount is applied to the slave channel as a gain factor during the mixing. It would also be nice to know what is the resolution of the mixing and of the final result output on the channel to insure fine control.

I guess it doesn't matter which is the master and which is the slave -- you just want to assign steering to the master or slave such that it has the adjustable gain applied to it.

Connections:

CH1 = steering servo, normal function

CH2 = not connected

CH3 = left throttle, mixing mode

CH4 = right throttle, mixing mode

-Paul

[mike_o]

Paul,

thanks for your input. When I read the Futaba manual, I felt strangely familiar with the text Getting a sneaking suspicion, I dug out the Graupner XS-6 manual from the bottom of my archives and realized that it, too, has that mixer function. Since it is in German, I never took the time to read all of it.

I tried it out this morning, and I can indeed set channel 3 as a function (gain and off set) of the steering (or throttle) :D :D

So I can now have the channel 1 operate the steering servo at normal gain, channel 2 (normal throttle) feed the elevator input of the V-tail mixer, and channel 3 - set to 30% of steering_r and steering_l - feed the aileron input of the mixer, giving me what i want.

The most critical part of it - adjusting the differential power split - was there all the time, just waiting for me to see it. Thank you for pointing me in the right direction!

I'm so exited

[Yonez]

Paul,

thanks for your input. When I read the Futaba manual, I felt strangely familiar with the text Getting a sneaking suspicion, I dug out the Graupner XS-6 manual from the bottom of my archives and realized that it, too, has that mixer function. Since it is in German, I never took the time to read all of it.

I tried it out this morning, and I can indeed set channel 3 as a function (gain and off set) of the steering (or throttle) :D :D

So I can now have the channel 1 operate the steering servo at normal gain, channel 2 (normal throttle) feed the elevator input of the V-tail mixer, and channel 3 - set to 30% of steering_r and steering_l - feed the aileron input of the mixer, giving me what i want.

The most critical part of it - adjusting the differential power split - was there all the time, just waiting for me to see it. Thank you for pointing me in the right direction!

I'm so exited

Great thread,

Getting the steering algorithms just right is the main challenge here, but in a few years when all competition racers and hybrid cars use this design you can look back and say "I did this first"

For off-road cars this will be even better, with computer controlled diffs automatically compensating for traction loss and stuck wheels.

Living in snowy Norway I'm so looking forward to when this becomes standard in 1:1 cars. Traction control does some of it now of course, but mimicking independent control by applying brakes individually is just so crude.

I also love that you've put the motors on the chassis, limiting unsprung mass.

[speedy_w_beans]

Ahhhh, you beat me to it!

When I woke up this morning I started thinking about the "special mixing" mode on my 3PL, which is even less expensive than the 4PK or the 4PL. The 3PL works with the R2004GF receiver which has 4 channels on it. The 4th channel can be configured for 4WS for crawlers. I hooked up an extra servo to an existing setup and started playing with the transmitter settings, and was able to set up channel 4 for 4WS, normal or reverse, and use the end point adjustments to set the overall gain relative to steering on channel 1. I was also able to set the trim for channel 4 independently of the trim on channel 1.

So I was going down the path of using 4WS to feed the V-tail mixer as well, and just wanted to make sure the gain could be controlled with EPA. It worked well; I could go as low as 1% EPA in both directions and could control the offset with independent trim, although I noticed some range reductions when adjusting the trim. If your V-tail mixer has a null point adjustment, that would be better than adjusting trim on the radio.

Looks like you solved the problem of adjusting the V-tail mixer gain as well!

-Paul

[mike_o]

Paul, Yonez,

thank you. I've spend part of this Sunday pushing the project a bit further. It is now "complete" with front suspension and wheels.

I have even done a bit of driving, concluding that I need as little as 10-12% throttle gain mixing for the V-tail aileron input. A small VIDEO HERE

Next steps will be to sort out the suspension (springs too soft) and to fit smaller ESCs

1 already in stock, but 2nd ordered and stuck in Christmas mail pile somewhere. 2 x 25A at 11.1V should provide a peak power of 555 watts, so I won't need anything heavier than that.

As you can see from the video, there is quite some delay from front wheels turning until motors/ESCs react so I'll also need to look into the ESC setting for it to keep up with the servo...

[speedy_w_beans]

Very nice, that's looking good. I think the concept is coming together.

I think some of what you're observing is due to commanding very little throttle (to keep speed under control), the rotational inertia of the wheels and tires, and having no drag brake to pull energy out of the wheels while coasting. In the video the car is sitting on a stand, but I bet once you put the car on the ground the dynamics will change. The real tuning has to be done with the car on the ground. I'd be pretty excited to see how different tuning causes the car to understeer or oversteer; that would be a key tuning feature of this chassis and enable the front tires to steer more effectively.

-Paul