Yet another three motor swerve module… but this time more compact, and with maximum commonality to the proven and battle tested SDS MK4i module.
I’ve tried to retain as many components as possible from a MK4i, the only parts you need to assemble this are the three plates, and the falcon spline converter + pinion. This cad has it set up with a L2 gearing, and 18T pinions for a 5.25:1 overall reduction. All of the models are derived from a parent part, so it should be quite editible for different ratio selection.
Forgive me if this is trivial as I have no first-hand experience with SDS modules, but is it easy to switch this out to the L3 ratio? The main driver for adding another motor is to achieve the same or higher max speed with a decreased acceleration time, so going with anything less than the fastest ratio seems like you aren’t getting the maximum benefits out of the system.
The L1 L2 and L3 don’t change any center-distances, so you could use this module with L3 as is. Worth considering though is that once your drivetrain is current limited (which with 8 motors it will be), your best bet for more power is finding a way to increase the efficency of the motors - which gearing a little higher will acheive. Where that balance lies is something that will depend on your robot weight, cycle distance, etc
Not this one, at least yet. My team ran three motor this year, but on a billet chassis so this was my personal project of adapting it to use tube.
That being said, I machined three iterations of modules during the season using on our router, so if you have any questions with manufacturing I can probably answer them.
Well that’s awesome Jack. Much appreciated. Ill reach out when we get closer to giving it a try. We have a Laguna Router. Hopefully it is up to the task. I can’t see why it wouldn’t be. I was trying to find a way to avoid the need to cut a new lower plate, but I think the only way to make that work would be to invert or rather un-invert the second drive Falcon.
Electrical system limits from a non-ideal battery - more motors means more current draw (even all at the same current limit), which reduces the battery voltage, which in turn reduces the maximum current that can be pumped through the motor, which limits torque and therefor limits drivetrain acceleration.
What I haven’t played with is whether there’s a lower current limit that better trades off battery voltage drop for torque.
I think theres two distinct “problems” that the extra motor solves. Say we’re in a scenario with a hard cap of 200A system draw, theres two avenues to getting more power onto the carpet -
Keeping the battery current draw saturated at high RPM:
Pretty self explanatory. As the speed increases the current is going to drop off - the second motor is going to delay that, allowing you to continue accelerating and drawing max current higher into your RPM range. Instead of hitting your speed plateau at 4000rpm you might make it to 5000.
Increased efficiency at low RPM:
With our 200A system budget, that will be 50A for a 4 motor swerve, or 25A for 8 motor.
When you’re starting from a stop, the motor is going to be dominated by copper losses in the resistance of the windings. Because there is no back emf (or very little at low speed), you could imagine the motor as simply a resistor.
If you halve the amount of amps you want to flow a resistor, you can halve the supply voltage. You can think of the controller as a variable voltage DC-DC converter, turning 12V into whatever fractional duty cycle the motor is set to, because thats basically what high frequency pwm control is. Therefore from the battery’s perspective, the 8 motor instance is creating the same torque, but only drawing half the current.
Am I then really saying the 25A instance is twice as efficient as the 50A though?
The vex dyno test agrees with me - compare how much torque you get per amp stalling at 12V vs 6V. 12V gives 0.0182 N⋅m/A compared to 0.0355 N⋅m/A at 6V. Thats a 1.95x improvement.
Here, using vex’s dyno csv numbers, I’ve graphed 6V vs 12V efficiency on the same x-axis. The efficiency delta tracks 12V, again supporting the 2x efficiency argument at low RPM.
So while its very true that two motors is completely controlled by current limits, the main benefit is a large efficiency gain at low RPM, and an ability to accelerate closer to the free speed of the motor.
The second motor extends the time when the drive train is current limited
There is less resistive losses with two motors in parallel, and that makes a bigger difference at (very) low rpm
that more motors means cooler motors, and cooler motors are more efficient (you didn’t make this point, but others have)
My rebuttal is this: In a real world acceleration, those factors offer very small gains compared to the basic “torque per current” factor (Kt) that is doing the vast majority of the acceleration “work”, and is the same no matter how matter how many motors.
Another counterpoint: 4 extra motors & controllers will add ~4lb to a robot. All other things being equal, that’s a ~3% increase in robot mass, which could have a similar scale effect on low speed acceleration.
I don’t know if you trust the iLite simulator, but I’m assuming it has lookup tables that take into account all of the efficiency gains you list above. If so, @gerthworm’s post suggests the efficiency gains don’t have much (simulated) effect. If you think the sim is wrong, I would be happy to try to help create a better one.
I’m sorry but I’m not really sure what you mean by
What I understand from that statement is that you think because Kt is fixed, RPMRevolutions Per Minute should be the only variable affecting efficiency. If that’s true I think you’ve misappropriated current and power.
While it’s true Kt is going to be constant, that applies to the current through the windings, not to the supply current from the battery which is the constraint here.
If you halve the stator current target, with the same coil resistance, the amount of volts you have to output to achieve said current decreases, irregardless of how much back-emf there is. The motor controller is acting as a buck converter here. Instead of sending those volts to be dropped across a resistor into heat, they never get drawn.
In your argument, how would you explain the 30% efficiency delta shown in vex’s own dyno testing? That energy is going somewhere, and that somewhere isn’t heat. Saying Kt doesn’t change therefore rpm is the only large contributor to efficiency is disregarding input voltage, and how much you’re just burning on drop across a resistor.
The 6V graph sees the same efficiency curve, compressed by a factor of 2 on the x-axis. How is this possible if rpm is the only factor for efficiency?
I will concede that it will never be that large of a gap. A four motor drive train is still limited to maybe 50A per motor, so it is already much lower than the unrestricted 257A the falcon would like to draw. But the premise that dropping your current target shifts the efficiency peak towards the left still holds.
Sadly there aren’t many options for pinion sizing. SDS sells a 16T pinion for the falcon spline, but that’s been out of stock at least since February when I first started designing modules. The 14T pinion doesn’t really work for the motor spacing.
Your only real option is to use a 8mm/spline to hex adapter, with an 18T gear (smallest for hex).
High tide was running a 25T pinion with L1 iirc though, so the 18t L3 combo should still be a nice balance.
Thank you! Our team has been developing 3 motor swerve modules for about a week now and got stuck when it came to getting the right gearing. We’ve also ran into the problem of limited pinion gears, luckily we have access to machining custom pinions.
So my team is working on building these swerves, and I noticed that the positions of parts on the swerves are different an example of this would be the distance of the 2 by 1 mountain points on this is different from the MK4I. why was it designed like this is it stronger? why did you add a fourth mounting hole did you find that with the added motor it would break the swerves? Along with the fact that this changes makes you have to remake the bottom mounting plate, wouldn’t it be easer and faster just to keep that sizing the same?
Also could you explain how you when about cading this because there were some new tools I saw that I have never seen such as Move face property manager, and the one with the main plate that recalled a different file?
