The huge shaft length underneath the top plate but above the module was bothering me, so I added a rudimentary PTO.
As shown with all hardware, this weighs 8.51lbs. A full chassis with this swerve pod weighs 39.82lbs with all hardware (minus 12 aluminum rivetscand ~2lbs of chain).
Ratios are 3.23:1 and 6.45:1 for high and low gear respectively, resulting in free speeds of 19.7fps and 9.8fps with a 2.75" wheel. The PTO powers a 26 tooth #35 sprocket when in low gear. This means that the efficiency is slightly lowered in low gear due to the chain of the PTO moving around. In order to engage the PTO fully, there needs to be a shifting gearbox elsewhere in the chassis to engage whatever device it is meant to power. Again, there will be losses because the low gear is active as well.
Alternatively, this can be used with a three-position cylinder, but I chickened out to cull some weight. The tradeoff here is that weight will need to be added elsewhere for a shifter. It really depends on the game; for lifting endgames like 2013, this would work because the wheels aren’t touching the ground when the PTO is engaged. Plus, due to the characteristics of a swerve drive (and the sheer power of this chassis), low gear would only be used in counter-defense or offensive scenarios, and you would be moving slowly.
A Bimba 3-position cylinder would be used if needed.
Based heavily of off Bryce2471’s work. The only difference is that the encoders are in different places, and this shifting shaft has to be fully custom. At this point I’m not even worrying about manuacturability. I’ll get the CAD out as soon as I fix up the shifting gears a little bit (adding in ball grooves, etc) and probably add in mounting options for a 3-position cylinder.
I’m really going to have to make a manually machinable version of the single-speed one.
EDIT: I forgot to mention: I can’t figure out where to put the dive encoder without adding more plates and spacers. I was thinking of using a hall effect sensor with magnets embedded into the large pulley.
Why did you choose those gear ratios? Most swerve is ~ 10fps and your low gear is 12fps. I haven’t done the math, but based on our experience, you’ll have problems accelerating in high gear. A 5-8 fps low and 10-15 fps high would be better IMO. Also, what’s the sprocket at the top of the wheel module for?
This is a 4-cim, 4-minicim drive.
This year, we ran a free speed of around 17fps with 4 cims only and had no problems accelerating at all. 1678’s 6-cim drive was geared for around 22fps. This is an even more powerful drive, so I bumped it up to 24fps. The math works out; for anything more than a couple feet it is easy to use high gear (with < 0.05s of lag behind low gear).
The sprocket is the output of the PTO. It is not mounted to the module, but shifts on the shifting shaft.
Have you done the calculations to estimate current draw on this? 1678 had to take extra steps to reduce the chance of popping their main breaker. Same goes with 254 and many other teams. The only reason 6-CIM drivetrains are even possible is because we know the breaker can hold more than 40A for longer than the spec sheet says.
You’re adding 230W of power AND reducing torque. It’s not a matter of acceleration, it’s a matter of keeping your robot powered while you accelerate.
Looks sweet, but I am concerned about the turning gears, they’re so close to the ground that if anything bounced up, it could potentially break or chip a gear tooth.
I don’t think the math will work out on those speeds when you take into account the voltage drop that happens when you run all twelve motors on at the same time.
The current drop from the turning motors is a significant factor here, I think.
I would be very surprised if a swerve like this didn’t blow the main breaker in a pushing match, even in low gear. IIRC FRC 2451 had current-related problems pushing an immovable object (read = our robot) last year with just 4 CIMS at a lower speed.
This will blow the breaker in a pushing match due to the massive power behind it. However, as with any high-geared 6+ cim drive, there must be good autoshifting code to prevent disasters. IIRC 1678 had a main breaker blow in the finals at SVR due to their autoshifting code having problems.
With swerve drives, if we get into a pushing match, we have three
options:
Translate away. Good for offensive tactics, especially with this blistering speed.
Push back. Useful only against lightweight robots or if we are stuck somehow. Or to slow other’s robots down.
Cross the wheels. If the wheels are in an X-formation, then we can’t be pushed unless they lift us (cue opponent’s breaker blowing). Esentially, the driver has a trigger to lock the wheels, and when the opponent tries to go another direction our driver releases the trigger and moves. This is an extremely effective defense (in theory) as long as our driver can move to block them when they try to get past. I’m confident our speed is high enough for this.
The purpose of low gear is to allow for precise positioning and very short drives. That’s why it’s geared so high here.
The 40a breakers are on each motor. The main breaker is 120a, which is usually the problem.
Thank you for the input. I hadn’t considered that. However, this only supports slightly more power than 6 cims, so it might not be that bad.
There is around 0.25" clearance to the ground. Not much we can do about that, but at least they’re vex gears so they’re easy to replace. Many swerve drives use gears so close to the ground, and I haven’t heard of so many problems.
EDIT: One of our mentors mentioned that a bigger problem would be the current spike killing us on startup. Some things we have to solve with software; this kind of drive won’t be used for at least a year (2016 game) so we have plenty of time to do tests.
In order to move the robot, the pushing robot would be acting against the traction wheels. Any direction that they are pushing from they would need to scrub the tires.
You could turn the wheels 90* from their push too.
D’oh. The low gear is actually 9.8fps in the photo. I accidentally marked it down as 36 tooth instead of 44 tooth in the calculator.
Then the robot is definitely traction limited. It draws 240 amps at maximum (lifting) torque. I was worried about increasing the pulley to get the free speed under 12fps, but it turns out that won’t be an issue.
In order to move the robot, the pushing robot would be acting against the traction wheels. Any direction that they are pushing from they would need to scrub the tires.
You could turn the wheels 90* from their push too.[/quote]
Assuming your contact patch is square an X with 45° wheels is the most effective (I know that a 1" wide, 4" dia wheel with Blue Nitrile is approximately square) because you are applying force along the longest line you can draw in that shape. If you measure your contact patch and find it is a rectangle you can calculate the most effective angle to put your wheels at when you make this maneuver. (Not sure how much difference that would make tho…)
It would increase contact area, but I want to be able to lock the wheels without having to spend power on the motors. I guess putting them in brake mode would do the trick.
Since the contact area and amount of force would be the same regardless of the wheel angle, there might not be any difference. This is not taking the tread pattern into consideration though. Which is why the X formation might be beneficial. I think the X pattern will both reduce the wheel’s ability to rotate when pushed and utilize the effects of the tread “fingers”.
The contact area would remain the same.
Nathan, I know you said that 365 did testing on diagonal vs. normal tread. Did you guys happen to test the tread sliding sideways as well?
Oh wait, you’re right. I was thinking of another tread with diagonal tread in it…
Regardless, it shouldn’t matter. An x-configuration would allow the motors to just sit in brake mode I think.
Pushed sideways while in an X? That woule be very hard to do if you set the turning motors in brake as well; they are geared down 1:26 with a secondary of 30:84.
Sorry, we did the tests last year on our 8 wheel WCD. With a swerve that has straight cut tread it should be just as hard to do as being pushed backwards. (I will have to think about how best to “break” with diagonal tread)
Thank goodness. I was thinking, “Oh god, did my numbers come out totally wrong!?”
They did, as a matter of fact. This is only geared for 20fps, sadly. I messed up the numbers in the calculator (again). Fortunately, the speed is easy to increase, but hard to decrease.
EDIT: I added a 3-position cylinder to the CAD, but for some reason these cylinders from Bimba are around 6" tall. So you can technically put on one here, but I would probably learn to live with this version.
I’ll get the CAD out as soon as I fix up the shifting gears a little bit (adding in ball grooves, etc) and probably add in mounting options for a 3-position cylinder.