Looks great. Do you think you could eliminate a lot of this play with a hex hub? You could try to broach the bronze gear to the same shape.
well the play isn’t on the bronze gear, that mating is actually pretty good. it’s specifically between the andymark hub, keystock, and axle. Also we don’t have the tools for hex shafts and it would be a major pain to deal with the bearings considering how it has to be assembled. but even then there would be some inherent play anyway. One option would be to setscrew the key in place. or get the 3/8 hubs and then bore and broach it ourselves for a tighter fit. The axle was sanded down though so maybe whoever did that went too far. may not be worth messing with though
Are you able to turn the wheel by hand on the new version?
not a chance in hell. this drive is 1 way only(cim to wheels). Trying to go the other way will be met with extreme resistance and would cause something to break if you actually could turn the wheel. But in our case this is actually a desired trait. It allows us perfect control of motor speeds at all times and provides extreme braking when someone else is trying to push is around when we don’t want them to(providing the wheels grant enough traction)
Nice work aligning the worm and minimizing lash in the gearmesh. :]
Have you measured the free current?
it’s about 3.5 amps
This looks great guys. I think that this version will help you in many of the pushing battles that you all faced last year. Your omni-directional drivetrain was very fun advantage to pair up with last year during the elimination rounds and I wish we would have provided you all with some more firepower. Thanks again for selecting our team last year and I hope that we get to work together in Saint Louis again in 2014.
I don’t follow this logic.
Well just think about it. Any wheel drive where you can move the motor by turning the wheel you will have drifting wheel speeds depending on what else is going on. for example another robot is trying to push you around or you want to slow down but the inertia carries your robot further. With this setup where the wheel can’t move the motor. If we tell the motor to go at 100rpm, then it doesn’t matter what else is going on that motor is going to draw up to 100A or apply extreme braking to make that motor turn at 100 rpm, not any slower and not any faster. That means if u want the robot to stop, it’s going to come to a complete dead stop instantaneously. Also if we meet some resistance(another robot) then the PID loop will keep increasing force to the absolute limits to try to make that motor turn at the set speed. Also if another robot tries to push us around, even if we don’t have enough power to overpower them, simply not trying to do anything will provide tons of force and keep them from pushing us around, well as much resistance as the friction of the omni wheels provide. With this design we don’t need the force to over power other robots, we just need enough traction and that will prevent anyone from pushing us around.
I think what he means to say is that it gives them near perfect wheel control. As soon as the motors are stopped so do the wheels while requiring little to no active braking from the motors. Would make for a very jerky robot without proper PID control though.
not just a full stop, it also prevents any unwanted changes in speed. so if something is trying to accelerate or decelerate us that force is going to be put into the chasis instead of the motor. So as I stated if we want to go a set speed, any changes in speed will be met with extreme resistance other than what we tell it to be. Also I tested a PID loop earlier today that seems to work extremely well, so much so that the only issue was that when loaded down too much it would draw 80-100A and shut the jaguars down. I have a video of my PID loop with the old transmission, and the only difference a load makes is higher current draw, the tracking is just as good.
Regardless of gearing method, the motor will shift up and down it’s curve as the applied load changes (being pushed is an applied load).
If your statement were true, the gear train would magically be creating and dissipating energy.
indeed it will. I never said it wouldn’t however with this system is such that it makes the job much easier on the cim. For example if we were moving forward at a certain speed. A robot behind us is attempting to push us forward faster. It’s not going to happen. All him trying to do so would REDUCE the load on our drive motors and the extra force would be directed into our chasis. If we wanted to remain stationary, and someone tries to push us, provided we have enough traction our robot will move absolutely nowhere regardless of how hard they push. or we can advance forward at the rate we want to advance at and not any faster than we want, due to this type of setup. Now granted if we faced head on against another robot, if we don’t have more force than they do we won’t be able to push them back, but at the same time they wouldn’t be able to push us back either and all the cims simply have to do is not move and all the force is transferred into the transmission structure
Neat idea. However, I am fairly certain that your worm wheel will loose teeth when you have the weight of the robot behind it. If your robot is going fast, and suddenly you stop applying power, the worm wheel won’t be to turn the worm gear, so you’ll just snap a tooth off. It will be different than just testing it on a table.
In the video you talked about having to use a smaller worm wheel in order to get the same reduction. These are pretty fragile, so you could switch to a bigger worm wheel with a two-start worm.
A few other recommendations. Be sure to lubricate your gears! Make sure there isn’t any play sliding the worm or worm wheel up and down on their shafts and that the unsupported drive axle doesn’t wiggle around, but a little play in rotating the gears (like you showed in the video) is really important. That backlash makes the transmission operate more smoothly, and helps with lubrication, especially when you’re using gear like this, than “slide” instead of “roll” on each other.
oh indeed we are aware bench tests are different. Hence why we are prototyping. The sudden stop may or may not be an issue, we don’t know. If it is, we can address it by having programmed coast made to prevent such an instance or at least minimize the sudden load. Hell at full speed our robot may topple itself over even! For robot impacts we have bumpers so with those two things in mind I think we will be fine. But we will be testing for it in the upcoming weeks.
Not sure what you’re talking about using smaller worm. We didn’t need to do such a thing at all, we had to use smaller wheels or we wouldn’t have enough torque and we would load the cims too much so we had to use a smaller wheel. Also these worms are already 2 start.
I was thinking aboud greasing them, it would help reduce the contact friction but would also make a mess =\ And yup there’s absolutely zero vertical movement on the worm. Everything was spaced down to 0.001 of an inch. Being wedged between a thrust bearing and roller bearing on the other side it’s got absolutely no where to go(unless it destroys the roller bearing, but that can be addressed if it happens)
Depending on the exact specification of the worm wheel and worm this could be a very likely scenario. Like Magnets said, you could switch to a larger worm wheel and a two/four start worm to lessen the tooth loading, and it’ll have the added benefit of being more likely to back drive. Under extreme loading rather than locking up.
Your wheel choice is also going to play a huge role in the overall setup robustness, something on the lower end of the scale should start slipping before the transmission fails, where a high traction wheel might not slip until after the gearbox fails, its hard to say exactly without doing some calculations.
Magnets also brought up the subject of shock loading the worm wheel and worm during deceleration which is a very, very likely scenario. A decent bit of braking code could help to lessen the chance of hard stops during most matches, but the right hit or two could ruin your day. If you really wanted to stick with the worm setup and have it lock, a clutch setup could do some really cool stuff, or even some sort of flex plate/flexible coupling in the system to ease the load on the gear.
well the one we are using now is already a 2 start worm. We could go bigger but would probably try to find ones made out of harder material before we go bigger in size. The locking up thing is desired though, if it can survive that is.
Well we are using the vex omni, so i’m sure those will start dragging across the carpet before anything extremely drastic happens.
We are aware of the potential for failure, at least this system will still work with just 3/4 wheels(even 2/4 if opposite corners go out) and are extremely easily replaced, a single bolt, slide axle out, swap gear, and ur back on in less than a minutes worth of time. a clutch setup could help but then you run into complexity and reliability issues. We’ll see how far this will take us and if it can survive what we throw at it then we won’t be worried.
You really, really need to lubricate this. You’ll have little specks of brass dust all over the place if you don’t. I agree with the rest of the post though. It’s only a prototype, so you don’t really know what will happen. It could be that the transmission with backdrive before you brake anything, or that the wheel (omni wheels don’t have much grip) will loose traction with the ground before something goes wrong.
Overall, I really like how small this is. It’s perfect to fit in the corner of a robot.
I don’t know. so far from the abuse I put it through it’s not grinding, it’s actually putting a mirror polish finish on the bronze gear so unless grit/dirt gets inside them I don’t think there will be any bronze specks lol. But as stated yes it would be better but will make an absolute mess x.x
are some pictures of a complete drive chasis. So far everything u see in that picture only weighs about 27lbs and that’s actual weighed items.