pic: 20's IRI Carnage

[R.C.]

Originally Posted by JChang

I would like to know how 254/973/1323 connect the dog to the moving shaft inside the output axle (roll pin, bolt or something else).


Some suggestions for improvement:

• Make the OD of the dog larger, there is plenty of room before interfering with the 45t gear.
• Use roll pin or solid body 4-40, the current bolt will snap when inexperienced programmers attempt to write an auto-shift code.
• Thicken the thin part of the dog if there is room

-Josh

Josh,

We use a bolt and shift at a lower PSI than most teams. We were able to shift at about 30 PSI with the 9/16" cylinder we use to use. We shift at about 20-30 with the current 3/4" Pancake Cylinder. A decent amount of teams shift at 60 PSI.

We are going to be using the solid body screw similar to AndyMark (Waiting for them to come in). We also used to just switch the bolt out at the beginning of each competition. As you can get a box of 100 screws for $4-5. Now we go pretty much go 5+ competitions and demo's without switching.

-RC

[DampRobot]

Originally Posted by Cory

There are good reasons to keep the dog as small as possible. It opens up the ability to use a lot of other custom reductions that would otherwise be interfered with. We had to actually turn ours down a touch to get enough clearance for the ratio we ran.

We used the new Andymark shoulder bolts for our dogs.

I agree, the OD of the dog can be quite limiting to designs. It would be nice to be able to use a 50t gear instead of the 45t the DS has.

I wouldn't worry too much about this type of failure. This dog saw far more torque than it ever usually would. This is the only failure of a dog I've ever seen, usually its just a broken roll pin or something.

After more fully understanding the issue, I'm surprised the debate has revolved around the dog, when it was the wheel that failed when it shouldn't have.

[AdamHeard]

Originally Posted by DampRobot

I wouldn't worry too much about this type of failure. This dog saw far more torque than it ever usually would. This is the only failure of a dog I've ever seen, usually its just a broken roll pin or something.

I'm not understanding this point, how did it see more torque than it usually would?

In high gear, a full speed direction change will result in 2x stall torque being applied.

[AdamHeard]

Originally Posted by JChang

I would like to know how 254/973/1323 connect the dog to the moving shaft inside the output axle (roll pin, bolt or something else).
[/list]
-Josh

We don't use 3/4" bore cylinders at 60 psi, but historically have used 9/16" bore at 40-60 psi.

We used 7/16" in 2012, but that was a single CIM per setup and had no initial reduction.

We use a standard #4-40, have never broken one.

We limit the stroked so that fully extended or fully retracted the dog is not loaded into the face. Of course when it's shifting if the dog gear hits the other gear tooth to tooth it will experience load while one rotates, but other than that we never actually load the gears.

I think the combination of shimming and lower force is what has caused us to never break a screw.

[KrazyCarl92]

Originally Posted by AdamHeard

I'm not understanding this point, how did it see more torque than it usually would?

In high gear, a full speed direction change will result in 2x stall torque being applied.

That is in high gear. So 2 x 2 x 45 x 45 / 11 / 30 = 24.5 times stall torque. Also remember voltage drop and it's far more like 16 times CIM stall torque at the dog.

We were in low gear when the wheel broke and stayed there for a while.
2 x 2 x 45 x 60 / 11 / 15 = 65.5 times stall torque of a CIM seen at the dog. More like 44 times CIM stall torque when you account for voltage drop. Normally the wheels would have slipped long before that point was reached, but one was broken, so the torque was unable to turn the wheels. Don't know to what extent this is true but there was certainly more than the normal traction limited amount of torque being demanded of the dog.

[AdamHeard]

Originally Posted by KrazyCarl92

That is in high gear. So 2 x 2 x 45 x 45 / 11 / 30 = 24.5 times stall torque. Also remember voltage drop and it's far more like 16 times CIM stall torque at the dog.

We were in low gear when the wheel broke and stayed there for a while.
2 x 2 x 45 x 60 / 11 / 15 = 65.5 times stall torque of a CIM seen at the dog. More like 44 times CIM stall torque when you account for voltage drop. Normally the wheels would have slipped long before that point was reached, but one was broken, so the torque was unable to turn the wheels. Don't know to what extent this is true but there was certainly more than the normal traction limited amount of torque being demanded of the dog.

I didn't mean 2 CIM units of stall torque, but twice the torque of what stall would cause at that stage.

I did however forget the ratio difference between high and low, thanks for correcting me there.

[Aren Siekmeier]

Originally Posted by DampRobot

I wouldn't worry too much about this type of failure. This dog saw far more torque than it ever usually would. This is the only failure of a dog I've ever seen, usually its just a broken roll pin or something.

After more fully understanding the issue, I'm surprised the debate has revolved around the dog, when it was the wheel that failed when it shouldn't have.

I agree, it seems apparent to me (without actually quantifying the shock loads of normal driving vs. the loads of lifting your whole robot) that this was due to the broken wheel, and the fact that the gearbox essentially had to lift a significant portion of the robots on a 3" lever every time the wheel went around.

[apples000]

Originally Posted by apalrd

-It would be interesting to try Ti for this part.

We've done this before with a PTO/brake setup where the wheels would lock when the PTO was engaged and the momentum of the whole robot could rip apart our hardened steel dog.
Our Ti dog didn't break and we could make it even smaller, but it was expensive and a MASSIVE pain to manufacture, so we had a machine shop (that a mentor worked at) make our second set for us. Unless you have a real need, and a ton of machining resources, I would advise against this.

[Aren Siekmeier]

Originally Posted by apples000

... and a ton of machining resources ...

Ha, like if you're that team working our of Oakridge National Lab that printed a good chunk of their robot out of titanium this year

[apalrd]

Originally Posted by apples000

We've done this before with a PTO/brake setup where the wheels would lock when the PTO was engaged and the momentum of the whole robot could rip apart our hardened steel dog.
Our Ti dog didn't break and we could make it even smaller, but it was expensive and a MASSIVE pain to manufacture, so we had a machine shop (that a mentor worked at) make our second set for us. Unless you have a real need, and a ton of machining resources, I would advise against this.

The guy I asked happens to be a fan of Ti, which is why he mentioned it (He later suggested hardened steel would be perfectly adequate, in a different email).

Titanium alloys have a low modulus elasticity but high yield strength/ultimate tensile strength, and fairly low density. So it's springy but won't break easily, and it's lighter than steel for the same strength, which is perfect for this application. But machining it is more difficult than steel or aluminum, and welding it is also hard (not that this piece requires welding, but it's hard).

I like doing this kind of failure/material analysis. It's fun and I always learn a lot.

[KrazyCarl92]

One odd thing that we did with our implementation of the shifters was putting one of our talons in coast mode and one in brake mode on each side of the drive train. Our drivers did not like that the robot was coasting so much when we lined up to shoot from the corner and it significantly increased our time to line up for shots. We tried putting all of our talons in brake mode on the drive train and the motion was far too jerky to be desirable. We then decided to try one motor in brake mode and one in coast mode and we found it to be a great sweet spot. The motion was not too jerky and we were capable of lining up for shots much more quickly.

Could this have contributed to the dog failure? Intuitively it would not seem to make that much of a difference, but I do not know how many teams have tried running a configuration similar to this.

[MichaelBick]

Originally Posted by KrazyCarl92

One odd thing that we did with our implementation of the shifters was putting one of our talons in coast mode and one in brake mode on each side of the drive train. Our drivers did not like that the robot was coasting so much when we lined up to shoot from the corner and it significantly increased our time to line up for shots. We tried putting all of our talons in brake mode on the drive train and the motion was far too jerky to be desirable. We then decided to try one motor in brake mode and one in coast mode and we found it to be a great sweet spot. The motion was not too jerky and we were capable of lining up for shots much more quickly.

Could this have contributed to the dog failure? Intuitively it would not seem to make that much of a difference, but I do not know how many teams have tried running a configuration similar to this.

This seems like it could have contributed to the failure. What you may want to try in the future is to have the drivers pull back on the joysticks a little when they want the robot to stop. This will create a situation electrically very similar to the break mode on the speed controllers, without the extra jerkyness

[AdamHeard]

Originally Posted by MICHAELABICK

This seems like it could have contributed to the failure. What you may want to try in the future is to have the drivers pull back on the joysticks a little when they want the robot to stop. This will create a situation electrically very similar to the break mode on the speed controllers, without the extra jerkyness

This is some of the fancy control that "Cheesy Drive" implements.

[apalrd]

In any case (Talons/Victors/etc. in Coast, in Brake, one in Brake, driver pulls back on stick) the drive line will transmit torque in the opposite direction (from motors>wheels to wheels>motors or to -motors>-wheels) and the gearing will transition through all of the lash, then hit the other load surface (a shock load on the dog and all of the gears/sprockets). The motor will spin more freely when in Coast, but the lash transition will still happen. The force of this is relatively low in any brake-mode case, as the motor provides zero brake force at zero speed (it acts as a generator to power itself, so the torque it reacts is proportional to the speed it is forced at, when in brake mode, not including electrical/efficiency losses). The torque reaction from the motor with reverse power braking will be related to the applied voltage and drive speed/

I've always liked driving full coast, and learning to coast/slow down properly into what I'm going for (or just hit it off-throttle if it's solid). The Cheesy and Culver drives were both fantastic improvements on the two-stick skid steer in that there is guaranteed to be no twist from asymetrical motions between the two sticks, since the throttle and steering are separated.

Edit: The more I think about cases like this, the dog is under a TON of repeated shock loading like this.

[KrazyCarl92]

Originally Posted by apalrd

Edit: The more I think about cases like this, the dog is under a TON of repeated shock loading like this.

So are you saying that the brake mode may cause repeated shock loading which could contribute to the failure? Or that code that does this for you may have the same effect?