Dog Shifting Gearbox Feedback
 

I've recently designed my first 2-speed shifting gearbox (technically 2nd, but the 1st one was abandoned half way through). I learned how to design these completely just by examining models made by other teams, so please point out any mistakes that I have made.



Theoretical robot speed with 2 gearboxes, 4 in wheels, and 120 lbs load: 6 fps & 15 fps

Weight (according to Solidworks): 10.8 lbs

Right now I have #35 12t sprockets on this for a west coast or similar style drive. Is a larger size recommended? If so, why?

Model files: https://grabcad.com/library/shiftbox-2-1

Re: Dog Shifting Gearbox Feedback
 

Looking at the 3d model closely, it looks like the lower mount hole on your CIMs is partially blocked by the large dog gear. Easy enough to solve by rotating the CIMs a bit so that these holes move outwards.

Additionally, it's hard to tell, but you may not have left enough room on either side of the sprocket for the actual chain to comfortably exist without rubbing on things. I'd double check this before producing it, it looks awfully tight to me, especially for #35 chain.

Not so much a design flaw as something to be aware of, but it looks like some of your standoff holes are going to be inline with your frame if this is mounted to a conventional 2x1 box chassis. Make sure to add clearance holes for these, or even better, use them to provide extra mounting support.

Re: Dog Shifting Gearbox Feedback
 

Yeah I found out about the blocking of the holes 1/2 of the way while designing. Correct me if i'm wrong, but dog gears are supposed to be able to pop put of their bearings manually and moved inwards right?

I also found that out with the first iteration of the design, so I had to edit the plates so that the standoffs go in between the loop of chain. They're not in the way anymore, but its fine if they rest in between the loop of chain right? Whatever the case, I agree with you, I should've left more room for the chain; I'll work on that.

And yeah I plan to cut out clearances on the chassis to make room for the bolt heads, or use them for mounting if required like you said.

Re: Dog Shifting Gearbox Feedback
 

It's fine to go around a standoff. I'm more concerned about how close your sprocket is to the front plate and to the gear.

Additionally, the dog gear's bearing bores should be pretty tight -- not something I'd want to depend on for maintenance. Otherwise, they'd drift inwards and fall off when not engaged by the dog.

Re: Dog Shifting Gearbox Feedback
 

No. The dog gears should be solidly attached and retained to the bearings they ride on. The dog clutch moves back and forth into the gears, and if the gears move you laterally you are going to have a bad time.

Re: Dog Shifting Gearbox Feedback
 

Ah ok. I guess I didn't compensate enough for the size of the chain.


I know that there should be a tight fit with the gears, but is it possible to pop them out by hand or are they usually grab on for dear life kind of tight?

Re: Dog Shifting Gearbox Feedback
 

You do not want the gears moving on the bearing. If that gear gets tilted it will now not be the correct center distance through the entire gear and won't run right. Coming from vex they are usually held in with retaining compound, and if you have to replace them it would be a good idea to use retaining compound to hold the new bearing in. We had to use a press to remove the bearings, and you really don't want to be able to do it by hand.

Re: Dog Shifting Gearbox Feedback
 

I would not use #35 chain for pretty much anything on a robot, it is heavy and way stronger than it has any business being on a drivetrain. Working load(not breaking strength) on #25 chain is 88 pounds vs. 199 pounds for #35. I'm not sure what the safety factor on ANSI chain is, but I know that the 6 CIM drivetrain we had on our 2014 robot used #25 chain(the heavier duty VEX stuff with thicker side plates) and required very little maintenance nor suffered adversely from stretching.

15 fps in high gear is slow, very very slow for a 6 CIM drivetrain considering you are likely to lose 10% or more mechanical efficiency, leaving you in the 13 fps range. Our 2014 bot was geared to a theoretical 22 fps and we measured our actual top speed at just over 18 fps(18% efficiency loss). We likely could have geared it a little slower and gotten our top speed a little faster, maybe 19 fps. So I know you can reasonably push the top gear 10%-25% higher, the only reason to keep it in the geared to 15 fps range would be if you need rapid bursts of acceleration to cover sub-20ft range movements on the field.

Re: Dog Shifting Gearbox Feedback
 

There are a couple of situations where #35 makes sense, usually on arms.

Re: Dog Shifting Gearbox Feedback
 

While 15fps would be slow for California, in general robots out east tend to gear much lower.

Also, I'd expect a 22FPS theoretical 6CIM drive to have massive brownout issues on the RoboRIO without some sort of current management in the code.

Re: Dog Shifting Gearbox Feedback
 

That would be one of the times I would agree that it makes sense.

Gearing for an actual speed of 15 fps after losses would be a better idea to take advantage of all that power I feel. Gearing for 15 fps theoretical with 6 CIMs just seems like a waste of 2 horsepower(1500 watts) to me, you can do it with 4 CIMs.

Re: Dog Shifting Gearbox Feedback
 

Honestly, I often see the tradeoff between #25 and #35 chain in a drive train application severely underplayed here on ChiefDelphi. The truth is there are pros and cons to each, and teams should evaluate which size chain they use.

It is not uncommon for a robot to weigh 150 lbs with a battery and bumpers and then have a CoF on their wheels of 1.2. That means you'd have 180 lbs of tractive force or 90 lbs per side of your drive train. If your sprockets have a pitch diameter of approximately 2 inches and you have 4 inch wheels, that's up to 180 lbs of force going through your chain! Nothing close to this will usually happen on a WCD in most cases because the center wheels will likely be direct driven and carry the majority of the tractive force, but this depends on your CoG.

#25 Chain Pros:
-Lighter Weight (~1.0 lbs/foot)
-Takes up less width

#25 Chain Cons:
-May require tensioning (higher part count)
-Co-planar sprockets are more critical (more precise machining)
-May come close to pushing limits (have to do the math)

#35 Chain Pros:
-Can probably get away without tensioning (lower part count)
-Sprockets less sensitive to axial position (less precise machining)
-High safety factor on chain working load

#35 Chain Cons:
-Heavier (~2.5 lbs/foot)
-Takes up more width

For the top teams, the extra 10-15 lbs you'd be saving by going with #25 chain is well worth it. These teams have no problem getting their sprockets co-planar sprockets and have experience doing the math to see if they're doing anything risky. On a WCD, the direct drive of the dropped center wheels off of the gearbox reduced the amount of the force that will be in the chain. For teams that may benefit from a lower part count, might have difficulty making co-planar sprockets, and deem the peace of mind of something that will "just work" to be more valuable than the weight savings, an argument can be made for using #35 chain.

I'd estimate that for about 75% of FRC teams, if they were to use chain they should at the very least think about these tradeoffs. What works best for the best teams in the world may work best for them as a direct result of their goals and resources, while it may not necessarily be the blanket best solution for all teams across FRC.

Re: Dog Shifting Gearbox Feedback
 

Ok, the reason I was wondering is because one of the gears is partially blocking some CIM mounting holes, so I thought one solution was to just pop out the gear and unblock access. Another thing to consider while designing next time.

My rookie year was 2015, so I have pretty much no knowledge on games with defense. What would be a desired speed for such games and why?

Also, without any weight the gearbox is geared for ~19 fps, and ~15 fps with 120 lbs of load, considering that there are two gearboxes on the robot. However, now that you have brought up efficiency loss, I will have to edit my spreadsheet to account for these things as well as voltage drop. Thanks for bringing it up; next time I will aim for a much higher fps.

Yeah, when I was experimenting with different gear ratios I found out that gearing so high resulted in almost 50%+ efficiency loss even with 6 CIMs on the drivetrain.

I think I may have miscommunicated. When I say theoretical, I mean the theoretical speed after robot weight, torque loss, etc. though I have left out things like efficiency loss and voltage drop.

With 4 CIMs, according to my spreadsheet, I get around 2 fps less speed on high gear. I know that 6 CIM drivetrains don't have as much improvement as lets say a 2 CIM to 4CIM, but I assume that acceleration and pushing force is a big component to games with defense. Again, according to my calculations, a 4 CIM drivetrain would lose almost 300 inch-lbs of tourque on low gear when compared to a 6 CIM with my setup.

Re: Dog Shifting Gearbox Feedback
 

It looks to me as though this problem can be fully solved by rotating the CIM slightly. Even if there's an alternate means of getting at this screw, it's better to address it the "correct" way at this stage.

A couple extra minutes of CAD work is a couple extra minutes you have to work. A couple extra minutes of between-match maintenance is a couple extra minutes you have to work, a couple extra headaches, and potentially a couple crucial matches where your robot is dead on the field.

Re: Dog Shifting Gearbox Feedback
 

When calculating, you must account for the largest diameter component acting on the gearbox, right? i.e. If you have a 6 wheel setup with 4in wheels and 2in diam sprockets, you only account torque loss with the wheels because they have the largest diameter, or am I missing something here?

We have little experience with chain, negative experience really, and have used belt drives since at least 2014 if my memory is correct, without any problems (15mm, not 9mm as they snapped during a match in 2014). Are there any real benefits to using chain on a WCD opposed to using belts instead? The only thing compelling me towards chain is increased flexibility in design, less space consumption, and ease of replacement.

In terms of weight, I would actually be compelled to keeping #35 chain as even though it will weigh more, the CoG benefits would outweigh the weight benefits. After all, the drivetrain is the last subsystem you want to fail, so, from my PoV, #35 chain would be a good precaution to take.

Re: Dog Shifting Gearbox Feedback
 

Good point, I actually did solve this problem by rotating both CIMs inwards by 10 degrees, though it did take another 3 mins redoing the pockets.

Re: Dog Shifting Gearbox Feedback
 

These chain weights are off. #25 chain is about .1 lbs/ft and #35 is around 2-3x that.

Re: Dog Shifting Gearbox Feedback
 

You're right, thank you Adam. I meant 1.0 lbs/10 feet for #25 chain and 2.5 lbs/10 feet for #35 chain. That makes the difference in weight on the whole robot only about ~1.5 lbs.

I'm not sure what you mean here. I didn't talk about "torque loss" at all in my previous post.

If you want to know the max amount of tension possible in your chain is, a quick and dirty way of getting there is:

Take your robot's weight. Multiply that by your wheel's CoF on the playing surface. Now multiply that by the ratio of your wheel diameter to your sprocket pitch diameter, making sure you use the same units. Divide by 2 because each half of a tank drive is assumed to support half the weight of the robot.

Tension_max = Mass * CoF * D_wheel / D_sprocket / 2

This will tell you the necessary tension in your chain to make one of your wheels slip when it is supporting half of your robot's weight. Now in reality, one wheel will not be supporting half of the weight, it will be distributed among multiple wheels on that side of the robot. And the wheel that most commonly supports the majority of the weight of the robot will be the one directly driven off of a gearbox on a WCD (your dropped center wheel). And since you have a dropped center direct-drive wheel in a WCD, it is not necessarily a crippling issue to toss a chain.

Re: Dog Shifting Gearbox Feedback
 

I know 1114 uses #35 chain on their tankdrives because they run exact center to center and avoid tensioners. They also try to match their drive wheel diameters to the pitch diameters of the wheels (not really possible on most WCDs). #35 is definitely overkill but overkill is nice to have in a drivetrain. (low maintenance too)

Tangentially, I want some enterprising company to makes usable sprockets for bike chain. Bike chain is stronger and lighter per length than #35, more resistant to misalignment, more readily available and has readily available and awesome chain tools for relatively cheap. Also comes in cool colours.

Re: Dog Shifting Gearbox Feedback
 

Shashank, if we were to make this in the near future, you'd have to simplify the part, there is no way that this can be made on the bridge port, this would have to laser cut and we usually can't cash in on our sponsorship benefits from Superb until well into build season. Even if it will weigh a half a pound more, that's fine, just make an iteration that is feasible to make parts for on the mill if we are going to need it in the near future.

Re: Dog Shifting Gearbox Feedback
 

I've been looking into better optimizing parts for machine time over this off-season, especially gearbox plates like these which sucked up a ton of waterjet time for us. One thing I'm exploring is using unpocketed 1/8" plates with additional standoffs, after looking at 2826's robot.

Re: Dog Shifting Gearbox Feedback
 

That is totally viable, especially on anything that doesn't exit the frame perimeter.

Re: Dog Shifting Gearbox Feedback
 

Sorry, I think I got mixed up on who I was replying to haha. But my question was do you have to account for the size of the sprocket when calculating torque loss? I would guess not as the majority of force is applied on the wheel.

Re: Dog Shifting Gearbox Feedback
 

Ben, did you forget that we are getting a CNC mill?

PCNC 1100 Education Package: http://www.tormach.com/store/index.php?app=ecom&ns=catshow&ref=PCNC1100_Package s_Ed1

I'm pretty sure that 1114 has used 1/8 inch plates with bends for many of their drive boxes. I wouldn't think it would be a problem.

Re: Dog Shifting Gearbox Feedback
 

Yeah, I found plenty of examples of bent 1/8" gearbox plates but we don't have access to a decent brake.

Re: Dog Shifting Gearbox Feedback
 

If you do some searching on harbor freight, you will find many hand brakes for under $80. They would do just fine for simple bends.

It looks like there's a shop right in your city: 1800 Fort Harrison Rd # 21, Terre Haute, IN 47804

EDIT: My bad, it seems that the only brakes online at harbor freight that can bend 1/8" aluminum cost over $200. I'm sure that if you guys use a strong grade of aluminum with frequent standoffs, you should be just fine.

Re: Dog Shifting Gearbox Feedback
 

There is a strength gain just by using 6061 alloy over the 5052 alloy, which is usually used in sheet metal fabrication for its ease of bending.

Re: Dog Shifting Gearbox Feedback
 

^This

On second thought, It would probably be much better to just use 6061-T6 plates in place of bent 5052 as it provides greater rigidity in comparison as Michael stated. Just don't use any pockets.

Re: Dog Shifting Gearbox Feedback
 

To elaborate more, there is a strength advantage but not a stiffness advantage. So there are many applications where people might think their plate is too "weak" where changing alloy won't help at all, as what they really want is stiffness.

Re: Dog Shifting Gearbox Feedback
 

We have run unlightened 090 5052-H32 gearbox plates without any bends for a couple of years. The only issue we have had is that the VEX hex bearings (or at least the old ones) don't press well in thin plates due to the excessive undercut. We fixed that this year by using thunderhex. That said, our opinion is that the gearboxes are louder when made of thin plate. We are looking to go to .1875 plates with lightening this year due to getting some additional machining resources, we will see how those end up.

Re: Dog Shifting Gearbox Feedback
 

Wow, I didn't know this. I always thought that 6061-T6 is stiffer than 5052-H32 as it makes logical sense: 5052 bends and 6061 doesnt.

Modulus of elasticity for 6061-T6: 10000 ksi

And for 5052-H32: 10200 ksi

As Adam said, 5052 is actually stiffer than 6061-T6. You would also need a laser cutter to make the plates with 5052 as it cant effectively be machined with a mill.

Re: Dog Shifting Gearbox Feedback
 

I wouldn't say "can't." It doesn't machine as nicely as 6061, but it can certaintly be done effectively with care and the correct tools. We regulary perform secondary milling operations on 5052 bent parts with our mill.

Either way, unlightened, unbent plates as thin as .090 are definitely workable for gearboxes.

Re: Dog Shifting Gearbox Feedback
 

I was making the point that they are essentially the same stiffness.

Re: Dog Shifting Gearbox Feedback
 

As low as .090? Would bearings stay put with that thin of a material?

Re: Dog Shifting Gearbox Feedback
 

We use positive bearing retention on all of our laser cut gearboxes rather than relying on press fits.

Re: Dog Shifting Gearbox Feedback
 

Are you saying your relying on the geometry to keep your bearings in? Such as the shoulder of the shaft + the flange of the bearing?

Re: Dog Shifting Gearbox Feedback
 

No, should have clarified. Retention is almost always through rivets+washers clamping down on the bearing flange. We'll occasionally do it through other parts of the system, but never something as likely to be seperated from the gearbox during maintenance as a shaft.

Re: Dog Shifting Gearbox Feedback
 

On our robot two years ago, a ball shifting setup with two full-size CIM's to a 6 wheel drop center had tons of power, enough to spin four-inch taction wheels. I think the bottleneck is not the amount of motor power or gearbox torque after two CIM's but the traction and number of wheels contacting the ground.

Also think about how much power you consume every match. Its a lot faster to chage a 30% depleted battery than a 50 or 60% depleted battery. If you have the resources to charge enough batteries fast enough than you'd be fine. Just points to consider.