Sonic Shifter - recent feedback?

[aldaeron]

Originally Posted by BoilerMentor

Gear ratios were 18.75:1 and 7.08:1. Gearboxes were lubricated with moly grease containing Teflon. The gearbox direct drove a 6" x 2" wheel at the center of the robot which was chained to the other two wheel in the drive train side, also 6" x 2" wheels. All wheels were treaded with blue nitrile. Center direct drive shaft was supported with a bearing in the outer plate. Two other axles were dead shafts with bearings in the wheel.

I want to point out for folks who don't spend a lot of time looking at gearboxes that this is a very aggressive gearing with one of the highest CoF wheel tread materials. My quick calcs say this generates ~747 pounds of force at the wheel patch with 6 CIMs, well over the traction limit of 185 for a maximum weight robot in 2014. Meaning it is an extreme case and may not be the best one to guide the decision of an average team.

Overall this is great feedback. I have been wondering how the 3rd stage on the 3 CIM works at high loads. I always thought the sheet metal with standoffs 3rd stage looked a bit rickety. I am curious if you ever subbed out the 7075 aluminum gears for 4140 steel gears in the third stage of the 3-CIM shifter? Sounds like it was a misalignment issue more than a material strength issue. Did you ever try stacking on a second Vex third-stage plate to add stiffness to the third stage bearing? This would pickup more of the bearing race and possibly prevent angular deflection.

I am assuming the output of the gearbox was direct driving your center wheel in a tank drive. How do you think a 2 stage 3-CIM shifter would work if offset from the wheel axles and with a #35 chain reduction between the gearbox and wheel axles in lieu of the Vex 3rd stage?

Originally Posted by BoilerMentor

The high load situation I believe created the problem was encountered when the driver returned the control stick to a neutral position with the robot at a high speed. The auto shifting code would have immediately tried to shift the robot to low gear with all three cim motors braking.

Was there a particular reason for adding this "stop on a dime" feature? Did it work as you had hoped? Would you recommend it? It does seem like a recipe for gear teeth shearing.

Originally Posted by BoilerMentor

I've spent a fair amount of time "behind the glass" as it were. I'd challenge you to find a driver who pushed their robot harder and drove more aggressively.

Based on photos and gearing - I agree - you win!

-matto-

[cbale2000]

Originally Posted by aldaeron

I want to point out for folks who don't spend a lot of time looking at gearboxes that this is a very aggressive gearing with one of the highest CoF wheel tread materials. My quick calcs say this generates ~747 pounds of force at the wheel patch with 6 CIMs, well over the traction limit of 185 for a maximum weight robot in 2014. Meaning it is an extreme case and may not be the best one to guide the decision of an average team.

I second this.

Though that said, we used the same gearbox in 2014 with a 26.04:1 low gear and a 7.08:1 high gear on 4" wheels with the same tread.

Originally Posted by aldaeron

Overall this is great feedback. I have been wondering how the 3rd stage on the 3 CIM works at high loads. I always thought the sheet metal with standoffs 3rd stage looked a bit rickety. I am curious if you ever subbed out the 7075 aluminum gears for 4140 steel gears in the third stage of the 3-CIM shifter? Sounds like it was a misalignment issue more than a material strength issue. Did you ever try stacking on a second Vex third-stage plate to add stiffness to the third stage bearing? This would pickup more of the bearing race and possibly prevent angular deflection.

I don't think the issue is the plate stiffness, but the fact that the small output gear (the one with the missing teeth in the photo on the last page) is on a cantilevered shaft. If one was to modify the plate so that another bearing could be used on this shaft, there likely would not be any problem.

On a related note, our team actually did this recently with a pair of VexPro 2 CIM Ball Shifters that we had inadvertently ordered without the 3rd stage (and apparently the output shafts on the 2nd stage are longer when order it like this). So we made a pair of replacement plates so that the smaller 3rd stage gear could be retained by an additional bearing instead of having to remove the shaft and lathe a snap ring channel into it.

[aldaeron]

Originally Posted by cbale2000

I don't think the issue is the plate stiffness, but the fact that the small output gear (the one with the missing teeth in the photo on the last page) is on a cantilevered shaft. If one was to modify the plate so that another bearing could be used on this shaft, there likely would not be any problem.

On a related note, our team actually did this recently with a pair of VexPro 2 CIM Ball Shifters that we had inadvertently ordered without the 3rd stage (and apparently the output shafts on the 2nd stage are longer when order it like this). So we made a pair of replacement plates so that the smaller 3rd stage gear could be retained by an additional bearing instead of having to remove the shaft and lathe a snap ring channel into it.

Wow I had never noticed that the output gear is unsupported on the 3rd stage for both the 2-CIM and 3-CIM ball shifter. For the 3-CIM I saw the hole in the plate and assumed it was 1.125, but it is only 1.000. Perhaps Vex thought the extra bearing would over constrain the shaft? A rare miss by Vex. The being said, the gear separation is 84T, so you could add some of my new favorite part, the Face Bearing Mount.

I am curious if you mounted your wheel right the to the output shaft or if the shaft was supported? I would take the 3rd stage output shaft and put two chain sprockets on in, then pass it thru a VersaBlock and put the wheel on the other side of the tube for a WCD setup.

The more you learn ...

[Paul Copioli]

BoilerMentor,

Were you guys using the old ball shifter shaft without the pin in it or the new shaft?

I ask because the press fit in the original shaft would allow the deflection you speak of but the new shaft would most definitely not.

Also, a cantilevered shaft with the proper bearing spacing behind it is a perfectly legitimate design strategy especially with the cantilevered gear so close to the external bearing face.

Additionally, this is the first example of a failure like this that we have seen with the 3 CIM shifter so I really would like to get more information from you.

PM me if you would like me to email you.

Again, this failure mode is not normal in the typical 3 CIM ball shifter use case (even with your ratios you are within our normal use case).

Thanks,
Paul

[aldaeron]

Originally Posted by Paul Copioli

BoilerMentor,
Also, a cantilevered shaft with the proper bearing spacing behind it is a perfectly legitimate design strategy especially with the cantilevered gear so close to the external bearing face.

It is a legitimate approach, but will have more deflection than if the end of the shaft were supported. Is the difference in deflection sufficient to cause a problem? Sounds like it was in a few cases.

Based on your reply (and the comparison here) it sounds like there were a lot of design changes in the v2 Ball Shifter Shaft. Under what circumstances do you recommend teams replace this shaft? We just bought the upgrade, but have not installed it.

[Paul Copioli]

Originally Posted by aldaeron

It is a legitimate approach, but will have more deflection than if the end of the shaft were supported. Is the difference in deflection sufficient to cause a problem? Sounds like it was in a few cases.

Based on your reply (and the comparison here) it sounds like there were a lot of design changes in the v2 Ball Shifter Shaft. Under what circumstances do you recommend teams replace this shaft? We just bought the upgrade, but have not installed it.

I would use the upgrade. The upgraded shafts are much better and allow for the use of the ThunderHex bearing without any additional machining.

By my math, the deflection difference is not enough to cause their problem. I believe it was related to a loose fit between the hex and shifter shaft in the v1 version of the ball shifter shaft combined with their increased load case.

[waialua359]

Originally Posted by Paul Copioli

I would use the upgrade. The upgraded shafts are much better and allow for the use of the ThunderHex bearing without any additional machining.

By my math, the deflection difference is not enough to cause their problem. I believe it was related to a loose fit between the hex and shifter shaft in the v1 version of the ball shifter shaft combined with their increased load case.

I bought several v1 Ball Shifters Transmissions when they first came out which are still unopened. We have yet to use them in our drivetrains, and used some of them instead for other robot features such as our winch for the ball launcher in 2014.
I better have them checked to possibly get the upgraded shafts.

[cbale2000]

Originally Posted by Paul Copioli

BoilerMentor,

Were you guys using the old ball shifter shaft without the pin in it or the new shaft?

I ask because the press fit in the original shaft would allow the deflection you speak of but the new shaft would most definitely not.

Also, a cantilevered shaft with the proper bearing spacing behind it is a perfectly legitimate design strategy especially with the cantilevered gear so close to the external bearing face.

Additionally, this is the first example of a failure like this that we have seen with the 3 CIM shifter so I really would like to get more information from you.

PM me if you would like me to email you.

Again, this failure mode is not normal in the typical 3 CIM ball shifter use case (even with your ratios you are within our normal use case).

Thanks,
Paul

I was wondering if someone from Vex would drop by this thread. Glad to hear this issue is an unusual/unique failure mode. Makes me feel better about our teams continued use of Ball Shifter Gearboxes (which as I've said already, we've been very happy with so far).

[BoilerMentor]

Originally Posted by aldaeron

I want to point out for folks who don't spend a lot of time looking at gearboxes that this is a very aggressive gearing with one of the highest CoF wheel tread materials. My quick calcs say this generates ~747 pounds of force at the wheel patch with 6 CIMs, well over the traction limit of 185 for a maximum weight robot in 2014. Meaning it is an extreme case and may not be the best one to guide the decision of an average team.

The decision was based not on pushing power, but a time to distance calculation that considered a two speed transmission. That pair of ratios minimized field crossing timed on a tool that was written up in Excel taking the two speed gearbox into account. People perceived 1747's robot as being tippy that year, specifically because of the brutal acceleration this setup generated. It was the one thing we didn't take into account and actually down-regulated our peak acceleration. No doubt the robot was quick across the field though. I personally think time to distance is an important consideration most teams neglect, but there are certainly some games where it's not a useful piece of data i.e. 2015.

Originally Posted by aldaeron

Overall this is great feedback. I have been wondering how the 3rd stage on the 3 CIM works at high loads. I always thought the sheet metal with standoffs 3rd stage looked a bit rickety. I am curious if you ever subbed out the 7075 aluminum gears for 4140 steel gears in the third stage of the 3-CIM shifter? Sounds like it was a misalignment issue more than a material strength issue. Did you ever try stacking on a second Vex third-stage plate to add stiffness to the third stage bearing? This would pickup more of the bearing race and possibly prevent angular deflection.

We ended up machining a replacement for the shaft with a 1/2" round to sit in a bearing in a plate opposite the face of the plastic housing. The 3rd stage mounting plate wasn't used. The hole pattern was transferred onto our drive train plate. We haven't had a problem since implementing the modified shafts, which lends a great deal of credence to the theory that it was deflection in the cantilevered shaft that caused the failure.

Originally Posted by aldaeron

I am assuming the output of the gearbox was direct driving your center wheel in a tank drive. How do you think a 2 stage 3-CIM shifter would work if offset from the wheel axles and with a #35 chain reduction between the gearbox and wheel axles in lieu of the Vex 3rd stage?

The axle was supported with a bearing on the opposite side of the drive pod from the gearbox, so as long as the mounting of the gearbox was adequate the bearing in the face of the gearbox and the outer bearing we added should have picked up the load. I did forget to mention that we machined custom shafts from the get-go for the output of the gearbox. We needed a longer output shaft to span our drive pod. These shafts went to 4140 steel, instead of aluminum. Also, out of center might be a concern with a custom machined hex shaft. I verified they were centered prior to installation.

Originally Posted by aldaeron

Was there a particular reason for adding this "stop on a dime" feature? Did it work as you had hoped? Would you recommend it? It does seem like a recipe for gear teeth shearing.

The automatic shifting code was very simple. Outside of handling for turning cases, there were two shift thresholds based on gearbox encoder feedback. I do not remember what drove the specific value used for down shift, but I think it was based on the highest speed that allowed enough separation between values to prevent oscillation between gears. The reason that the downshift occurred before full stop was in consideration of a number of logical conclusions about robot-robot interaction. The downshift feature did have one drawback. We had to have a specific case to handle direction changes, because the robot would end up on its back otherwise.

Just to clarify, my responses haven't been about "winning and argument" I'm just try to establish that we covered all of our bases as far as a well put-together investigation of the cause of the failure. It was actually a great exercise to be able to work through with my students. You can imagine in a competition setting there's a great deal of anguish caused by a failure of this magnitude in the student's eyes. It's a great feeling for them to be able to say, "This was not a flaw in our design, it was a failure of an input part. Here's why:"

-Charlie

[aldaeron]

Originally Posted by BoilerMentor

The decision was based not on pushing power, but a time to distance calculation that considered a two speed transmission. That pair of ratios minimized field crossing timed on a tool that was written up in Excel taking the two speed gearbox into account. People perceived 1747's robot as being tippy that year, specifically because of the brutal acceleration this setup generated. It was the one thing we didn't take into account and actually down-regulated our peak acceleration. No doubt the robot was quick across the field though. I personally think time to distance is an important consideration most teams neglect, but there are certainly some games where it's not a useful piece of data i.e. 2015.

I agree with you that this can be an important parameter and we do calculate it, but it is game dependant. Sometimes that few tenths of a second gets you a game piece or allows you to avoid a defensive hit. There is always a lof of discussion on free speed, but I think time to X feet is a more relevant metric. Simbotics talks about it in their videos on strategic design and drivetrain design.

Thanks for the details on the design and fixes. Since we have never geared that aggressively, it is good to know some of the pitfalls in case we ever want to.

-matto-