Air Pressure Loss in Shifting Gearboxes

This hasn’t happened to our team as we’ve never used shifting gearboxes but I thought of this problem. If you are using dog shifters and lose air pressure, the dogs will stay engaged in the last gear ratio due to the shape of them. However, if ball shifters are being used, then a loss of air pressure would kill the drivetrain due to the way a ball shifter works.

Is there a workaround using shuttle valves, check valves, or whatever fancy valves that exist, that in the case pressure is lost before the solenoid, there is a way to maintain air pressure in the shifting mechanism so the robot isn’t beached?

1 Like

Hi there,

In my experience, when using a non-spring loaded cylinder in a dog shifter, it can still pop out of gear when you lose air pressure after a few direction changes.

A workaround for this is to either use a spring-biased air cylinder, or to add a small spring along the shift rod in one of these gearboxes. AndyMark EVO Shifters ship with a spring-extend cylinder because of this potential issue.


1 Like

To chime in with years of experience with ball shifters, they won’t disengage simply due to a loss of air pressure.


It’s difficult to imagine anything which would do this, and still be compliant with the long-time rule which was R93a this year:

R93. Any pressure vent plug must be:

  • A. connected to the pneumatic circuit such that, when manually operated, it will vent to the atmosphere to relieve all stored pressure in a reasonable amount of time, and

As Nick suggested above, the common failsafe for this case is to use a single-acting cylinder with spring return.

But spring return cylinders don’t offer full force at both ends of the stroke. So for a ball shifter I’m assuming, depending on the gearing, the gear could come off the balls. Is that correct for me to assume?

You may be misunderstanding how a ball shifter works. The balls are pushed outward into depressions on the gear by a shaft sliding in between them. The shaft comes to rest with the balls on the flats of the shaft - there is not axial force that would make the shaft move. A small spring should be more than enough to hold it in place in an emergency.


@ProPain37 your assumption that ball shifters disengage when pressure is lost is not accurate.

Perfect diagram. There’s no load on the shifter shaft besides inertia. We’ve taped them back and driven matches with failed air systems.

Spring-return air cylinders is an upgrade that I’d love to see for the Vex ball shifters.

1 Like

The ball shifter will not “pop out of gear” right away, but over time it tends to move and will eventually pop out of gear. This could be a few seconds or a few minutes, but we’ve had this happen plenty of times when we forgot to pressurize the system during practice sessions.

Adding a spring to hold it in one direction if there is no pressure would be a great idea. But if the spring was trying to hold the shifter rod in the “pulled out” direction and the pressure was lost when the rod was in the “pushed in” direction. The spring would actually cause the shifter to come out of gear. If the spring was strong enough to push it into the opposite gear, then you are good. But it would need to be more than a light duty spring to do that.

You could add an accumulator that was directly connected to one port of the shifter cylinder and pressurize that accumulator to 30 psi. The other port would be connected to a normal pneumatic channel. When activated, the 60 PSI working pressure would push the cylinder piston with 30 PSID of pressure (60 on one side and 30 on the other). When you de-activate that channel, the cylinder piston would push the other direction with 30 PSID of pressure (0 on one side and 30 on the other). The 30 psi accumulator would “bias” the cylinder in one direction.

The accumulator would not need to be as big as a clippard tank. It could be a much smaller “tank” as long as the volume was large enough that the pressure wouldn’t change much when the shifter cylinder changed positions.

If you have a regulator to step the system pressure down to 30 PSI into this accumulator tank and then a check valve to prevent the accumulator from leaking back into the main system if you lost main system pressure, you would be able to pull this off. Edit - as long as the vent valve is downstream of this second regulator, you can still dump the entire system pressure.

Edit: brain fart - the vent valve can still be located in the system to be able to dump the whole system. Sorry for all the convoluted attempts to design a legal system in my earlier post. They were unnecessary and I have removed them for my own sanity and to save you from having to read the ravings of a madman.


As Nick said, Dog gears will pop out of gear if air pressure is lost after a few turns.

It happened to us at Champs this year, we were playing defense on 4028 and 195 and lost air at about 80 seconds. I attempted to turn around, but the dog gear in our right side gearbox came out and we drove straight into the HAB. I knew we had lost air and had to keep the robot over there, trying to stay away from the hab and rocket to not get penalties. It looks like 195 tried shoving us into the rocket for the RP at ~20 seconds, but we weren’t quite lined up.

This is clever, thanks for the cool idea! I’ll try to keep this in mind next time we use shifters.

I’ll be another voice chiming in saying that ball shifters don’t pop out of gear when they lose pressure. We had a practice chassis sitting around for a while that had shifters, but we’d scrapped the pneumatics off of it. It held in gear just fine. Maybe once per day of use we had to manually put it back in gear.

Wouldn’t the check valve prevent air from coming out of the mini accumulator?

The check valve would be between the secondary regulator and the accumulator such that if the main system pressure dropped below 30 psi, the pressure would not flow out of the accumulator back into the main system. The pressure lines from the accumulator to the shifter cylinders would need to be downstream of this check valve.

In ancient times a hobby servo was used to shift. Back then the servo rules limited the power available. With the REV servo expansion hub, relaxed servo rules, is servo shifting now an option? The new Gobilda servo’s have vastly more power than back in ancient times. We use the REV servos and this year the Gobilda to move a whole motor assembly on our CVT.

looks like someone has already thought of that. I swear I thought there was a larger thread about it and how it has worked on the practice field but cant for the life of me find it. maybe @cpapplefamily could shed some more light on it?

this might be what I was thinking about.

So, this wouldn’t be that hard to do yourself. Spring return cylinders with 1/2" stroke are available from McMaster Carr for either spring extend or spring retract. These cylinders are designed so that you install them through a hole in a plate and then use the 1/2"-20 nut on the nose of the cylinder to secure it. So all you need is a plate with that hole in it and two other holes for the standoffs. I can’t find a drawing of the VEX pneumatic mounting bracket to see if the hole in that bracket is large enough to accept this nose of the piston, but it might be. If it is not, you would need to enlarge that hole and you might need to make a plate with a larger hole in it. You also might need to lengthen the standoffs or add some spacers between the standoff and the plate to account for the difference in the length of the rod between those pistons and the pancake piston that VEX designed the kit for. All that is pretty simple stuff.

So the thing with a spring return cylinder would be the robot would have to be going pretty slow to get the robot to shift in its default gear. Cause the spring would need to provide enough force to overcome the “ramp” in the ball shifting mechanism. Thats the only issue I see with using a spring return.

Also, if you go on SMC’s website, they do sell the identical cylinder to the one VEX uses except they also have spring return options so I’m assuming that would allow native support with all the shifting gearboxes vex and wcp sell.

Yes Here is the link to my with a small video of it in action. On of my biggest push is to have the team design redundancy into system when possible.


We Bought these cylenders before designing the thingivers thing. Turns out the spring return cylinders do NOT have rode seals so you can NOT use air to retract the cylinder. I did chat with them about this and had a quote to build a cylinder that both had air and Spring return. Since it’s not a catalog part the cost was $200+ each.

I thought this was ridicules. Turns out to go from a 1/2 stroke dual action to a 1/2 stroke spring return they move to a 3/4 cylinder body and replace the rod cap with a none sealing cap that has a spring pocket and adds a spacer to the piston to stop the extend stroke from crushing the spring.

What I wanted is to go to the 1" stroke body, keep the rod seal add a second spacer to the spring to stop the extend from crushing the spring. If FIRST was not so strict on modifying pneumatic I would have just bout the 1" body and made our own spacers. These parts are serviceable but the modifying rules prevent this internal stops In my Opinion

A final note on this is you loose the encoder option but it better to put the encoder closer to the output shaft anyhow. Or now you could take a step back and use the encoder built into your new brushless motor integrated encoder.

Yeah, the return spring is not very strong. It is only about 5 lbs of pull force. I’m not sure if that is enough to shift the gearbox. I would want to test it and see if it was strong enough to shift under reasonable loads.

@cpapplefamily - it sounds like you tried these? did they have enough spring force to shift? What force did your thingaverse return spring have?