Some basic Pneumatic Design Questions

Hello Everyone,

My team has recently started exploring in pnematics and we have some basic questions:

  1. How do we determine how many tanks we will need if we don’t want to carry a compressor? I know we can calculate how much air the system will use depending on how many times we fire, but is it that simple?

  2. When CADing, how do you go about working in the cylinders into your design? Are there specific models? Do you just take the stroke of the cylinder and adhere all designs to that?


  • Sunny G.

1)It is a smart choice not to carry a compressor,
If you keep the pressure above 60 psi, the answer is simple and as you have explained. Once the pressure in the storage tanks goes below the regulator pressure, the relationship is no longer linear. I will let someone else get into it because I am not sure how to calculate it.

  1. There are two kinds of motions you can get from pneumatics; linear or some curve using a linkage.

When designing, I like to make the best estimate I can as to how long a cylinder I can use. Then, I will often use one that is a bit longer.

When designing linearly remember that you can adjust how far the cylinder retracts easily by adding a spacer to the piston. Adjusting how far the cylinder extends takes a new mounting system.

For linkages:
Design the system so that you can mount the cylinder in many different places. A hole pattern spaced ~ .375 inches can do this nicely. And remember that you can restrict retraction with a spacer too.

One other note on linkages,
When the cylinder is perpendicular to the lever arm, you are generating the most torque. Anywhere less than perpendicular is generating less than the max possible torque. This is a good thing to keep in mind when you are designing clamps and wrist joints.

Warning: Oversimplification. If you’re just shifting a few times a match, and you have 3-4 tanks, that’s a good choice. If, OTOH, you happen to have a long-throw or large-bore (or both :yikes: ) piston firing a grabbing device dozens of times a match, on top of shifting and maybe an endgame pneumatic, you’re going to run out of air pretty quickly unless you have a massive amount of tanks. In that case, you probably want to carry a couple fewer tanks and a compressor.

  1. There are two kinds of motions you can get from pneumatics; linear or some curve using a linkage.

Don’t forget using rotary pneumatics. Those are legal under 2013 rules, IIRC.

  1. It’s that simple. 2 cubic inches of air at 60 PSI = 1 cu in of air at 120 PSI.

Step 1 is to determine the throw of the cylinder you need. Round up to the nearest inch - it is easy to limit throw, but it can’t easily be increased
Step 2 is to determine how much force you need. 60 PSI acting against 1 square inch (or piston area) gives 60 lbs of force (approx, there are friction losses). Use the smallest bore you can, since it uses least air. Note a larger bore can deliver more force, or the same force if the pressure is lower.

Step 3 is to Find the cubic inches at your given pressure. a cylinder with a 1 square inch piston that is 5 inches long uses 5 cubic inches.

Step 4 is to guess how many times you will actuate the cylinder. Extend and Retract are each “one time”.

Cubic inches * # of actuations = air needed. Multiply by (output pressure / storage pressure) (usually 60 and 120) and that’s how much storage you need “theoretically”. (Tanks have a cubic inch capacity specified) In reality, pressure drops as you draw against it, so it becomes a differential equation… So just round up and add a tank, empirically it will be close. Then test.

So, suppose a piston uses 1 cubic inch of air, and is fired thirty time. As per your math, it would be:

1 cu in * 30 actuations * (60 output psi / 120 psi) ~> 15 cu inches of “theoretical” storage?

And as far as the design question is concerned, my question, which I now understand was slightly misleading, was more in regards to CADing the piston.

Are there nice models with the appropriate motion constraints built in, or do people just model in a pole and restrict its range of motion?


  • Sunny G.

There are definitely some. Check the Bimba website. Usually, when you import them as a .step, you have to add in the motion constraints yourself.

Actually, Bimba has a really cool CAD generator that can hook into your CAD software and watch it be made in front of your eyes. I can’t recall if motion constraints are placed or not.

This is how we calculate air needs, but we usually multiply by 2, which would mean (using the simplified linear model) that you end the match with full working pressure, so that your actuators don’t lose force. Basically this also becomes a safety factor - you can continue using the system for more actuations if necessary, but the force will reduce with continued usage.

We pretty much exclusively use Bimba cylinders - between the free product voucher in the KoP and the fact that McMaster carries them. Bimba has configurable CAD models of all their cylinders on their site.

If you use the Direct Insert method (which controls your Solidworks/Inventor/ProE remotely using a Java app and effectively generates the part on the fly) I think it may add the constraints. (It’s been a while since I’ve used it, though, so I may be wrong.)

The Direct Insert process takes quite a while on my computer, so I usually opt for the STEP file download. STEP files don’t have constrains, but if I need motion simulation, then I group the parts into either two subassemblies: cylinder and piston, then apply a single constraint to enforce axial motion.

I use Inventor - if you use Solidworks or ProE, it looks like there are application-specific formats which may include constrains.

<up on soapbox>

Please they are not called pistons they are called cylinders.

<off of soapbox>

Yes. That 15 in^3 would drop to zero if it remained at 120 PSI until the last actuation. In reality, it would drop about 4 PSI (1/30 * 120) each actuation, so after 15 actuations your storage pressure would be 60 PSI. You can do the math to see where you’ll end up. Ryan’s “double it” is reasonable.

Be sure to take into account all leaks, which can be non-trival. Also, storage is often not at 120, but a bit less.

An AndyMark AM-2478 ($15) is 30 in^3, weighs 0.56 pound, two of those puppies will keep you moving 1 in^3 well above 30 actuations. On the other hand, a 12" cylinder with a 2" bore would work maybe twice at full pressure.

But would “wow so cylinder” look as good in our shooter?

Keep in mind the difference between gauge and absolute pressure, and your calculations will work out pretty well (or they have for us in the past). 60 psi gauge pressure is 74.7 psi actual pressure, and 120 psi gauge pressure is 134.7 psi actual pressure. This means each cu. in. of storage air corresponds to roughly 1.8 cu. in. of air in max working pressure. Another source of error in these calculations is assuming that the air is behaving as an ideal gas, but this is close enough for determining the number of storage tanks needed on an FRC robot, especially considering it makes sense to include a safety factor anyway.

A couple of finer details to consider.

If you are working with the constraints of only using tanks, remember you have to feed the cylinder on both the extend and retract if it is a repetitive motion. In these situations, it may be favorable to look at cylinder that has a spring to return for one of the motions, so you don’t have to supply air from the tank to return the cylinder.

Also there is what is called rod end of the cylinder. This is the side the working arm attaches to the piston in the cylinder. The working area on the rod end, is the area of the cylinder bore, less the area of the rod bore. So the rod end has less force at the same psi. Also the volume of air is less on the rod end of the cylinder.

In the drop calculations, I’m assuming that the 4 PSI is dependent on the cylinder size? Or is that more of a rough constant?

  • Sunny G.

We pretty much exclusively use Bimba cylinders - between the free product voucher in the KoP and the fact that McMaster carries them. Bimba has configurable CAD models of all their cylinders on their site.

Actually, you are not guaranteed to get a Bimba from McMaster-Carr. We have gotten several non-Bimba (but compatible) cylinders from them in the past two years.

Until you run out of pressure, or forget to recharge between matches, or don’t have time to recharge during matches (think eliminations).

The new compressor is light and small and only needs a spike to control it. You will be investing in all of these things anyway since the robot needs to control the compressor to charge it’s tanks per 2013 rules.

There are legal rotary actuators available per 2013 rules.

This is very conservative, and unnecessary in my experience. It will cost extra air even if you restrict throw on the cylinder somehow, it adds extra weight, and cannot increase your force capability without re-mounting the cylinder.

If you want to be conservative I would suggest getting the right stroke air cylinder, but in a diameter slightly larger than what you need. To conserve air: regulate your operating pressure down below 60psi. This gives you some headroom to adjust pressure (and thus force) up later if need be. It does add some weight too, of course, but is more efficient with compressed air than restricting stroke.

McMaster and Bimba have very good CAD models of their air cylinders available in a large number of CAD formats (Solidworks Native IIRC). The model includes a deployed and retracted position, so it’s very easy to figure out what length cylinder you need and where you need to mount it.

Note that basically every air cylinder I’ve ever worked with has a threaded end. This allows for a large amount of adjustment forward and back of the rod end on the cylinder. These adjustments can be made with the cylinder in place too. This is a great way to fine-tune the overall length of an air cylinder. It would be excessive to have a hole pattern spaced smaller than the amount of adjustment the threaded end gives you, but this varies per cylinder. A decent rule of thumb is that you only need 3 complete turns of thread to engage the full strength of the rod-end, so there is a lot of adjustment to be had!

I don’t mean to sound too critical of you Jeffy :slight_smile: I’ve been designing FRC pneumatic systems since I was a freshman in 2001. I used to design things the same way you’re suggesting for a while. It works, but it’s not the best way. I have found what I think are better design practices in the intervening 12 or so years.