Elevators: Cascaded vs Continuous

I have only ever had experience with the continuous one (where one stage moves up at a time), and I can say that I’d prefer cascade based on how 2415 used it, and their increase in speed. However, I suppose such a speed could be achieved by any system assuming the proper balancing of the lift.

One issue with continuous, if you’re going to truly go that way, is in my experience, the intermediate stage tends to go up first. I’m not entirely sure how to circumvent that problem, but I suppose if you could get around that, then continuous would be preferable.

  • Sunny G.

We ran a continuous system in 2011. It was good.

Initially it’s because we wanted the carrier stage to be able to go all the way to it’s travel limit before the intermediate stage lifted, so we could rotate the arm over backwards. Eventually, we learned that this skill was completely useless because the human players were so good, but we still designed around it.

It gave us no troubles. the elevator was 1x2 box welded to the 1x2 box cantilevered axle drivetrain (‘west coast’ style), the elevator was driven via a large (2") drum which spanned the width of the space between the two elevator boxes (it was mounted to a riveted triangle gusset). It was driven by 2x 550, no efficiency issues to think of.

We counterforced it with a pair of 20lb constant-force springs, one drum-mounted to the top of the fixed stage and connected to the bottom of the intermediate stage, and the other wound around a structural tube of the carrier and connected to the top of the intermediate stage. It was a lot of force, slightly more than gravity but not enough to overcome gravity and friction, so the elevator would generally stay where you left it if the robot was on it’s feet and gravity was down. If you disconnected the motors, it would drift up over time, actually.

If you run a continuous system, then the length of string you have to deal with on the drum is equal to the lengh of travel of the elevator. If you run a cascaded system with a string and drum, then you only have to deal with half of the string but twice the force, requiring 2x gear reduction (more gears = heavier = less efficiency there). It is possible to run a continuous system with chain, although that would be a lot of chain, and chain is heavy.

The speed of any system is purely determined by power exerted by the motors. This means, to achieve the fastest speed, one must:
-Minimize the weight seen by the drivetrain (lightweight components, counterforce)
-Minimize efficiency loss
-Maximize motor power
-Gear optimally
It has absolutely nothing to do with string geometry. The lifted weight is still the same.

We’ve done both.

The first was continuous. Routing the wires was our biggest issue, finally solved through using energy chain to help guide them along the correct path. We didn’t have any real mechanical issues, once we got it figured out.

The second was cascade. This made wire routing extremely easy, and we never had any issues with that. However, this elevator had a lot more mechanical issues, partly around binding and partly due to a design that forced us to stall the FP motor in both the up and down positions. I think we just had a slightly worse design for this particular one, and for some reason had a harder time keeping everything square.

The biggest issue is ensuring there isn’t binding. We had an extremely overbuilt design the first time, so everything stayed square and didn’t bind. While we used a similar design for the moving parts, things didn’t hold as square and there was a lot of binding the second time. To alleviate this issue, I really loved what Team Titanium did with their elevator in 2011. Basically, you can look at the outer stage as two pieces of angle iron, and the inner stage interfaced with that using roller blade wheels. A top down view would look something like this:
<- ->
The fixed angle iron is <>, the roller blade wheels that roll up and down it are - -.

I think the decision on which to use really comes down to motor power and weight. Cascade typically requires more power, as each segment essentially doubles the force needed to lift your weight. Continuous doesn’t have this problem, and the force required is directly related to the weight of the elevator + game piece. If you have motor power to spare, I would go with Cascade to make wire routing easier.

Also, take note of how long the wires are if you have motors on the end of the elevator! Those long wires can significantly drop the voltage the motors see. You can compensate for this in two ways: First, use larger wire. This can be tricky, as larger wire tends to be less flexible. Second, use multiple wires for each motor. Run two wires from each motor lead through the robot and to the same connector on the speed controller (this way you are still only powering it through 1 speed controller!). This will cut the voltage drop in half, and deliver more power to your motor! Expect to be questioned on this by your inspector, as inspectors aren’t used to seeing multiple wires run out of each connector on the speed controller or the motor, but you should be able to prove (with a multimeter) to his/her satisfaction that it is actually legal!

We cascaded our elevator in 2011 and it worked very well. A couple of advantages: Movement of the intermediate stage is predictable and proportional to the overall lift. With continuous rigging the intermediate stage can theoretically move first or last or somewhere in between, depending on where the friction is. With cascade the total lift is 2X the length of the cable wound on the winch, so you get 2X the lift speed with 1/2 the motor revolutions, and 1/2 the sensor revolutions if you are measuring. By cascading we were also able to route an umbilical to the arm (air line and solenoid wires) by simply going over a pulley on the intermediate stage, which moved in proportion to the lift.

Can we get some photo examples of each?

1 Like

I have mostly worked with continuous elevator designs. The first that I worked on was the 2004 GRR robot. We used 80/20 extruded aluminum with the recommended sliders to create a version of an extension ladder. A CIM motor powered the elevator design with timing belt and pulleys.

The main concern that we had was the friction that this design created. If you watch the video link below, you will notice that it takes about eight seconds to get to ten feet. The friction between the 80/20 and sliders proved to be a constant battle. We learned that the weight of the 1" X 1" 80/20 created a bowing effect that was hard to overcome. Using graphite in the sliders improved the design.

Overall it worked very well and accomplished the game tasks of getting a hook to a ten foot bar. We continued to improve on the design in future years on the team.

There is much more detail on the robot on FRC Designs: Behind the Design. Check out this link for more information (http://www.frc-designs.com/btd/2004_340.html).

Here is a photo example of the 2004 GRR robot.


Not a photo… but


Is a part of the larger presentation from 2001 found here.

I have often heard that cascaded elevators require a return drum of a different diameter than the feed. But wouldn’t the following work?

Each segment of the elevator can have a closed loop of belt/chain attaching it to the prior stage. It should be anchored at the top of the previous stage (or motor for the first stage), and the bottom of the next stage. It’s important that the total length of the loop doesn’t change as it travels up and down.

You still have the speed and force multiplier effect with each stage, but you won’t have to worry different feed and return speeds.

Yes this will and does work. I just finished designing an elevator that uses this principle with a rack and pinion style. I will soon be finishing a fee things and create a thread or something. As I put it on my facebook I found out from Adam that 973 did this exact style for their 2011 offseason robot. If you havent seen it you can look it up here on Chief Delphi, or also go to FRC -Designs.com and the CAD is there fr download, really nice looking if tou haven’t downloaded already.

Jared you can go on my facebook an look at it there in my Robotics Album. I uploaded it last night, but you can’t reallysee the design very well of what you are describing. Thats why I will upload it in a thread soon and will likely add my design to FRC designs.

We used a very effective cascade elevator in 2011. The main fixed mast was 2 pieces of 2 x 1 x 1/8 channel with Igus linear slides mounted inside. The first stage was 1 1/2 x 1 x 1/8 rectangular aluminum tube with the igus shuttles mounted. It was driven up by a Gates belt driven by a 775 motor and a Bane Bots 64:1 gearbox. It was helped down by a constant force spring. The second stage also rode on Igus linear slides and was pulled up and down by lightweight cable. Once we got it fine-tuned, it worked really well.


Ignoring the fact that our drive train was slow as death, 1:40-1:45 in this video gives a good picture of our 2008 cascade forklift (and the World’s Biggest Roller Claw) in action.

We made one enormous mistake in its design, and that was using steel cable instead of belts. Belts good. Cable bad.

A second mistake was that the CIM was utterly unassisted by any kind of force spring or other mechanism… My only excuse for that is that we didn’t really know what we were doing, and it was well better than our tetra lift design a few years prior!

Even so, it worked pretty well…

By general feelings on cascade lifts is that they work just fine and are not difficult to build, and in the right circumstance are a good solution for lifting things.

Our general design that year was very similar… except we used continuous. We also used steel cable, but we had no issues with it the entire season. You can see ours in action here from 1:00-1:15. It was powered with a single FP motor, and if we had really known what we were doing, we probably would have put a bit more power behind it and ran it faster.

We’ve used Spectra cable for several elevators and been very happy with it.

Our 2008 and 2011 elevators both used a thin variant of it with 800lb rated load, generally used as high-strength kite string. No issues with 2011 robot, ever. It’s super lightweight, too.

Our 2012 stinger initially used the 800lb cable, but we eventually broke it by landing on the stinger. We replaced it with three 800lb cables braided at that tournament, then found some higher-force thicker Spectra harpoon fishing cable, which lasted the rest of the season.

We used a continuous elevator in 2011 and we absolutely loved it. The “clean-ness” of one stage coming out a time was really nice in my opinion. It meant for an intermediate lift there were less parts of the robot moving and less load on the system for a lift where the entire length of the elevator wasn’t needed.

We used Dyneema sailing line which is basically another brand name of Spectra cable. The nice thing is it can be picked up from West Marine locations (we have a ton around us). It’s also available in several colors, which is always a plus. We ran the same line the entire year through ~90-95 matches with no issues. The line is still in great shape this day. Link for reference.


My aging brain is not visualizing this, perhaps you could sketch it. The purpose of a return/take-up cable is to drive the elevator down as well as up. You could create a closed loop cable between the 2nd and 3rd stage, but it would not drive the 3rd stage. Remember that the lift cable for the 3rd stage is anchored to the 1st stage, not the 2nd. It therefore needs a corresponding return cable which is also based on the 1st stage (the 2X take-up). It is a closed loop overall, but not between stages.
We solved the 2X take-up issue by making our winch drum a 2-stage drum (photo) with the return stage 2X diameter of the payout stage. Made the cable rigging quite simple.

Funny, we did exactly the same thing when we did cascade!

Here is a picture of what I was envisioning.

The motor is directly connected to a continuous belt/chain/cable around only the first stage which is anchored to the bottom of the second stage. Thus the motor can drive the second stage both up and down. The rest of the system is constrained based on these two moving parts.

Basically, it is a conventional cascade lift, attached to a cascade “lowering” system.


Each moving stage acts as a closed loop, but the linkage involves the two stages before it. For example, the cable (belt, chain, etc.) would be anchored to stage 3, run over pulleys at each end of stage 2, and be anchored to stage 1. The stage 2 cable would be anchored to stage 2, run over pulleys on stage 1, and be anchored to the non-moving chassis/frame. The stage 1 cable would be anchored on stage 1, run over pulleys on the chassis/frame, and be anchored/attached to the drive drum. The ends of each cable terminate at one anchor point which includes the tensioning adjustment. The other anchor point for each cable is normally a clamp restraint on a continuous section of the cable. When loosened, it allows for adjustment of the location of that stage relative to the other stages. It doesn’t matter which of the stages the different anchor points are on.

We have used both methods (continuous and cascading). Both work fine if done well, and both stink if done poorly. My advice is to use a design that allows for easy access to cable tension adjustment and replacement, and make it so that the cable cannot come off the pulleys when tension is lost.

Now I get it. Clever. Gets rather complex with double the cables/belts and pulleys, but it would work. I would probably still choose the 2X take-up reel, it is easy to do and greatly simplifies the rigging.

330 did that for multiple years (1999, 2000, 2001, 2004, with a single-stage moving piece in 2002). No variation in drum diameter between up and down cables; the return cable simply attached at the top and fed back to the drum. The main issues with those lifts (all blue, all cable-driven) were when cable needed restringing–break out the aircraft cable crimpers and sneak them in somehow to make the crimp. I can’t recall tensioners being used at all.

The other issue I can recall was a direct result of two FPs trying to drive the lift down past its hard stop in 2004. I think we were able to bend the C-channel back to normal; I do know that it held up through the rest of the regional and the Championship and offseasons, and if we powered it today it’d probably still function.:smiley: