paper: Apalrd's Choo-Choo Analysis Spreadsheet

Thread created automatically to discuss a document in CD-Media.

Apalrd’s Choo-Choo Analysis Spreadsheet
by: apalrd

Basic design spreadsheet to analyze the pull-in process of a choo-choo type mechanism, for determining the optimal gear ratio and geometry given a constantly-increasing spring load during pull-in

I wrote this spreadsheet to assist with the Killer Bees 2014 robot design. This spreadsheet is a slightly cleaned-up version.

With this simulator, you can look at the motor speed and torque during pull-in and optimize the gearbox and choo-choo design accordingly. The spring forces you are using (both fully extended and fully retracted) are used to calculate the torque on the motor through the choo-choo mechanism.

It is notable that the length of the choo-choo fling length can be used to adjust the motor torque during the ‘first hump’ and ‘second hump’ of choo-choo pull in separately, so the motor load remains more constant. As the spring force increases with pull-in, a shorter fling length will cause the torque arm length to decrease on the second hump, so the motor will see less torque, and more length will be pulled in during the first hump vs the second hump. You can play with all of these lengths to optimize the design based on your spring setup.

The top graph shows motor torque and reference torque (set at a % of motor stall torque). Try to keep the two humps under the reference. I used 50% stall torque for this, since the choo-choo is not pulled in very many times in a match and I felt that the CIM could handle 2 seconds of heating. The second graph shows motor RPM during the pull in process. The simulation will repeat over and over for a fixed number of iterations, so the graphs will repeat also.

On 33’s 2014 robot there was a lead-screw based spring adjuster, so I ran the numbers in each of the positions it operated at to determine the worst-case choo-choo design and the others operated with a safe margin.

Let me know if there are any issues with the file, I saved it from OpenOffice.

choo_choo_math.xls (760 KB)

By ‘Choo-Choo’ do you mean the kick-driving mechanism that 148 did in 2010? I’m not familiar with your 2014 bot’s inner workings.

148/217’s 2010 mechanism is a snail cam, not quite like this.

Winnovation (1625) 2010 used a choo-choo, but it was mostly popularized by the 2014 Build Blitz where Team JVN used it.

Here’s a graphic I found on CD a few years ago (but couldn’t find it with a google image search). Thankfully I saved it. Here it is:

In my spreadsheet, the blue link is the ‘fling’ arm and the red link is a cable which connects to the throwing arm. The length of the red link is not important as long as it pulls straight down on the throwing arm. Some designs use a hard link instead of a cable for the red link.

Also used by team JVN in the 2014 BuildBlitz

FYI - the choo-choo cam was used by Winnovation in 2010 and JVN’s 2014 Build Blitz. Those two teams had one thing in common (Aren Hill) and that is no small coincidence. ;o)

I definitely like how the launcher can be held fully cocked with little torque required from the motor; probably little enough to let the gearbox hold it (especially if that last 12:1 reduction is through a worm gear).

Is the driven axle assumed to be located midway between the “pivot” and the “pin” mounted to the disc? Have different axle locations been investigated?

If the axle was not in the line between the pivot and the pin (below this line when horizontal), it appears that the two drive periods would overlap and merge into a single steadier pull, at the cost of introducing a null sector at the opposite part of the cycle.

You can also use something like a ratcheting wrench to be an anti backdrive. Unless you want to be able to slowly release, the choo choo only ever needs to rotate in 1 direction.

I also believe it was in their 2008 robot also if memory serves me correct.

We were in an alliance with Winnovation when they used the choo-choo to win the Colorado Regional in 2008. At the 2014 kickoff we were making train sounds as soon as we saw the game! We ended up using a choo-choo cam and made the hub of the sprocket a ratchet then used a pawl to hold it in the firing position thus not stressing the motors.

33 2014 used a NBD (‘Nothing But Dewalts’) transmission with the CIM interface. We had ordered a huge number of DeWalt replacement parts when NBDs were popular and are slowly using them. I wouldn’t recommend it for new designs. If I did it again I would probably use the ratcheting wrench trick (using a 1/2" ratcheting wrench on a 1/2" hex shaft) on some intermediate shaft in the gearbox.

Also the example gearing is not what 33 ended up with, but it was an early iteration.

The entire purpose of the spreadsheet was to investigate different locations for the pivot and the pin, based on the start/end force of the spring.

There are two geometry points on the choo-choo. The first is the diameter of the disc. The second is the length of the ‘fling’ arm (blue in the illustration above).

During the first half of the rotation, the pull-in length will be 1 diameter of the disc and the torque lever (torque applied to the motor) will be 1 radius of the disc. During the second half of the rotation, the pull-in length will be (fling length - radius) * 2and the torque lever will be (fling length - radius).

If the spring force starts at 0, it would be ideal to make the fling 3/4 of the diameter. The first hump would then pull in 2/3 of the total pull-in length with a longer torque lever (more torque on the motor), the second hump would pull in the other 1/3 of the total pull-in length but the torque lever would be shorter. Since the spring force is constantly increasing, the torque seen by the motor will be equal between the two humps.

As your spring equation changes, you can play with the diameter and fling length to get the total pull-in distance that you need along with balanced torque humps for fastest pull in.

The pin location doesn’t matter at all, it can be anywhere on the disc as long as it pushes the fling arm. 33 2014 used the driving axle as the pin and made a notch in the fling arm for the axle (with a thick washer on the end) to nest into. I’ll see if I can find pictures.

This is true. However, the ‘null sector’ would actually be a slight reversing of direction and the total pull in distance would be reduced (it would pull in the mechanism a bit too far, let it out a little, then let it go).

Does this mean that the “cocked” position could require exactly zero force to maintain (though it would be metastable, at the “top of the hill”)? If so, I like it even better.

The Choo Choo is a favorite mechanism of mine, Dillon Carey and myself have a fancy version of a Choo Choo sheet with graphs and a rewind time calculator as well. (He did most of the work)

The first rendition of this mechanism I used in FRC was 2008 on 1625, with a fisher price motor in a Nothing but Dewalts setup for anti-backdrive.
Second implementation was 2010 for the kicker, same deal.

Then when 2014 rolled around and Build Blitz was happening…yup
Also Versaplanetary and a ratcheting wrench? way easier than dewalts.

Killer Bees were just the first people I’ve seen actual math from.


I did a reverse image search on my saved copy and it didn’t find it anywhere on the internet. I thought it had been lost. I guess it doesn’t index CD attachments.