designing a three motor gearbox

[1683cadder]

I want to design a custom gearbox for practicing my engineering skills. I have been thinking about designing a two speed gearbox with 2 cims and 1 minicim. Would that be a legal gearbox? What are the formulas I need to consider? Other words of advice? Is there a way to test the gearbox in CAD? Can my team use the gearbox design next season if I publish the CAD?

[1683cadder]

also is there other good engineering info I should know? Like designing custom gears?

[Walter Deitzler]

Originally Posted by 1683cadder

also is there other good engineering info I should know? Like designing custom gears?

I would not go with with custom gears unless you realistically have some way to maufacture them. Designing something that you cannot make is a bit of a letdown.

I would download the WCP gears .step files, as well as the one from AndyMark and VexPro, these are a good place to start.

2 CIMS and 1 miniCIM will be legal for this year, but as for next year, we do not know (hopefully it will be!)

As for formulas, download JVN's gearbox design spreadsheet. It is one of the most useful thing you can ever use for gearboxes.

[1683cadder]

Originally Posted by LedLover96

I would not go with with custom gears unless you realistically have some way to maufacture them. Designing something that you cannot make is a bit of a letdown.

I would download the WCP gears .step files, as well as the one from AndyMark and VexPro, these are a good place to start.

2 CIMS and 1 miniCIM will be legal for this year, but as for next year, we do not know (hopefully it will be!)

As for formulas, download JVN's gearbox design spreadsheet. It is one of the most useful thing you can ever use for gearboxes.

Shouldn't the minicim be different than regular com? I thought they had significantly different rpm
Edit: sorry for the misquote

[Tristan Lall]

Originally Posted by pmangels17

Designing for two CIMs and a miniCIM is the same as designing for three CIMs, they have the exact same mounting style.

Originally Posted by 1683cadder

Shouldn't the minicim be different than regular com? I thought they had significantly different rpm

To a first approximation, you can substitute one for the other, but if building a gearbox, you'll probably want to consider whether interchangeability or optimizing the operating state is more important. For most FRC applications, I'd lean towards running all the motors at the same points on the power curves (but because they're different motors, the magnitudes will be different). That means slightly different gear ratios, with less interchangeability between CIMs and Mini CIMS.

[1683cadder]

Originally Posted by Tristan Lall

To a first approximation, you can substitute one for the other, but if building a gearbox, you'll probably want to consider whether interchangeability or optimizing the operating state is more important. For most FRC applications, I'd lean towards running all the motors at the same points on the power curves (but because they're different motors, the magnitudes will be different). That means slightly different gear ratios, with less interchangeability between CIMs and Mini CIMS.

please enlighten me; i honestly have not much of a clue. How do i make two different motors work together?

[EricH]

Originally Posted by 1683cadder

please enlighten me; i honestly have not much of a clue. How do i make two different motors work together?

You match the output speeds of the motors.

Here's an example: Let's say that I have a CIM (free speed 5500 RPM) that I want to have work with an RS775-18 (free speed 19500 RPM), in the same gearbox. 19500/5500 is 3.545, so I need an extra gear ratio included somewhere between the RS775-18 and the CIM that is 3.545:1. (There are other considerations--see Tristan's post--but that's a reasonable starting point.)


What Tristan is saying is that if you're doing a quick-and-dirty gearbox design (first iteration), the MiniCIM and CIM can be considered interchangeable. But, your performance may suffer as a result of that consideration, due to the motors being different (say, by 1000 RPM at free speed). So maybe you want to optimize your motors' performance in a later, more refined version. The way you do that is you look at the motor performance curves (like the one for the CIM, down in the lower middle of the page). As speed goes down, torque goes up; as torque goes up, so do the amps the motor pulls. I don't see the power line, but if you remember that power=torque*RPM you should be able to construct it (it should be a curve).

For example, I seem to remember that a particular team in a particular year designed an arm around operating a particular motor (not in the current KOP) at the top of the power curve. Anybody who has some experience doing the same thing knows what comes next... Unexpected loads caused some loss of power, resulting in the arm failing pretty badly at times. The team has since designed farther to the left on that curve to have extra power in such situations.

Anyways, if two motors are different, matching their speed will work for a quick solution. But to optimize, you probably want to match their available power at a given RPM, which will require slightly different gear ratios than just matching speeds.

[1683cadder]

Originally Posted by EricH

Anyways, if two motors are different, matching their speed will work for a quick solution. But to optimize, you probably want to match their available power at a given RPM, which will require slightly different gear ratios than just matching speeds.

How do you do that?

[Tristan Lall]

Originally Posted by 1683cadder

How do you do that?

The speed/current/power vs. torque curves are shaped similarly for DC permanent magnet brush motors (i.e. all current FRC motors). Motors differ in the magnitudes of these quantities. You plot those curves (either experimentally or theoretically), and then decide what ranges you'd like your mechanism to operate in. Then you scale the curves using gear ratios (to multiply the torque by some factor). Look at the JVN DesignCalc spreadsheet (here, here and here) and Andrew Kesic's motor curve spreadsheet (here and here) for examples.¹

If you want the motors to share the load proportionally to their maximum output power, then you want to gear them so that the speed vs. torque curves of all motors are superimposed (or that at least they're as close as possible within the relevant range). If you want motors to share unequally at different loads,² then you can orient those motor curves arbitrarily.

¹ I have some incremental updates to these I've been meaning to complete and post. (Mostly updated motor lists for 2012, and minor formatting/functionality improvements.) One of these days....
² That's a pretty rare requirement. Maybe one motor is a can motor with a fan that has a minimum effective speed, and you are willing to load it more at high speed, in order reduce its load when high torque is demanded (at low speed when cooling is ineffective).

[pmangels17]

My apologies. I forgot for a moment that the MiniCIM and CIM were different RPMS. However, going from all MiniCIM to all CIM wouldn't be a problem. You should definitely use a slight reduction on the MiniCIM, though I am not positive on the exact ratio required.

[cgmv123]

Originally Posted by pmangels17

My apologies. I forgot for a moment that the MiniCIM and CIM were different RPMS. However, going from all MiniCIM to all CIM wouldn't be a problem. You should definitely use a slight reduction on the MiniCIM, though I am not positive on the exact ratio required.

1.1676082862523540489642184557439:1 or 620:531 or 62:53* or 6:5*

*Approximate

[42!]

Designing a gearbox is a wonderful engineering exercise. In fact it's one I use when I want to teach someone the basics of FRC design. To start off, it is never completely sure what motors we will be allowed next year, however the combination you listed would be legal for the 2013 game year. As far as gearbox design, the best tool in the world is jvn design calculator. http://www.chiefdelphi.com/media/papers/2755
If you're smart you'll most likely figure it out quickly. Among other things this table can show the expected robot speed based on your gearing motors and wheel size. The easiest way to do this is to pick a wheel size, the pick gears from either of these sites http://www.vexrobotics.com/vexpro/ge...pro-gears.html and http://wcproducts.net/gears-20-dp/
Both of these have high quality aluminium gears which work well in robot drive trains. As far as what final speed you should aim for. most single speed transmissions are are 12-14 fp/s. Once you've figured what gears you want you can use these tools
http://wcproducts.net/how-to-gears/
To figure out how far apart your gears have to be to mesh correctly. This will allow you to figure out where you need bearing holes on your gearbox plate. After this you would draw up your gearbox plates, shafts to hold the gears as they spin and import cads of the gears (these are available from both vex-pro and WCP in the cad section of their site). Then you simply constrain everything together. This is obviously a simplification of the process of gearbox design, but feel free to ask any questions on here if you don't understand something. Also one more link to a helpful article on the subject.
http://www.frc-designs.com/btd/roboticstutorial1.html
Happy cadding!

[1683cadder]

Is there a list of formulas? I know there are calculators, but I want to know formulas. I wouldn't make custom gears for FIRST but for BEST we sometimes make gears with jigsaws.

[EricH]

Originally Posted by 1683cadder

Is there a list of formulas? I know there are calculators, but I want to know formulas. I wouldn't make custom gears for FIRST but for BEST we sometimes make gears with jigsaws.

Ahh, geartrain formulas. A really nice resource that I use is my college-level engineering textbook on Machine Design. To understand the WHY of those formulas without calculus is roughly equivalent to being very lost without a map of the area. To use them without calculus, OTOH, is actually possible.

The good news is that if you're just calculating a gearbox ratio, you just need to know the intermediate ratios. Let's take a single-speed gearbox--oh, let's grab a pair of AM Stackerboxes just for grins. As simple as it gets... one input gear, one output gear. 3.57:1 ratio (from here on out expressed as 3.57/1). Take two in series, and you get (3.57/1) * (3.57/1)= 12.7449/1.

Now, that formula (ratio 1*ratio 2=gearbox ratio) is expandable to any number of intermediate stages.

What that ratio does: It multiplies torque at the expense of a divided speed. Using the stacked Stackerboxes with a single CIM, running unloaded, you end up with 5500 RPM/12.7449=431.5 RPM at the output--but your torque went up by the same factor. If you had a 1:X gearbox, where X>1, it would have a reverse effect, speeding up the motor at the expense of torque.

Most folks who design custom FRC gearboxes will space the holes using the gears' pitch diameter (where they mesh with other gears)--take the sum of the pitch radii, add about 0.003", and usually that works.


I could go into gear design itself, but that's pretty ugly, or gets that way after a little while. Shaft design might be more helpful, but that's one where I'd just use COTS shafts rather than deal with squares inside of square roots inside a cube root.


Going on to some of your other questions:

Sure, you can test the gearbox in CAD. Do I know exactly how? No, due to not playing around with the motion constraints and other similar stuff, but I'm willing to bet there's a tutorial or someone can help you.

Can you use it next season if you guys publish the CAD? I'll let you know on January 4, 2014, but under the 2013 rules, that would be a Gray Area governed by R16. The CAD would be publicly available, but the gearbox would not be. Whether CAD counts the same as software is up for debate. But if you were to make changes to the gearbox design, and build a new one, you'd be quite legal. (My personal opinion, not valid at inspection, is that it would not be legal without the modifications--second example in the blue box.)

[Ian Curtis]

Originally Posted by EricH

I could go into gear design itself, but that's pretty ugly, or gets that way after a little while. Shaft design might be more helpful, but that's one where I'd just use COTS shafts rather than deal with squares inside of square roots inside a cube root.

This kind of attitude is unhelpful. It does no good to say, "Look at this really complicated formula I know!" and then not even explain it. Furthermore, for 99% of FRC applications there is no need to do really crazy math, just a serious need to understand core concepts.

This thread is a really great place to start, and this post by Adam Heard is really good. As Adam suggests, I think you are much better off starting by doing a simple custom gearbox that consists of all off the shelf pieces except for the sideplates and working your way up. There is a huge wealth of information on Chiefdelphi, and don't be afraid to ask technical questions. 1 simple technical question is worth 100 "Is XXXX better than YYYY even though OPR says otherwise?!" threads.

Making gears is probably worth a separate thread. It has definitely been done. If you have access to some type of CNC machine you can waterjet/mill/route/lasercut larger gears that don't move quite so fast, but typically the smaller gears you'd find in a gearbox are hobbed, and that requires special tooling. It is much easier to call up a supplier, but if you dead set on making your own gears I'm sure someone on CD has done it.

Hope that helps get you started!