pic: Schreiber Take on West Coast Drive

[Chris is me]

The funny thing here is that while you're preaching to us how doing the math is better than a pretty render, I think your drivetrain won't perform adequately with the gearing you've outlined. 10 fps with 4 CIMs isn't the best acceleration, and you are nowhere close to traction limited. Pushing matches will probably trip your breaker faster than I would be comfortable with. Your gearing would probably eat through batteries faster than I would want as well. You could increase your reduction a bit to become traction limited at much more feasible currents, while increasing your acceleration enough to virtually make up for the speed loss in a Breakaway like field.

This is assuming you bring your robot to full weight. At the 27 pounds pictured, I bet you would have zero problems.

[MrForbes]

Chris, the two speed sketch is on the reverse side of the napkin, you can't see it

I like to move straight from NCAD to MAC machining (manual analog control)

[JamesCH95]

Good stuff. An excellent ME/Designer that used to be my mentor said that if I couldn't design it on a napkin then I couldn't design it at all.

[Andrew Schreiber]

Originally Posted by JamesCH95

Good stuff. An excellent ME/Designer that used to be my mentor said that if I couldn't design it on a napkin then I couldn't design it at all.

Couldn't or Shouldn't? I know a lot of guys who are of the opinion that if you can't explain the idea using a napkin and a pen you are probably over complicating it.


Specifically, what part of my math did I do wrong? (Remember folks, I am a programmer, I do this in my spare time) I worked out what the speed of the motor at 40 amps was and based the speed off that. Is a 5.97:1 reduction too little for a 4" wheel?

[JamesCH95]

Originally Posted by Andrew Schreiber

Couldn't or Shouldn't? I know a lot of guys who are of the opinion that if you can't explain the idea using a napkin and a pen you are probably over complicating it.

Well I suppose one probably could still design it, but like you said, it's probably not an elegant solution.

Originally Posted by Andrew Schreiber

Specifically, what part of my math did I do wrong? (Remember folks, I am a programmer, I do this in my spare time) I worked out what the speed of the motor at 40 amps was and based the speed off that. Is a 5.97:1 reduction too little for a 4" wheel?

I think your powertrain calcs are probably fine, and I wouldn't worry too much about acceleration. My freshman year (2001) we ran 2 Bosch drill motors geared for 10ft/s and it worked wonderfully, 4 cims should have plenty of torque to get you by. You probably won't win a pushing match against a hardcore pushing robot, but that's what shifting transmissions are for if you chose to use them.

[Andrew Schreiber]

Originally Posted by JamesCH95

I think your powertrain calcs are probably fine, and I wouldn't worry too much about acceleration. My freshman year (2001) we ran 2 Bosch drill motors geared for 10ft/s and it worked wonderfully, 4 cims should have plenty of torque to get you by. You probably won't win a pushing match against a hardcore pushing robot, but that's what shifting transmissions are for if you chose to use them.

Yeah, looking into it, interfacing a Super Shifter in there with a 45:20 gear set would be feasible. I may have to buy an extended output shaft. The ratios wouldn't be optimal though. By tweaking the way I drive the center wheel I could also use the Gen 1 shifters which offer slightly better ratios. But that would get pricey real quick.

[JamesCH95]

What is "optimal" for this robot? IMHO 5-12 ft/s is is a very generous range. Any faster and an operator can't control it too well, any slower and virtually anyone will be able to dodge you. Just a thought.

[Andrew Schreiber]

Originally Posted by JamesCH95

What is "optimal" for this robot? IMHO 5-12 ft/s is is a very generous range. Any faster and an operator can't control it too well, any slower and virtually anyone will be able to dodge you. Just a thought.

That would depend on the game, obviously if it were a game like last year I would say 8fps. For a game like 2008? 10fps seems more than fast enough for me. There are games that I would prefer more oomph (2007) and would settle for 6fps (For reference, in 2007 the team I was on never crossed past midfield) This is one of the major reasons I like AM products, they have a wide range of options that do not change the mounting/interface options. Heck, if I wanted to switch to 6" wheels I could do so relatively easily with a simple swap of gears in the transmission.

[JamesCH95]

So if I'm understanding you right 5-12fps would cover you, with a reasonable margin, for any game within recent memory I like where your design is going

Just a little thing I gambled on this year that paid off: keep your CG behind the bumper zone, not as low as possible (assuming the bumper zone isn't insanely high, of course). 96.3% of the time robot-robot contact is bumper-to-bumper and having your CG/center of inertia in the bumper zone will help keep your robot from flipping. I'll let you do the fizix.

[Jared Russell]

Originally Posted by Andrew Schreiber

Specifically, what part of my math did I do wrong? (Remember folks, I am a programmer, I do this in my spare time) I worked out what the speed of the motor at 40 amps was and based the speed off that. Is a 5.97:1 reduction too little for a 4" wheel?

Here's how I would design a two-speed transmission. I always try to have specific goals for my low and high gearing to "optimize" the way that I'm using the motors' mechanical power.

LOW GEAR ANALYSIS:
Goal: Be traction limited at 40A per motor
Rationale: This will let us push as hard as we can without ever popping breakers.

Assuming the load is something like 145 lbs (robot+battery+bumpers) and you are using roughtop with a static CoF ~= 1.3, the maximum amount of force that you can transfer to the ground is about 190lbs.

Assuming 4 drive motors (CIMs), and 4 inch wheels, each motor can contribute up to (190 lbs) / (4 motors) * (2 inch wheel radius) = 95 in-lb of torque before you slip your wheels.

Stall torque of a CIM is about 21.5 in-lb (at 133A). Torque at 40A is more like 6.2 in-lb. And then there's gear-train inefficiency - let's say you have a good spur gearbox for a total efficiency of 90%. Now your torque is 5.5 in-lb per motor.

So if you want to be able to push ~190lbs across the floor all the live-long day, a reduction like 95/5.5 = 17.2:1 would get the job done. When not pushing, you would expect to move at a clip of about 4.5-5 fps (I generally find that 80-85% of the "free speed" of the motor, divided by the gear ratio, gives a good top speed estimate. It is never exactly the free speed because of drag in the gear train).

HIGH GEAR ANALYSIS:
Goal: Be traction limited at 75% stall (100A per motor)
Rationale: We will be able to turn in the upper 25% of the motor power curve (since we need to skid our wheels in order to turn in a tank drive). At the same time, our max speed will be very high. We don't worry about the 40A breakers, because they won't trip unless there is a constant (> a couple seconds) heavy draw, and we will downshift if there is a need for that...

Assume all the same things as before...4 CIMs outputting 21.5 oz-in of torque and 190 lbs of total load. But now at 75% stall and 90% efficiency, we expect ~14 oz-in of torque instead of 5.5 as before. 95/14 ~= 6.8:1. We expect a top speed of about 11-12 fps.

Obviously, the goals at each design point will vary widely depending on what you are trying to do - these are only two possible criteria. For a single speed drive, you'd need to trade off between popping breakers/being traction limited and having a good degree of speed and agility.

Likewise, the missing "last step" that I would always go through would be to compare my desired ratios with those that I can easily achieve by utilizing COTS components and my teams' manufacturing capabilities. For example, a SuperShifter (with the last stage removed) can achieve ratios of 10.67:1 and 4.17:1. Coupled with a chain reduction of about 1.6:1, I get 17:1 and 6.7:1, which are pretty close to my theoretical goals using 4 inch wheels.

[JesseK]

Andrew, I don't know a single employer who would take a napkin sketch and arbitrary numbers over a full CAD given a choice between the two. It's a shame that you're mocking everyone who requests feedback on various frame designs while also learning.

CAD is never a crutch; it's simply one of the many pieces to the equation of a high quality product.

[AdamHeard]

Originally Posted by JesseK

Andrew, I don't know a single employer who would take a napkin sketch and arbitrary numbers over a full CAD given a choice between the two. It's a shame that you're mocking everyone who requests feedback on various frame designs while also learning.

CAD is never a crutch; it's simply one of the many pieces to the equation of a high quality product.

There are a good amount of people who post pretty renders of designs that aren't even fully functional in my opinion, yet they spent the time to make a pretty render and CAD arbitrary details.

I agree with Andrew, you should be able to plan what actually is going to do the business (and all the details of it) in your head before you start making pretty CAD.

I don't think Andrew intends to go into season with a napkin sketch and machine from there. A nice worked out concept will save much time when it comes to the detailed design in CAD.

973 works the same way; we CAD 100% and machine 100% to print, but almost every system is sketched out by hand, on a whiteboard, made from paper, etc. before we begin designing. We'll often work out ideal ratios, sizes and stages of reduction before CAD even begins; it allows much more efficient design.

[Jared Russell]

Originally Posted by JamesCH95

Just a little thing I gambled on this year that paid off: keep your CG behind the bumper zone, not as low as possible (assuming the bumper zone isn't insanely high, of course). 96.3% of the time robot-robot contact is bumper-to-bumper and having your CG/center of inertia in the bumper zone will help keep your robot from flipping. I'll let you do the fizix.

I am struggling with why having a high CoG would prevent you from flipping in the event of a collision.

If anything, lowering the traction of your wheels would seem to do the most good since it would prevent your entire robot from becoming a lever arm rooted in the ground.

[JamesCH95]

I think the point is that if a designer/engineer has a very clear idea of what an elegant solution is, could be, they should be able to explain or convey the needed information with simple sketches, and then spend a minimum amount of time detailing the design in a CAD program. The N-CAD skill can be invaluable and it seems as if many students are missing it.

As an example: 2 months after I was hired I was responsible for designing a number of replacement parts for a high-g-load testing machine that was not working properly. This machine was too dangerous to run in our main engineering building, so a technician and I were over an hour away from any CAD-capable computer, but only 10 minutes from a machine shop when we were debugging it. My employer was VERY happy for me to make hand sketches of parts to give to the machine shop because saved time and money on the project as the deadline loomed nearer (not to mention many miles on my car!)

[JesseK]

If you re-read Andrew's post as if you were brand new to these forums, you may find that (specifically) the final paragraph is incredibly off-putting to someone trying to do a CAD by themselves for the first time, not knowing what they're really supposed to know in order to do things correctly. Andrew has exemplified the common CD knowledge paradox. Sure, all of the recent CAD prototypes are 'pretty', and most of them are missing details, yet the negative implications of Andrew's tone are as common as the missing details themselves.