Mecanum Einstein this year

[Nick Lawrence]

1503 used 6X4" wheels, 1/8" drop. We used blue nitrile tread and didn't change tread once.

Our gearboxes were little innocent CIMple boxes with a secondary 12:18 sprocket reduction.

Fun fact about the CIMple boxes; if you didn't Loctite your motors in, they would work themselves loose after about two minutes of usage.

-Nick

[Chris is me]

It's also worth noting that it's somewhat easier to use bigger wheels for newer teams.

Personally, I dig small wheels. 4 inch wheels are a biiiig weight cut when you add all the parts up.

[BrendanB]

Originally Posted by apalrd

We (33) built a live-axle drive inspired by team 254. We made bearing blocks and cam tensioners out of 1/4" plate and cut them on a waterjet, but the bearing blocks were seriously just three holes in a line and not hard to make by hand. The shafts were lathed by hand (we used keyed shafts since they were easier to machine), and the wheels, hubs, and transmission components all came from Andymark. If you simplify the design, it isn't hard to make.

The Cheesy Poofs transmission, on the other hand, is quite difficult to replicate as basically all of the gears are custom-made by them.

33 robot stats (We were on the top 25, so that qualifies us?)
-8 wheel (6" wheels) live-axle DualDrive combining Plaction, KOP, and Lunacy wheels. Two wheels are actuated by pneumatic pistons, automatically.
-2-speed, 4-CIM transmission (5.5 and 12 ft/sec)

Any pics of your drivebase?

[Akash Rastogi]

Originally Posted by BrendanB

Any pics of your drivebase?

Jim Zondag also posted their CAD in CD media.

[Peter Matteson]

Originally Posted by Ether

What was your rationale for the small wheels?

I'm thinking 1) less gear reduction necessary and 2) lower center of gravity

are there other reasons?

Adam pretty much nailed it below.

Originally Posted by AdamHeard

The question I always ask the people at competitions who question are small wheels is, "Well, why do you use big wheels?".

Usually the answer is either they don't know, or they always have.

We use small wheels for a lot of reasons, but they mostly stem from two of ours teams primary design goals; less weight, and less friction.

Small wheels are physically smaller, which is less weight.
Small wheels require less torque to turn to achieve the same force on the ground, less force in the shafts/sprockets means smaller and lighter parts.
Small wheels need less reduction, which is both a direct decrease in weight, but also a decrease in friction losses as we can run less stages of gear reduction total.
Small wheels let you have a slightly longer wheelbase for all other factors the same.
Small wheels are cheaper for us to make, as it's a smaller diameter stock, and has much less wasted material.

Also, experimental data has shown that for rough top tread, smaller diameter wheels have more traction.

Historically we used the 4wd 2001-2005 (disclaimer: I haven't really looked at any of our robots older than this) with 6" pnuematic wheelchair wheels and 6" skyways in 2001 & 2003. In 2004 and 2005 we used 6" pnuematic and 6" omnis with a T-kats based 2 speed.

In 2005 we had a lot of drive issues, marking the carpet with our omnis and breaking KOP skyways when we switched over to them. That was the last year I let our team leader use the excuse "This is easy we've done it before." as a logical arguement in the decision of what drivetrain to use. In 2004 we had pushing power and the robot was designed to operate in the limited perimeter of the field. 2005 we were outgunned and spent the whole season fixing issues.

In 2006 we clean sheeted our drivetrain ideas and arived at going with a 6WD drop center with AM shifters an d 4" wheels. We made bearing blocks that bolted to a 1x1x1/16" box Al space frame. The performance was good for the way we played the game but we saw weaknesses in the design with the reliability, wheel alignment and maitenance. We used the brand new IFI wheels that year, which were good but did not reach the robustness of where they are now at that time.

In 2007 I designed the first parallel plate design that the Bobcats used as a way to keep the performance of the 2006 robot but improve the robustness of the 2006 design. This purpose of this was to improve mainenance, modularity, and reliability. In doing this we used AM shifters with custom output shafts and lost the outer steel plate from AM to shed weight and remove redundancy. We also made our own wheels because we wanted wide wheels in the corners(these can be seen elsewhere on CD). We never lost a chain, replaced a tread or had any issue with the drive. We still use this robot to practice although we probably need to replace bearings and motors at this point.

In 2008 we had to speed up the robot for the game, so we took the opportunity to improve packaging further and tighten everything up. At thispoint the drivetrain was mostly designed by another engineer on the team. We also narrowed our wheels because we did the math and realized the wider wheels didn't really do anything.

In 2009 we did our first live axle and cantilevered the wheels as well. We packaged the transmissions and all in a 2x2 box because we needed to package low and tight for our design that year. After this with the introduction of the AM hex bearings we hex broached everything.

Our 2011 drive was an evolution of the 2009 combined with our parallel plate system from previous years. We put the wheels and all inside the box this year combining the best of the 2008 and 2009 designs. This is what we planned to do for 2010 until we saw that game.

I hope this explains some of the thought behind how we got where we are as an example for other teams. I will openly admit we think there is another jump in highly mobile drive trains that teams are starting to go through right now and we think we have to put some work in to plan ahead for it.

[Cory]

Originally Posted by Chris is me

Do keep in mind though, the semifinals came entirely down to the minibot coin flip in match 1, and the finals were 2v3. While 254 / 111 / 973 was undoubtedly the better alliance, it was certainly not predetermined.

First, what Mike said.

Second, the real difference in Semi 1-1 was 254 missing an ubertube in autonomous, not the minibot (and given that 254/111/973 won in 2, there would still have been a second chance had 973 not taken first in the minibot race).

[Nick Lawrence]

Can we just agree that they outplayed us and leave it at that?

-Nick

[Karthik]

Originally Posted by Mike Soukup

I know that Karthik's term "minibot coin flip" has become a popular way to describe the end game this year. I'm not a big fan of Logomotion's end game, and I think the description is true, to a point. But, eventually people have to recognize and appreciate greatness. 973 had the fastest minibot on Galileo and potentially the fastest minibot at Championship. 254 chose them to guarantee 1st place in the minibot race every match.

From now on, if you use the term "minibot coin flip" to describe our matches, know that our alliance was using a coin with two heads, we called heads every match, and won every toss.

The "minibot coin flip" term is meant to refer to matches where the point differential on the rack became irrelevant, and was to be decided by relatively random nature of the minibots. However, there was nothing random about the minibot success of 973. They were consistently the fastest on Galileo. The minibot coin flip determined many matches throughout Logomotion, but it never really came into play for the World Champion alliance. They definitely didn't leave anything to luck.

[JesseK]

Originally Posted by apalrd

We (33) built a live-axle drive inspired by team 254. We made bearing blocks and cam tensioners out of 1/4" plate and cut them on a waterjet, but the bearing blocks were seriously just three holes in a line and not hard to make by hand. The shafts were lathed by hand (we used keyed shafts since they were easier to machine), and the wheels, hubs, and transmission components all came from Andymark. If you simplify the design, it isn't hard to make.

The Cheesy Poofs transmission, on the other hand, is quite difficult to replicate as basically all of the gears are custom-made by them.

It's more about their overall design rather than the WCD-sliding blocks or their custom gears (which I suspect *could* be COTS gears if they chose).

3.5" wheels -- they can't go smaller because of the large gear on the shifting shaft
They can't make the first stage reduction greater because of CIM spacing (2.55 or 2.6" between CIMs) and the internal shaft spacing of the large gear on the first stage. The AndyMark super shifter runs into this same issue.
So they're 'stuck' with 18 ft/s, in a sense. All in the name of removing the 3rd gearing stage from the transmission to increase efficiency.

Then
- For any sort of decent acceleration, they need incredible efficiency on the overall drive train since it's only 4 CIMs (unless there were other motors hiding under the CIMs...)
- Weighing only 100 lbs + battery, bumper, lightweight minibot helps acceleration too

So their strategy, at 18 ft/s for an 'open field' where defense rules were constricting and effective "anti-flow" strategies forbidden, was actually a great idea in hindsight. 7 ft/s would help them get through the average defensive robot in a pinch, though the tradeoff was lack of torque -- which did get them into trouble once.

I'll admit, I didn't count gear teeth, and a bit of this is reverse-engineered estimation; so I don't know the true numbers for their gearing itself. Yet if you design a 2-reduction 2-stage gearbox (I tried the week after champs) you'll see that it's not quite as easy as slapping COTS parts together. So most of us wouldn't be able to do it quite like they do.

[Chris is me]

Jesse - They could break 18FPS with bigger CIM pinions, however CIMs just so happen to take a massive acceleration hit right at the 18 FPS mark.

[Cory]

Originally Posted by JesseK

It's more about their overall design rather than the WCD-sliding blocks or their custom gears (which I suspect *could* be COTS gears if they chose).

All our gears are COTS in the sense that they're off the shelf products from either Martin or AndyMark.

This year we were forced to use more gears from Martin than we'd have liked and fewer from AndyMark. That required us to do a lot of machining to remove hubs, machine dog teeth, lightening pockets, and hex broach bores. In that regard the Martin gears are not truly COTS.

[AdamHeard]

Originally Posted by JesseK

It's more about their overall design rather than the WCD-sliding blocks or their custom gears (which I suspect *could* be COTS gears if they chose).

3.5" wheels -- they can't go smaller because of the large gear on the shifting shaft
They can't make the first stage reduction greater because of CIM spacing (2.55 or 2.6" between CIMs) and the internal shaft spacing of the large gear on the first stage. The AndyMark super shifter runs into this same issue.
So they're 'stuck' with 18 ft/s, in a sense. All in the name of removing the 3rd gearing stage from the transmission to increase efficiency.

Then
- For any sort of decent acceleration, they need incredible efficiency on the overall drive train since it's only 4 CIMs (unless there were other motors hiding under the CIMs...)
- Weighing only 100 lbs + battery, bumper, lightweight minibot helps acceleration too

So their strategy, at 18 ft/s for an 'open field' where defense rules were constricting and effective "anti-flow" strategies forbidden, was actually a great idea in hindsight. 7 ft/s would help them get through the average defensive robot in a pinch, though the tradeoff was lack of torque -- which did get them into trouble once.

I'll admit, I didn't count gear teeth, and a bit of this is reverse-engineered estimation; so I don't know the true numbers for their gearing itself. Yet if you design a 2-reduction 2-stage gearbox (I tried the week after champs) you'll see that it's not quite as easy as slapping COTS parts together. So most of us wouldn't be able to do it quite like they do.

Not to pick on you, but there are quite a few flawed assumptions here.

I can only speak for us, but I would assume 254 has the same reasoning.

We don't gear for top speed, but for a "sprint distance".

Low gear is well past traction limited. 254 was likely pushed in fm1 due to being underweight. There was no lack of torque. 99% of robots would not have been able to do that to them, 469 was really an edge case with their drivetrain.

Using the same geometry, you can easily gear to get a top speed in a smaller range (in 2010 we were 13.5 fps with LARGER wheels...). We're not stuck with 18fps.

[JesseK]

Originally Posted by AdamHeard

Not to pick on you, but there are quite a few flawed assumptions here.

I can only speak for us, but I would assume 254 has the same reasoning.

We don't gear for top speed, but for a "sprint distance".

Low gear is well past traction limited. 254 was likely pushed in fm1 due to being underweight. There was no lack of torque. 99% of robots would not have been able to do that to them, 469 was really an edge case with their drivetrain.

Using the same geometry, you can easily gear to get a top speed in a smaller range (in 2010 we were 13.5 fps with LARGER wheels...). We're not stuck with 18fps.

I get that 'sprint distance' concept; it's why I made my "gearing vs time" charts in my recent calculator. Other than that, a couple of things --

I was under the impression, from talking to kids on all 3 teams, that the drive train gearing between 254, 973, & 1868 were identical. Due to the collaborative nature of the machine shop setup (which is great, imo) I presumed many discussions and similar conclusions were made.

I'm having difficulty getting the geometry for less than 18ft/s correct by using 3.5" wheels, the 15:48 & 28:35 combos for the shifting stage, and 11:40 for the first stage; this is why the presumptions were made. It's easy to get less than 18ft/s after adding a 3rd stage, but that's extra weight/inefficiency. Increasing the 40T gear on the first stage interferes with the dog gear; decreasing the 11T pinion further to a 10T pinion without increasing the 40T gear causes the CIMs to touch (at least). For anyone wishing to use the COTS solutions from AndyMark without extra milling on the gears for the dog gear setup (such as my team), this is what we'd have to use.

So really, I have difficulty in understanding how you get the dimensions needed for 13.5 ft/s in the same geometry (2 stages of reduction); it doesn't work out for me.

[AdamHeard]

Although functionally the same, our drivetrains are very different at the part level. 254 makes no parts for us, and we make no parts for them. We're certainly buddies with 254 (and they/968/60 certainly inspired our design style), but we each run our own show.

What I meant by the same geometry, was the same overall gear setup. We don't use AM's stock shifter setup. We get dog gears custom cut each year per whatever ratio we need. It's a pretty simple operation for a CNC, so sponsors are willing to do it.

I also left this out of my previous reply, but it's possible to make this style drivetrain with zero CNC equipment (more parts on our current drivetrain are made on manuals than you would think); 973 did that from 2005-2008 (single speed through 2007, SuperShifters in 2008).

Originally Posted by JesseK

I get that 'sprint distance' concept; it's why I made my "gearing vs time" charts in my recent calculator. Other than that, a couple of things --

I was under the impression, from talking to kids on all 3 teams, that the drive train gearing between 254, 973, & 1868 were identical. Due to the collaborative nature of the machine shop setup (which is great, imo) I presumed many discussions and similar conclusions were made.

I'm having difficulty getting the geometry for less than 18ft/s correct by using 3.5" wheels, the 15:48 & 28:35 combos for the shifting stage, and 11:40 for the first stage; this is why the presumptions were made. It's easy to get less than 18ft/s after adding a 3rd stage, but that's extra weight/inefficiency. Increasing the 40T gear on the first stage interferes with the dog gear; decreasing the 11T pinion further to a 10T pinion without increasing the 40T gear causes the CIMs to touch (at least). For anyone wishing to use the COTS solutions from AndyMark without extra milling on the gears for the dog gear setup (such as my team), this is what we'd have to use.

So really, I have difficulty in understanding how you get the dimensions needed for 13.5 ft/s in the same geometry (2 stages of reduction); it doesn't work out for me.