6 CIM Drivetrains... What's Your Experience?

[thefro526]

We ran a 6 CIM Drivetrain in 2013, and other than a couple of hiccups, it worked really well for us.

We were geared a bit on the higher side of things, 6:1 for an effective max speed of ~12.5-13fps. Compared to a 4 CIM Drive with Similar Gearing or a 4 CIM 2-Speed Drive, it seemed like our acceleration performance and/or general "Zippy-ness" of the drive was on par with most of the front-runners in the field.

We did experience current draw problems a handful of times over the course of the season, most of which were triggered by defense being played by, or on us. It's also worth noting that we had 5 other motors on our robot, (4) of which were RS550's, and one of which was an AM-9015, so if any of them were running (especially being pulsated or stalled) it didn't exactly help the problem. In either case, even though we thought that the drive should have popped the PD Breakers (40A) prior to popping the 120A main, this was rarely, if ever, the case.

After the season, we did some digging into the problem, and found that the cause was due to a misunderstanding (or assumption) made about the PD and Main Breakers. As Mark posted above, the 40A Snap action breakers are capable of handling some amount of current in excess of 40A for some finite period of time - which basically means that a drive can draw more than enough current to pop the 120A Main breaker well before any of the 40A PD Breakers pop.

If you reference the data sheets for the 120 (http://www.waytekwire.com/datasheet/...UITBREAKER.pdf) and the 40A Snap Action breaker (http://www.snapaction.net/pdf/MX5%20Spec%20Sheet.pdf) you'll be able to fairly accurately estimate when your breakers will trip, if ever, during a match. In the case of an aggressively geared robot (like ours) the drive motors can pull somewhere between 60 and 70a each, for a total of between 360A and 420A. If you reference the curves for the snap action breakers, you'll see that at 150% rated current (60A for a 40A) the breaker can last between 3.9 and 47 seconds, but the Main (loaded at 300%) should pop in 5 to 15 seconds. The numbers are even worse at 420A, the 40A PD Breakers will only hold for 2.2 - 9.2 seconds, and the main will last for 2.5 - 8 seconds. It's also worth nothing that these calculations don't take any other current draws into account, so if were running a 6 CIM Drive and had some sort of RS550 powered load pulsating (stop/start) or a compressor kick on, you'd more than likely spend the rest of the match with a robot turned off.

The easiest way to get around the current draw problem seems to be to gear the drive more conservatively, use a wheel with a lower CoF, or to avoid pushing matches... But none of those are really that 'fun', so the other method would be/could be to use a 2-speed transmission along with a 6 Motor Setup. If you were to gear for ~40A Current Draw at maximum load in low gear, you should see very few problems with extreme current draw - if anything, you may have issues with voltage drops and minor brown outs - but those are still better than looking at your robot while it sits dead on the field.

FWIW, circuit breakers are also very, very susceptible to temperature, so keep that in mind. If the venue is abnormally hot, or abnormally cold, or your robot is still warm from the previous match, the breaker performance can change. More often than not, a hot breaker will be able to hold less current, and a cold breaker is able to hold more. As we monitored breaker performance throughout the off season, I noticed that the main breaker was 'warm' to the touch after the majority of our elimination matches.... Anyway, should you decide that monitoring the temperature of the breakers is something you'd like to do, it's best done with proper supervision and safety equipment (Always) - and non-contact tools (think IR Thermometer) when possible.

[MichaelBick]

We didn't use a 6 CIM drive, but did use a 6 motor drive(4 CIM 2 550). Regardless, we ran into a couple problems. We were geared really high(19fps at 100% efficiency) with high traction wheels(2" wide vex traction). We never blew the main breaker in a pushing match, but going from full speed forward to full speed backwards bl;ew it a couple times in practice and competition. We ended up inserting a 1/8 second ramp up on big velocity changes like this and never encountered the problem again.

[philso]

Originally Posted by MichaelBick

We ended up inserting a 1/8 second ramp up on big velocity changes like this and never encountered the problem again.

I think half or more of the 3-phase motor drives we build at work are used to reduce or eliminate the starting surge of the motor like you have with your ramp up. It also reduces the mechanical stresses in your drivetrain.

[Jonathan Norris]

Originally Posted by Nathan Streeter

So far, I only know of 254 using 6 CIMs in 2013 with their 2-speed+PTO gearbox and of 610 using them this year in their 1-speed+PTO gearbox. I heard though that when 610 set aside their climber mechanism they also removed the 2 "extra" CIMs... according to their website they designed for a top speed of about 9fps.

Any recommendations, cautions, or further information?

I wasn't directly involved in designing the drivetrain last year, but the whole thinking around the drive system was to minimize the time it took to get from the loading station to the pyramid. Surprisingly when you crunch the numbers it doesn't mean building a 2-speed gearbox with a 22ft/s high gear, actually a single speed drive system with 6 CIMs geared at ~10-11ft/s is the fastest way to get from the loading station to the pyramid.

The main reason for this is acceleration, having 6 CIMs geared at a 'slower' top speed gave us incredible acceleration, actually allowing us to reach that top speed quickly. It also allowed our driver to be more creative and precise trying to get around defence, our drive team figured out quickly that we could accelerate away from anyone. So when teams tried to play D on us, it would almost seem like they just bounced off us, because when we got into a collision we could spin our robot around them and get away quickly. I believe the drive system was one of our keys to success last year.

I highly recommend teams analyze the game and use that analysis to figure out your optimal drive system. 610 has been doing this analysis for years, there is no 'perfect' drive system that works every year, the best drive system is the one thats optimized for your team's game strategy.

(pro-hint: the acceleration difference between a 6 CIM and 4 CIM drive geared at 10ft/s isn't that much different. Start thinking more about acceleration/agility/control vs. top speed, its a real tradeoff, you need an amazing driver to control a robot above ~15ft/s)

[Ether]

Originally Posted by Jonathan Norris

Surprisingly when you crunch the numbers...

Could you please elaborate a little on what method you used to crunch the numbers?

the acceleration difference between a 6 CIM and 4 CIM drive geared at 10ft/s isn't that much different.

I'm assuming you are referring to acceleration at low speeds, where you are traction limited, correct? Once you going fast enough to be torque limited, there should be a significant difference in acceleration.

[gpetilli]

Originally Posted by Jonathan Norris

(pro-hint: the acceleration difference between a 6 CIM and 4 CIM drive geared at 10ft/s isn't that much different. Start thinking more about acceleration/agility/control vs. top speed, its a real tradeoff, you need an amazing driver to control a robot above ~15ft/s)

I totally agree. I don't understand why some teams that are going to 6 CIMs AND a two (or three) speed gearbox.

In order to take advantage of the extra torque, do you need some form of traction control? Are you using a PID loop to control the voltage ramp? Are you imposing an open loop ramp on the driver or motor commands? Do you monitor motor currents for stall? Do you simply let the wheels spin and degenerate to the kinetic CoF for a small time?

[AdamHeard]

Originally Posted by gpetilli

I totally agree. I don't understand why some teams that are going to 6 CIMs AND a two (or three) speed gearbox.

In order to take advantage of the extra torque, do you need some form of traction control? Are you using a PID loop to control the voltage ramp? Are you imposing an open loop ramp on the driver or motor commands? Do you monitor motor currents for stall? Do you simply let the wheels spin and degenerate to the kinetic CoF for a small time?

Mathematically probable increases in acceleration and usable speed (even without any fancy effects). Is the increase worth the cost to most teams? Probably not. Certainly there are some teams where it is worth the cost however.

[JamesTerm]

Originally Posted by SoftwareBug2.0

We didn't have any delays to spread out current peaks. We used six CIMs and each of our wheels was powered by two CIMs, so we had three wheels. Our drivetrain would have just been very hard to control if we had turned all the motors on full at once because it would have wanted to spin very quickly.

You said... would have been hard to control... so how did you solve it to where it didn't turn all the motors on full (or where the wheels didn't spin) so quickly? Was this a mechanical or software solution? If software... please elaborate... thanks.

BTW... this is the first video I've seen where 1114 successfully climbs... thanks for including it... I've been wanting to see that for quite some time.

Originally Posted by Jonathan Norris

(pro-hint: the acceleration difference between a 6 CIM and 4 CIM drive geared at 10ft/s isn't that much different. Start thinking more about acceleration/agility/control vs. top speed, its a real tradeoff, you need an amazing driver to control a robot above ~15ft/s)

Do the extra motors help with control? I never thought of that as an effect of a 6wd vs a 4wd.

[Chris is me]

Originally Posted by Andrew Lawrence

Do the extra motors help with control? I never thought of that as an effect of a 6wd vs a 4wd.

He's implying a lower top speed helps with control. If a 10 FPS 4 CIM drivetrain gets you there as fast as a 15 FPS drive, then you gain controllability by going down in speed.

Originally Posted by Chris is me

He's implying a lower top speed helps with control. If a 10 FPS 4 CIM drivetrain gets you there as fast as a 15 FPS drive, then you gain controllability by going down in speed.

Ah, that makes sense.

[SoftwareBug2.0]

Originally Posted by JamesTerm

You said... would have been hard to control... so how did you solve it to where it didn't turn all the motors on full (or where the wheels didn't spin) so quickly? Was this a mechanical or software solution? If software... please elaborate... thanks.

BTW... this is the first video I've seen where 1114 successfully climbs... thanks for including it... I've been wanting to see that for quite some time.

Having all six motors on for even one second would probably make you spin faster than you would want to. If you build a three-wheel holonomic drive you will find very quickly that it likes to turn. We had a gyro-based feedback loop keeping us from turning too much.

[MichaelBick]

Originally Posted by SoftwareBug2.0

Having all six motors on for even one second would probably make you spin faster than you would want to. If you build a three-wheel holonomic drive you will find very quickly that it likes to turn. We had a gyro-based feedback loop keeping us from turning too much.

Do you really have any data to back this up? 254, 610, 973 among others(including us) all had 6 motors and were extremely controllable.

[ErvinI]

Originally Posted by MichaelBick

Do you really have any data to back this up? 254, 610, 973 among others(including us) all had 6 motors and were extremely controllable.

I believe he's talking about a three-wheel holonomic, not a skid-steer.

[JamesTerm]

Originally Posted by SoftwareBug2.0

Having all six motors on for even one second would probably make you spin faster than you would want to. If you build a three-wheel holonomic drive you will find very quickly that it likes to turn. We had a gyro-based feedback loop keeping us from turning too much.

Ah ok, thats what I thought you did (or something along these lines)... did you make some trapezoidal motion profiling on the anguar velocity to do this? Or was it a fixed angular velocity?