Re: Why not to shift?
 

Why we haven't shifted in the past:

1. Adds complexity to the design
2. Adds weight to the design
3. Servo shifters can be unpredictable
4. We often don't have pneumatics on the robot already; adding the compressor for one purpose is a poor use of weight/real estate
5. Adds complexity to the programming
6. Adds cost
7. In some games, we don't see ourselves driving far and fast often enough to justify shifting mechanisms
8. Physical barriers on the field reduce effectiveness of shifters
9. We had a non-shifting drivebase done in a matter of days - we could use this for testing, practice, etc.

Re: Why not to shift?
 

In the last 2 years, having a single speed transmission allowed us to actually complete our robot before ship date and under weight. This is HUGE. Do not underestimate this. We've had our best-ever on-field success in the last 2 years because of this. Go listen to Karthik's presentation. Go now! Then come back.

As the coach on the field, at no point in the last 2 years did I WISH I had a shifting transmission to power through a given situation. Sure, it may have helped inherently ease one or two situations on the bridges in 2012, yet the issue was more about the bridge lowerer than the control getting up.

Thus, unless we're climbing some incline like 2010 (and even then, only if climbing the incline provides enough reward) or we're forced to traverse a wide open field like 2008, I doubt we'll do a shifting transmission in 2013.

Re: Why not to shift?
 

THIS is a key point. In 2001 we found that a low gear allowed the driver to better align the robot for hanging tubes. In 2012 it allowed the driver to shift into a lower gear and

1) climb onto the bridge with more control

2) Play defense and push an opponent

One other thing we've always done when shifting is to definitely go for the air shifter as long as pneumatics are on the machine for another reason. The servo shifters do work but we feel they're not as good as 'shifting on the fly' as pneumatics.

As far as the driver getting used to it? Driving practice is a HUGE help.

Re: Why not to shift?
 

I hope everyone posting on this thread understands the diferamce between power, torque, and pushing force. When a robot is in low gear, it generates moretorque at a given power, which resuts in more robot pushing force.

Re: Why not to shift?
 

This is straight our of 67's 2012 Engineering Design log. Now, this is beyond my level of mechanical comprehension to really understand what's going on. But the jist of it is 67 used a window motor instead of a servo to shift their two-speed gearboxes. Because the window motor is more powerful, it allowed for shifting on the fly, as opposed to a delay of seconds when using servos.

For those who are opposing shifting when pneumatics aren't being used elsewhere on the robot. It's an interesting idea. If you're willing to give up the motors for it.

Re: Why not to shift?
 

All though this is off topic slightly. Our team has triggers and buttons on our controls that are slow mode and turbo mode. No physical shifting required I know it is not the same thing but I am curious as to the benefits between them programming method and the actual shifting method.

Re: Why not to shift?
 

Shifting gives you an actual increase in pushing power. This means that it is useful for getting around defense in addition to having more controllability.

Re: Why not to shift?
 

If I understand what you're saying right, 1511 doesn't shift, rather, change the speed at which the motors run?

I gave heavy consideration to doing this to help line up for shots at Ruckus. Not much time to do so, I'm afraid. This method does offer controlled movements and speed when necessary. For teams looking for those things, it's a good alternative to shifting. Obviously it won't offer much in terms of advanced power. But a good alternative nevertheless.

As to the topic of the thread, "Why not to shift?", you wouldn't shift if you required controlled movements and high speeds, but not a change of power in the drive system, I suppose.

Re: Why not to shift?
 

I agree that the pushing power is different but both slow modes provide control and precision.

Re: Why not to shift?
 

Shifting reduces the torque demand on the motors, which brings the motor's top speed and loaded speed closer together. That means a smaller jump in speed when you go from heavily loaded to lightly loaded, which means better control in many situations, such as making small rotation adjustments on a sticky 6-wheel drivetrain.

Simply scaling in programming is never going to achieve this because the problem is in the dynamics of the electromechanical system. Speed or position feedback control of the drivetrain could help with this, if it's tuned well enough, however.

If anyone's interested, I'll add a post with all the math explaining why this is so.

Re: Why not to shift?
 

To add to that:

Power = torque * speed
Power is work / time
Work is force * distance
So, torque * rotations / seconds = power

Acceleration requires torque. To accelerate faster, you need more torque (unless you're traction limited, then you need more controlled torque output or less torque).

For DC motors, since the power curves are linear (or extremely close), peak power usually occurs at 1/2 speed (which is also 1/2 torque), so peak power is indicated as (1/4)(torque)(speed) if you convert units correctly. You have 0 power at stall torque or free speed. This point splits the 'good side' and the 'bad side' of the power curve. Motors are very very unhappy with the bad side of the curve (lower speed, less torque). Don't design to be on the bad side for too long.

For DC motors, speed is inversely proportional to torque and proportional to voltage. That means, at free speed, you are producing no torque. Since you always require torque due to friction losses, you will never be at free speed.

Torque is also proportional to current, and motor heat output.

Voltage = current * resistance. The entire electrical system can be represented as a resistor, so you can estimate voltage drop based on motor current.

A gear reduction adjusts speed and torque while maintaining power. If you cut speed by 1/3, you get 3x torque and power remains the same (neglecting the slight efficiency loss)

If you gear for too high of a speed, you will see several issues:
-Increased torque at the motor during acceleration increases current draw, draining the battery faster, possibly tripping breakers, and increasing voltage drop in the electrical system.
-Running too long on the 'bad side' of the power curve could thermally damage the motor. Some motors are more notorious for letting out the smoke than others.
-You will take longer to accelerate if you gear higher than a certain point, as you will be torque limited by the motor at a lower speed. Essentially, you need torque to accelerate, if you don't have enough, it will take longer to accelerate to the same speed (even if your top speed is higher)

In a car, shifting is done for several reasons:
-Engine power curves are very nonlinear, they usually want to stay in the efficient operating region and sometimes in the peak power or peak torque region
-Acceleration is faster if you're shifting at the optimal points. Imagine graphing the wheel torque vs speed in all gears on the same graph. There are intersections where it becomes more efficient to shift as torque drops off. If you shift at the right place, you will stay in the most efficient place. A CVT would always stay at the optimal place, transmissions with many gears also shift often, but it's a tradeoff between staying in the optimal operating region and shifting a lot.

In a FRC robot, where shifting is generally manual, shifting is done for slightly different reasons:
-To optimize pushing power, to become traction limited at a point where you can't trip the breakers in match play
-To optimize speed so you can still drive fast during the rest of the game

You could alternatively optimize two gears to accelerate faster, if you shift at the correct points.

With so much power in the motors now, it's getting more and more reasonable to design around a moderately high single speed (10fps-11fps) rather than use a 2-speed of ~5fps and ~13fps and shifting to accelerate. There's always something to be said for throwing more power at the problem.

Re: Why not to shift?
 

The danger is calling it "pushing power" when what one really intends is "pushing force" or "pushing ability". Most people will understand what the intention is, but some people will get power and force mixed up if they're not familiar with the difference between the two. I think that's what DampRobot was getting at.

/semantic threadjack

OP: all the main points have been hit: shifters add weight, complexity, cost, and are not always utilized enough to offset the first three drawbacks.

We're leaning towards going back to a single speed drive this year, mostly for a cost and timeliness concern, but also because we didn't use the shifters much in the last two years.

Re: Why not to shift?
 

This is neat feature we use and I know others do as well.

In 2010 IIRC we had three buttons that limited power to the drive motors. I think (1/2, 1/3, 1/4). It was a great feature for picking up balls of the wall and corners and to cut the power while turning. Add these buttons to a two speed transmission and our drivers had six different speeds to play with on the field!

In 2012 we had a 50%-75% (I don't remember the exact number one of our programmers played with the power setting to get a good balance) power button. This was great for lining up on the key, climbing the bridge either as a solo balance or the first bot on, and turning while in high gear. So technically we were a four speed transmission. :cool:

Similar to your team, you can use a single speed transmission and run constantly with a power of say 85%-90% and have a button for a low speed and a turbo button to give you an extra boost of speed to act like a high gear. Before pursuing this route, do the math!

As others have mentioned it doesn't give you an edge with pushing power but gives your drivers flexibility to pull off moves you can't do with a single speed. In 2011 if a low speed was all you wanted for placing tubes and lining up your minibot a wiser decision would be to have a programmed lower speed rather using shifters because you don't need the extra torque. This also allows you to keep a lighter robot which will equal faster speeds and less torque required to move!

For many teams, shifting is the answer but in most cases you can accomplish what you want with lighter, reliable single speeds and some basic programming additions!

Re: Why not to shift?
 

We preferred to line up for shots using a low gear vs a software-slowed high gear. Since we run an 18-20+ fps high gear, long acceleration times made precise movement difficult for our drivers.

Re: Why not to shift?
 

Understand the difference between "top speed" and "time to destination". A single speed robot geared for a top speed of 6 to 9 fps will get you from where you are to where you need to be very nicely, and still spin the tires in a shoving match. How often to you cross the entire field without stopping or turning? How many times to you accelerate from a dead stop? How much time & driver focus is spent shifting?

We have never used a shifting transmission. I won't say we never will, but I haven't seen a game (even Overdrive) where I regretted not having one.