Transmissions

[Tristan Lall]

Originally Posted by sanddrag

Many teams do this and I can't figure out why you would ever want to limit your power. Why not just give it all from the start? Yes a "tubo boost" may give a nice surprise when pressed but during normal times why in the world would you use the program to limit the capabilites of the mechanism? To me that seems foolish.

Back in 2000, Woburn's Blizzard 1 had just such a control. But it wasn't included to surprise people--there was a bit more pragmatism involved! As it happened, that robot had front-wheel-drive (old Johnson drills) and rear casters, and because the weight distribution was very biased toward the rear, the robot had an annoying tendency to over-rotate when trying to manoeuvre. That button was in there so that full speed could be achieved if necessary (and usually with a fixed rear wheel deployed, raising the robot off of its casters), but under normal conditions, the robot wouldn't be so difficult to control. I'm not so sure it was of much actual use, though, since that robot was by far the worst-driving of any of the Woburn family (of those that I've taken for a spin, at least--that's the last six, plus one of the ancient Canada FIRST robots). It certainly wasn't an efficient or elegant solution, as sanddrag pointed out.

This sort of solution is only useful if you don't care about the power output (in 2000, we knew we had enough power to drive around at half-throttle), or conversely, don't care about the current and efficiency (we should have cared--we threw 30 A breakers on some occasions, that year, and were limited to a 60 A main fuse, not a 120 A main breaker). You're effectively decreasing the voltage of the motors--for instance, by half--which just doubles the current needed to sustain a certain power level. This is why it is fundamentally crazy to control something like a (DC permanent magnet brush) motor at low throttle settings (i.e. the Victor's output voltage is a fraction of the motor's rated quantity); you're just going to force the current way up, and the efficiency way down. Low efficiency just means heat losses, and to compound the problem, the internal fans don't work well at low speed.

So what's the right way? Shift gears. You'll have two speed ranges that allow the driver to avoid overstressing the motors by running them too slowly.

Of course, sanddrag already knows all of this, so I suppose that this is for everyone else out there....

[sanddrag]

We made a 6 motor drive system with two speeds. I believe 188 Woburn did something similar in years past but I believe we are the only ones in the country with a 6 motor two speed this year. It's shifting mechanism is based roughly on the 2003 Technokats design. Everything has been beefed up (1/2" wide gears, 5/8" hex shaft, 1/4" sideplates) for the power of six motors. We could have gotten away with making it lighter and a little smaller but since this was our first year doing 6 motors or shifting, we decided to not take any chances. Since this worked with no problems whatsoever, we will be going smaller and lighter next year.

One nice thing is that al the reduction is done in the box to directly drive a 5.5 inch wheel. This is nice so you don't have to have the little sprocket - big sprocket thing going on.

Unfortunately, the way this year's game turned out to be played was quite different from what we imagined. We never really got to exhibit the drivetrain's full potential. But maybe that's just because the threat of it made opponents run away. In a game like 2002's Zone Zeal or 2003's stack attack, this gearbox would have been totally awesome.

While it was cool to have potentially the most powerful drive system in the nation, next year two of those 6 motors will probably be put to better use into an arm or ball scooper or something.

And just as a final note, having 6 motors is advantageous if your drivers can use them. We are geared for 15.5 fps (no load) in high gear and can still push a robot sideways (in high gear) without tripping any breakers.

[Max Lobovsky]

A software filter for the drive train is not a bad thing in itself, but a turbo button is not really the best solution. There is enough resolution in human movement of joysticks to be able to control any robot I have seen (We have a omniwheel holonomic drive with no feedback. It is likely one of the most difficult to control FIRST drivetrains that exists). All that is required is a good input/output curve. The only time a "turbo button" is actually neccesary is when the thing you are controlling actually has a wider raneg of outputs than you can reasonably input. The turbo button lets you effectively double your input range.

[UCGL_Guy]

Our team worked over the summer to ue 222's design - best shift on the fly made. After finally getting the bugs worked out they work beautifully. We stuck with the three speeds a low low a middle and a high.
Thanks to Brad and 222 -- we did modify and tweak your design somewhat and if time ever permits will share with others.

[Rombus]

Heres my whole take on the "Turbo Button" issue:
i would have set it up to be locked in high all the time then have a slow mode for Perscis movement to cap or get a tetra from the loading zone. We tried the 4 speed transmission this year but ended up having to lock it into 2nd gear and we did OK with it. I dint know the exact reasons why we chose not to go fully shifting, but i personally didn't see much of a need for a 4 speed tranny on such a small field.

[Dave.Norton]

Originally Posted by Max Lobovsky

We have a omniwheel holonomic drive with no feedback. It is likely one of the most difficult to control FIRST drivetrains that exists)

We used the same wheels, and after looking at your site it looks like we had an identical layout. One difference though, ours was very easy to control. When we prototyped it the drive wouldn't go in a straight line to save itself, problem with 3 points defining a plane, and four being over constrained. Which is to say every time it hit a bump it would steer into a new direction. By adding a rotational accelerometer and suspension to the platform we got it under control (little joke there) and now it is fun to drive.

[Veselin Kolev]

Team 1072 (Harker Robotics Team) used a TechnoKat style dog shifter this year. Two speeds, two CIMS, geared to go 6 feet and 11 feet per second. Man.. what a beauty. Of course we totally ignored the white papers and designed everything ourselves... just to improve even more on an already very good design. We first of all did not use a flex coupler, instead we put a chain sprocket directly onto the output shaft of the transmission. We also made the whole thing with .25" and .125" plates screwed together, you could literally take the shifter OUT by unscrewing 5 bolts. And the thing was sturdy... even thought the transmissions were tiny they could pack a punch. For the record, or robot in low can push TWO other robots with the kit transmission. And we did this during competition. That is how we know..

As for shifting, it was very smooth and quiet, no jolt or sudden lurch of the robot. I was thinking about putting sycros on if shifting was too rough, but we tested it and they shifted fine. I highly reccomend any teams with access to a mill and lathe with the proper tooling and the urge to make a 2 speed transmission to consider a dog-style shifter like this.

[bradleym1559]

Originally Posted by Tristan Lall

Back in 2000, Woburn's Blizzard 1 had just such a control. But it wasn't included to surprise people--there was a bit more pragmatism involved! As it happened, that robot had front-wheel-drive (old Johnson drills) and rear casters, and because the weight distribution was very biased toward the rear, the robot had an annoying tendency to over-rotate when trying to manoeuvre. That button was in there so that full speed could be achieved if necessary (and usually with a fixed rear wheel deployed, raising the robot off of its casters), but under normal conditions, the robot wouldn't be so difficult to control. I'm not so sure it was of much actual use, though, since that robot was by far the worst-driving of any of the Woburn family (of those that I've taken for a spin, at least--that's the last six, plus one of the ancient Canada FIRST robots). It certainly wasn't an efficient or elegant solution, as sanddrag pointed out.

This sort of solution is only useful if you don't care about the power output (in 2000, we knew we had enough power to drive around at half-throttle), or conversely, don't care about the current and efficiency (we should have cared--we threw 30 A breakers on some occasions, that year, and were limited to a 60 A main fuse, not a 120 A main breaker). You're effectively decreasing the voltage of the motors--for instance, by half--which just doubles the current needed to sustain a certain power level. This is why it is fundamentally crazy to control something like a (DC permanent magnet brush) motor at low throttle settings (i.e. the Victor's output voltage is a fraction of the motor's rated quantity); you're just going to force the current way up, and the efficiency way down. Low efficiency just means heat losses, and to compound the problem, the internal fans don't work well at low speed.

So what's the right way? Shift gears. You'll have two speed ranges that allow the driver to avoid overstressing the motors by running them too slowly.

Of course, sanddrag already knows all of this, so I suppose that this is for everyone else out there....

That was exactly our problem (casters)...we had to limit our power because the robot was uncontrollable with the program we were provided with...so we had to change the gearing...but to save money we did it in the program. During the competition we increased the speed just to see what would happen..Divide by 2 instead of 3, and i couldnt control it (I was the driver)...all our mods to the program seemed to become more ineffective the faster the robot got. But next year we will need a more concrete solution to our gearing...because that was really our only issue--our robot was too slow. First year mistakes...we learn. This is what CD is for! Thanks for all the input!