pic: 6 speed?? Great in theory - lets see how it works!

[sanddrag]

Originally Posted by ahecht

2". Now, if you ... attached a larger gear to each of the six existing gears .

Thank you. My train of thought almost got there but it derailed. Thanks for getting it back on track. I'm not sure if that is the way the drawing is but that is certainly the way it would be done. Mystery solved by ahecht.

[Sean Schuff]

I hope this comes across as a simple explanation for the 6-speed transmission - you know how us teachers can complicate even the simplest of things!

First off, I've posted a new picture of only the gear reduction system of the transmission. The first gear (top red) on the primary shaft (on the left) is an initial input- from the CIM and drill motors. It is a 1.25:1 reduction from the gear on the CIM motor.

From the first double gear there is a simple tranisition of 1:1 to the first double gear (top blue) on the secondary shaft on the right. The overall reduction from the original input is now 1.25:1.

The first doulble gear on the secondary shaft reduces the speed (1.67:1) back to the second red gear on the primary shaft. Overall reduction at this point is now 2.0875:1

The second primary red gear transitions back to the second secondary gear at 1:1.

The second secondary gear reduces back to the third primary gear at 1.67:1. Overall reduction to this point is now 3.486:1.

The reduction continues down the system to the sixth gear.

The gear on the hexed output shaft (shown in the original picture) slides across the six blue double gears and engages at each one of them to provide six different output speeds.

Friction, while a concern, is negligible since all of the gears on the gear train (red and blue) are pressed on bearings and rolling rather freely. Granted the friction between the gears from one end to the next creates quite a bit of friction, I don't think it will be a huge concern. Like I've said before, we're going to start with a two-speed variation and work up from there.

I hope thsi explains how this system works a little better. Once you understand the stair-stepping concept between the gears, it is easy to see how you can set this up with as few or as many speeds as you need without great modification.

I will keep you all posted as we progress with the fabrication, construction and testing. I will also update the model to include gears with actual teeth so it is easier to see what meshes with what and how it all plays together.

Sean

[Biff]

There are threads about multi positioning with air cylinders. If I was going to do it that way I would use a control valve with a center off, no power all ports closed. A slider pot attached to the piston, and a really good programer.
You can stop a piston in any point of it's travel with the first, you can tell where it is with the voltage generated from the slider on a analoge input to the controler. The bounce and drag may cause occilation problems. And if your coding is not up to it, all bets are off. Other wise I would use shaft with very coarse thread, almost like that of a Yankee screwdriver, and a multi turn pot; driven as suggested by a globe motor. Good luck

[Tristan Lall]

I like the concept, but there are a few little things that I'd watch out for when finalizing the design.

I'm making a couple of assumptions about those gears, namely that they're 1/4" face width, and 24 pitch (the pitch is a guess, based on the face width, and the face width is an estimate based on the size of the other components). Using a spreadsheet which calculates the number of horsepower that the gears can accept, using the Lewis formula (let's all hope that my spreadsheet skills are up to par), those gears would have to have approx. 85 000 psi tensile strength. (The Lewis formula isn't perfect, but it ought to be fine for these purposes.) Fortunately, SDP/SI sells gears that exceed that spec nicely at 180 000 psi, but at around $50 (USD) each, you'd be spending vast amounts of money. (Backtracking a step, I'm only saying 180 000 psi, because SDP/SI quotes min. R_C 35 [Edit: That's the Rockwell C hardness scale] or so, which, in steel tends to exhibit strength of 180 000 psi. Actually, they quote R_C 55 for the teeth, but I'd rather not use that number, for safety's sake. Note that I've never tested this, or received verification from SDP/SI!) Are you planning to make your own gears (slices of gear stock, maybe?), and have them heat-treated?

I'm really starting to like the possibilities here, especially because of the expandability of the thing. It would be a trivial matter (except in terms of weight) to increase the face width of the gears to 3/8", if you needed to.

Make sure that you verify your gear strength calculations, and take into account the fact that the gears will be "crashing" together a little bit, due to the teeth engaging one gear after another (sideways). I don't know how durable the gears will be under that kind of loading, especially when shifting on the fly. Those through-hardened gears are rather brittle--this is definitely a good reason to test them!

The lead screw idea sounds good, and M's suggestion of the Geneva mechanism is an interesting one.

Finally, what's the purpose of the small slots in the sides of the gears? Some sort of aperture for an optical or Hall effect sensor? I don't think they'd be necessary everywhere, in that case.

[Sean Schuff]

Couple of clarifications -

1. The drawings I have posted are for illustration purposes only. It was, more or less, a quick and dirty way to prove a concept. And, I'm pleased to say, it worked. At least in the theoretical universe of Autodesk Inventor 8 Pro.

2. The slots in the gears (the blue and red components that are missing actual gear teeth) are there so it is possible to see the rate of rotation of each gear as it steps down. Motion constraints have been applied to all of them so you can actually see the reduction taking place when the constraints are driven.

3. I am working on a four speed variation (can't imagine the need for 6 gears unless, as JVN pointed out, the field gets REALLY big!) of the same design using actual gears so the concept is a bit more clear. The assembly is about 90% complete and weighs in at just a shade over 8 pounds without motors or a shifting input device (pneumatics/lead screw/etc.). I'll post a pic after I submit this post.

This new design uses standard hubless spur gears (20 deg. pressure angle) ranging in size from 15T to 45T. Their face width is 1/2" and they are tied together in sets of two (face to face) using drilled holes and roll pins. I can probably reduce a fairly significant amount of weight by dropping down to 3/8" face width. I've tried to standardize as best I can for ease of component acquisition.

Tristan - thanks for the insight into the Lewis formula. I'll have to keep that in mind when selecting components that can withstand the abuse of FIRST robotics! $50 each - YIKES! I better find a more financially responsible solution!

Sean

[sanddrag]

Originally Posted by Sean Schuff

Their face width is 1/2" and they are tied together in sets of two (face to face) using drilled holes and roll pins.

If there's a way you can keep the tolerance to press in dowel pins I would recommend that over roll pins. They have a much higher shear strength.

[Tristan Lall]

Originally Posted by Sean Schuff

Tristan - thanks for the insight into the Lewis formula. I'll have to keep that in mind when selecting components that can withstand the abuse of FIRST robotics! $50 each - YIKES! I better find a more financially responsible solution!

Depending on what sort of gear strength you actually need, you can use cheap (and relatively soft) gears in the high-speed stages (where they are less stressed) and only go with high-strength gears for the lower reductions. Also, though I've never tried it myself, getting carbon steel spur gear stock and having it hardened is probably an excellent way to control cost here. SDP/SI (and others) typically sell several inches of the stuff for the same price as a single gear. All you need to do is machine the bore, cut off a reasonable length, and add a keyway (or dowels, or whatever is necessary). Then find somewhere to get it hardened, or maybe do it yourself (and hope it doesn't distort--you'd have to inquire about the specific hardening process to find out if that's an issue).

By the way, for a 6- or even a 4-speed version to work well (it seems to me, anyway) that you'd need to be able to shift on the fly, even while pushing something, or crashing at high speed into something. (Because the chances will be higher that you'll be in the wrong gear for something, as you're doing it.) I think you've addressed that, but just in case it wasn't a priority--please make it one! Part of that involves using sufficiently robust gears, since a partially engaged gear does not evenly apply its load to the entire face width!

[Sean Schuff]

Originally Posted by Tristan Lall

By the way, for a 6- or even a 4-speed version to work well (it seems to me, anyway) that you'd need to be able to shift on the fly, even while pushing something, or crashing at high speed into something. (Because the chances will be higher that you'll be in the wrong gear for something, as you're doing it.) I think you've addressed that, but just in case it wasn't a priority--please make it one! Part of that involves using sufficiently robust gears, since a partially engaged gear does not evenly apply its load to the entire face width!

The design intent is shift-on-the-fly capability. How well it works will remain to be seen. We plan on auto shifting - requiring a driver to also manage which gear he/she is in is not something we want to do.

We would like our shifting mechanism to be fast AND accurate enough to engage the vast majoirty of the gear even in the heat of battle. Double-edged sword whether to go with wider or narrower gears. Wider gears mean more strength but also mean more travel to fully engage and more weight. Narrower gears are lighter and quicker to engage but may not have the necessary strength. Time to add some math to the design mix.

Sean

[Validius]

Just one question..... why?

[kmcclary]

I just ran across this old thread... Very cool idea.

OOC, Did it ever get made? If so, how did you solve the shifter problem?

My $0.02... Several methods of shifting immediately come to mind:

Method 1 - Threaded rod, Globe Motor and 10-turn pot feedback:

Use a high pitch threaded rod to drive the indexer, such that the entire run from one end to the other is less than 10 rotations.

On one end, attach a Globe gearmotor, with a "slippery" fitting (ex: use a short section of vinyl tubing as a "coupler"). This prevents damage in case of a software error, as the drive will slip if it hits end of travel.

On the other end, attach a 10-turn potentiometer. Wiring: one end to +5V, other end to Ground, wiper to an analog input. Attach it with a hard coupling, but set it so that the pot never over-runs. This gives you an analog voltage into the RC, proportional to the position of the indexer.

Calibration: Manually set the indexer into the center of each gear position. Record that voltage value into a table. Jog it around to record the RANGE of values that still legitimately set a solid gear ratio for each gear setting as well. Taking these error values (epsilons), look them over and decide upon a good maximum error band value you'll allow that still guarantees you are In Gear no matter which gear you select. Call this your Epsilon.

(I'm assuming here all of your drive motor Victors are set to "coast", to protect the gearbox from meshing damage.)

Now, to shift: Your software simply "hunts" for that value, within "+/- Epsilon". Look at the current pot value. Idle the drive motor, Turn on a "Shifting" light on the Operator's Panel. Move the Globe motor in the right direction to increase or decrease the value. Once within Epsilon, STOP, turn off the Shifting light, and re-enable the drive motor. This helps keep you from tearing up the gearbox while shifting, no matter what the operator is requesting of the drive.

Method 2 - Threaded rod, Geneva Wheel idea for indexing, Globe Motor, 10-turn Potentiometer, Home Switch for Geneva drive wheel:

Use a threaded rod for the indexer motion, and add a 10-turn pot to it for feedback, as before. This is now your "gear selected indicator"

On the other end, design and add a Geneva Wheel as an indexer.

See: http://en.wikipedia.org/wiki/Geneva_mechanism
Mechanism: http://upload.wikimedia.org/wikipedi..._animation.gif

Drive it with a Globe Motor. On the drive wheel, add a notch for the green wheel's "the pin is away from the Geneva Wheel" position, and monitor it with either a micro switch or an opto interrupter, connected to one bit of Digital Input. This forms the Home position of the drive wheel for each gear position, where you will stop the motor. (I can show you how to wire this, if you need more information.)

Software: Again, record the pot's analog value for each gear position, and calculate the overall Epsilon.

This time though, the analog value will be used as an INDICATOR of which gear you are in. You still need the Epsilon, as the analog input values WILL drift with slightly with temperature shifts (a known effect of A/D conversion systems). That is easily implemented by a CASE statement, that says if whenever the values is BETWEEN (value-epsilon) and (value+epsilon), AND the HOME is hit (IOW the globe shifter motor is OFF), then it's in THIS gear...

Shifting: As before, coast the drive motor, turn on the SHIFTING indicator, and use the current value to determine which way to kick on the shifting motor. Once the HOME switch is off, use that signal to keep the motor on until the HOME switch is hit again (one gear step). If you're still not in range, keep going. Once the POT is in range, keep going until the HOME switch is off as well, then stop. Turn off the SHIFTING indicator, and re-enable the drive motors.

Method 3 - Two Indexer Cylinders (Up, Down), ratchets, and a Potentiometer or switches for feedback

Two tiny cylinders: 3/4" bore, and 1/2" or 1" stroke (whatever the design calls for). One cylinder is INDEX UP, the other INDEX Down.

Mechanically set it up such that each full cycle of one of the cylinders runs a ratchet step system, that either increases or decreased the gear by one step. ("The design of which shall be left as an exercise for the reader"... Don't you just HATE that when your teachers do that???)

Current position is indicated by either switches, or a potentiometer. One way to connect the potentiometer is via an inverse rack and pinion system, watching the throw of the shifter. An alternative is to use micro switches at each position.

If you use switches, but become short of digital input lines (you need them for other reasons), wiring lines can be saved by:
a) attaching the switches to a Priority Encoder chip, which converts single lines into a binary value. (This is roughly a $1 IC in most logic chip families.) or:
b) attach the switches to an R2R resistor ladder to encode them as an analog voltage, which is then connected to one channel of Analog Input. Decode like a pot as with method 2 (bands of values equal each gear position), as again the A/D converters do drift, and values are rarely EXACTLY the same from one sample to the next.

For safety to keep the mechanism from tearing itself apart if an indicator switch fails, I'd also make sure you either have some End of Travel switches, and/or set up the mechanics such that attempting to index beyond the end of travel will cause something to slip, vs tear up.

Does this make sense?

- Keith

[gorrilla]

I dont see the value of anything more then a two speed unless u want to waste a lot of weight

[kmcclary]

Originally Posted by gorrilla

I dont see the value of anything more then a two speed unless u want to waste a lot of weight

Well, for most of the contests I've seen to date I'd agree, but I can imagine exceptions.

The two primary gears I want are of course "dash around unhindered - high gear" and "position yourself for scoring - low gear".

However, having a "sumo fight - VERY low gear" can be extremely useful in some contests, where you have to hold ground or displace another bot from their scoring position, but it's too slow for efficient scoring. (But that can often BE the low gear...)

Beyond those cases though, the only time I'd want more than three gears would be in a contest where you could score in more than one way and they require different speeds not easily attained by the two (or three) gear system.

One game that comes to mind is Zone Zeal. It could easily use up to 4 ratios, optimized for minimum time consumption at full speed:
- Dash while empty
- Drag 1 loaded goal as quickly as possible without stalling
- Drag 2 loaded goals as quickly as possible without stalling, and
- SUMO mode - tank crawl against a blocker while holding 2 goals (or displacing an opponent holding 2 goals).

<edit>
I do have a hard time envisioning use for anything over a four speed transmission, though.
</edit>

- Keith

[gorrilla]

oh that would be better than just two, and i have a feeling we will be doing something like zone zeal pretty soon

[DMetalKong]

Originally Posted by kmcclary

Well, for most of the contests I've seen to date I'd agree, but I can imagine exceptions.

The two primary gears I want are of course "dash around unhindered - high gear" and "position yourself for scoring - low gear".

However, having a "sumo fight - VERY low gear" can be extremely useful in some contests, where you have to hold ground or displace another bot from their scoring position, but it's too slow for efficient scoring. (But that can often BE the low gear...)

Beyond those cases though, the only time I'd want more than three gears would be in a contest where you could score in more than one way and they require different speeds not easily attained by the two (or three) gear system.

One game that comes to mind is Zone Zeal. It could easily use up to 4 ratios, optimized for minimum time consumption at full speed:
- Dash while empty
- Drag 1 loaded goal as quickly as possible without stalling
- Drag 2 loaded goals as quickly as possible without stalling, and
- SUMO mode - tank crawl against a blocker while holding 2 goals (or displacing an opponent holding 2 goals).

<edit>
I do have a hard time envisioning use for anything over a four speed transmission, though.
</edit>

- Keith

It seems to me that a better way to get four speeds would be to use two dogs, that way you do not have the gears smashing into each other. You could just have something like this:

   In
    |      |      |
    |     ---X---------
    |      |      |
  -----X-------   D
    |      |      |
    D    -----X-------
    |      |      |
   ---X---------  |
           |      |
                 Out

X's are interfaces between the gears,
D's are the dogs,
----- is a gear,
| is an axle.

[Tom Line]

Originally Posted by Sean Schuff

My guess - and it is ONLY a guess - is about 8 pounds without the motors. I'm probably ballparking high at the moment. Once all of the details are filled in I will give you an actual weight including motors and shifting mechanism. All of this is very much in the preliminary stages.

Does anyone know how to stop a pneumatic cylinder at 6 specific, pre-determined points in the stroke??

Glad to share!

Schuffman

Use a leadscrew on a banebot or other low-gear motor.

Doh - looks like kmcclary beat me to the punch.