31-07-2014 22:42

ekapalka

Re: pic: 3216 Swerve

Another angle

01-08-2014 09:42

coalhot

Re: pic: 3216 Swerve

I'd double check some of the VP kits, I distinctly remember there being an extended output shaft add on/kit.

01-08-2014 10:43

glennword

Re: pic: 3216 Swerve

Have you thought about utilizing the tapped hole in the end of the VP output shafts? you could either use a 1/4-20 screw and somehow clamp on the sensor shaft that way, or you could drill it out, and put a tapped setscrew hole in the side of the shaft to lock onto the sensor shaft.

01-08-2014 16:00

Adrian Clark

Re: pic: 3216 Swerve

Using the tapped hole on the end of the shaft is a good idea, it's worked for us in the past but not exactly as you describe it. In 2013 our shooter wheels were on two vp gearboxes and we needed to attach encoders to each for control. Our solution was to thread in a bolt with the head removed and couple this to the encoder via a short piece of surgical tubing held on by zip ties.

-Adrian

01-08-2014 17:01

anthonyttu

Re: pic: 3216 Swerve

A tension system for the belts. It could be very difficult to assemble without something. Also you might want to look into sliverthin bearings http://www.silverthin.com/ They are a few X more expensive then the ones you used but can be worth it with the weight and size.

01-08-2014 19:50

ekapalka

Re: pic: 3216 Swerve

The problem I'm facing is that the absolute encoder/potentiometer needs to rotate with the module 1:1. The only solution I have been able to come up with is to have a separate series of gears originating from the VersaPlanetary at the same ratio as the belts, but I've run into found a few issues with that (stemming from the short VersaPlanetary output and/or vex and andymark gear thickness). I'm trying to keep as much as possible contained within the plates that make up the body of the module (and I'm trying to go no higher than 1.25" from one to the other). I'm sure there are a lot of ways around this I haven't thought of... [EDIT] I could use a motor other than the BAG motor with a lower RPM and have 1:1 pulleys turning the module. Then mount the sensor right on top of the VersaPlanetary

Hmm... I was told that properly spaced pulleys wouldn't require a tension system :/ I guess I'll get looking into how that's done. As for the bearings, do you have a ballpark estimate for the prices they offer? I like the weight and look of them, but the bearings I've found are only $19.

Even if my team actually decides to go through with this and build it (which I'm almost certain they won't), we're probably not going to use it in competition because of the technical problems that undoubtedly go along with swerve.

I was wondering if anyone had any thoughts about holding the pivot yoke to the rotating module using screws (8xM5). Its not something I've seen done, so there's probably a reason...

01-08-2014 20:15

Re: pic: 3216 Swerve

[edited]

01-08-2014 20:22

Jared

Re: pic: 3216 Swerve

If you can get the center to center distance correctly, like if you're doing it on a mill/CNC, you won't need tensioning.

Silverthin bearings are nice, but really pricey. If you've got the room for thicker ones, you could save some money. I know 1640 uses a large ball bearing, a large thrust bearing, and a bushing on top.

01-08-2014 21:37

ekapalka

Re: pic: 3216 Swerve

Yeah but that means that the modules have to be perfectly aligned every time its turned on or the wheels have to be in the same orientation when the robot is turned on as when it was turned off (and the last angle value would have to be persistently saved so it remains between reboots). I'm also the programmer, so I'm trying to avoid having to do that

02-08-2014 04:21

asid61

Re: pic: 3216 Swerve

Okay, so I have a few questions and suggestions for you:
1. Why do you need the encoder to have a 1:1 ratio? You can still tell if the wheels are out of alignment, and realistically it would be better to just have some way of keeping them in line for calibration mechanically, with something such as surgical tubing between the axis of the wheels. I think 1640 had something on keeping modules in line on their Swerve Central site.

2. Is this a shifting design? If not, you are almost certainly using way too many gears and pulleys. You can probably cull almost every gear by using a sprocket reduction to the wheel. For example, using a 10t #25 sprocket on the shaft with the bevel gear and a 42t #25 sprocket on the 4" wheel (I assume it's 4 inches) will net you around 18fps adjusted, and you can lower that via a single pulley reduction going from CIM to turning module. Less gears means cheaper and less complex.

3. Mount the encoder to the end of the versaplanetary and save yourself some time. I think Western Digital sells 10mm shaft absolute encoders, so you can drill out the 1/4"-20 tap on the end of the shaft and add a set screw.

4. Is the center of the turning module, looking down from the top, equidistant from both of the sides that mount the swerve module? That way you don't need to worry about module orientation when putting it on a chassis, and programming becomes a bit easier.

5. How thick are the top and bottom plates, and why? Just curious.

6. You don't need to use roller bearings for turning the module. Bushings can support tons of weight at low rpms, which you are running at anyway. Even a thick plastic bushing on the top plate can provide a strong interface. Ball bearings will work fine, but I think a flanged bushing would work better so you don't need to depend on a press fit or put a lot of axial load on ball bearings.

7. What bevel// miter gears are you running? I've never found a good place to get them cheaply at other then Vex, and the Vex bevel gears are pretty large.

8. How much does this weigh? If it weighs more than 8-9lbs, you need to rethink weight distribution. It's definitely possible to get it lower than that.

Overall, it looks very slick. I like the bearing mount on the top of the module. I hope your team can build a swerve!

02-08-2014 23:25

ekapalka

Re: pic: 3216 Swerve

The encoder I'm intending to use is a USDigital MA3 absolute magnetic shaft encoder (not the incremental encoder used to measure speed). Preferably, the robot should be ready to go as soon as its turned on with little or no calibration (which I guess might be unrealistic, but I'd still like to make the code relatively straightforward). I'm not sure what you mean by a mechanical solution... something to align the wheels with human assistance, or something else?

Its not a shifting design. The max free speed is supposedly around 16.5fps. Aside from the weight difference, I'm not sure why I chose to go with belts. I chose the particular gears I'm using to take up the least space. I tried with various other ratios, but the one I chose allowed me to make the plates that hold the module together significantly smaller by having the gears be positioned close around the bearing. Like this.

I'm currently re-designing the body, but the version shown in the pictures is equidistant. Unintentionally, though - I'll make sure to do it this way in the final design as well.

0.25in, because its the thickness of the WCP SS and DS gearboxes. Additionally, its the thickness of most Vex/AndyMark bearings (minus the flange), to there's a lot of nice flush edges.

Vex 15t Bevel gear. I haven't found a problem with them other than the fact that they're 3/8in hex and most everything else is .5in hex, so the shaft going to it has to be milled from one to the other and it needs to have a set screw put in it...

With quite a bit of hardware missing, it weighs 10.25lbs, but once I add the missing hardware, cut out relief pockets, and (maybe) trade out one of the heavy bearings for a bushing, it should be around 9-10lbs. Its not for competition, so I'll be satisfied with anything under 11lbs.

Thank you for your input!

03-08-2014 22:18

Tyler2517

Re: pic: 3216 Swerve

What is the use of all the extra material below the wheels axial?

The module looks supper tall. This will give you a higher center of gravity decreasing performance.

The co-axil looks complex.

I don't see the need for the gears. Most teams that i have seen that don't use shifting go directly from the cim to the co-axie with a timing belt and the majority of reduction in the yoke. Removing the gears will give you a higher mechanical efficiency and less moving parts in the high speed parts of the transmission.

The top of the modules bearing can definitely be using a bushing saving weight and cost. I personal like the thrust bearings riding on the bearing like 1640.

The plates look complex with a lot of milling operations on places that are not holding weight. Mainly where the bearings are why is this? It would be lighter and cheaper not to have to get plates that thick and mill them down even if the beaing plate are not smooth.
Over all a great start

05-08-2014 14:07

kk052

Re: pic: 3216 Swerve

screws can be inacuate, also how are you planning on putting a pully on if you cant change the center distance in any way?

05-08-2014 14:15

Andrew Schreiber

Re: pic: 3216 Swerve

You could always use an incremental encoder and an index (limit switch of some sort) as a zero.

06-08-2014 06:09

asid61

Re: pic: 3216 Swerve

That's a good encoder. To align the wheels, you can use a hall effect sensor and magnet aligned such that the wheel are all pointing the same way when the hall effect sensor is activated. Alternatively, putting a hole directly above each wheel axle and stretching surgical tubing between the wheel axles will make a pair wheels face the same direction. Then you just set constants in the program to get them aligned properly.

Belts are fine. They offer greater efficiency at the cost of width/ space. I just think that you could use a large reduction to the wheel axle instead of using so many gears. Then you could just use a belt or just a pair of gears to go to the coaxial axle from the cim.

0.25in is pretty hefty. Keep in mind the WCP gearboxes are heavily pocketed. When you pocket your swerve modules, which you should, make sure to get the pocket design checked out by a mentor/ engineer to maximize strength and minimize bending. Get the wheel very close to the edge to stop bending issues.

Vex bevel gears are fine. The low pitch gives them resistance to bending issues. What do you mean by a set screw? Avoid set screws on shafts whenever possible. A machine shop, or even a basic benchtop mill will be able to cut a 3/8" hex on the end of a 1/2" hex shaft without problems.

10lbs? I'm a huge weight freak when it comes to drivebases. Especially when it comes to swerve drives, the main setback that I see with them is size and weight. Size you are good on. 10lbs is very heavy though. Even if this will not be used in competition, I strongly advise you remove unnessesary weight. Like I mentioned with removing gears, there are ways to reduce weight.

I have noticed that designing a tall module with a cim facing up actually produces a similar COG when compared to a very short module with the cim facing down. Unless you can get the cim facing sideways, which requires more bevel gears and such, it's easier to keep a tall module.

06-08-2014 08:14

Gdeaver

Re: pic: 3216 Swerve

I don't have time to list all the little problems with this module but I see many. I would suggest you search out the cad files that are available from several teams that have done swerve coaxial. See how they have solved many of the module design problems.

06-08-2014 13:50

ekapalka

Re: pic: 3216 Swerve

Looks, I guess. It weighs a fraction of the modules weight, but I'll remove it if need-be to save on weight later.

Yep. It's 7.125in from the bottom plate to the ground. The centre of gravity, however, depends on your frame and how you mount it. Most modules I've seen are pretty tall, so I didn't pay it much attention. The thing I've most commonly seen is to lower the internal components within the frame so that is only has about 1" of clearance. Like this. I'm thinking of using gears instead of belts or chain to reduce the size, but the CIM is about three inches from dragging on the ground

...Yes. Quite. This is the only part we don't have the means to make ourselves.

If I were using gears in the rotating module, I would have a lot more options as far as ratios go, but VexPro pulleys only come in a few sizes, and I'm using the lowest reduction available. I'm still considering my options, though.

I have nothing against bushings, but I've never really used them. Are they loud? It seems to me that having no moving parts would cause some abrasion and unnecessary stress / heat. By bushing do you mean sleeve bearing? Can you make a recommendation? The bushings/sleeve bearings I have found on McMaster all have to be cut to length and are very similar in weight to the ball bearings I'm using now.

There's a lot of material that I intend to remove once I finalize the layout. I'm not sure I'll be able to live with non-flush bearings, though. They're 19mm thick.



Screws... What do you mean that they can be inaccurate? I'm hoping for strength by redundancy, which is why there's eight 5mm screws. As for the pulleys, I always figured that you could put the belt on the pulleys before you put the pulleys onto the shafts and then place it on as one piece... I guess that's not how it's done? I've never seen a swerve module using pulleys that did have a tension system.

I personally love hall effect sensors, and hope I'll be able to find a use for them in this project. Any idea if magnets will mess with the absolute magnetic encoder?


I'm not entirely sure how to do the pocketing, but the WCP gearboxes have no place on them where the metal is less than 0.25x0.25in (meaning that you can't have two pockets less than 0.25in from one another).

I hate set screws, but I don't know what else to do. Aside from the other bevel gear being in the way, there's nothing to hold the bevel gear in place (unless it's force fit onto the shaft). I'm not sure if this is a viable option. I'm glad to know that it's possible to go from one size to the other, though. We don't have too many resources at our disposal, but we do have a small mill.

The bearings weigh nearly 1lb each (according to Inventor, not the manufacturer), so I'd really like to do something about them.... There's a lot of unused space in the body, so I'll be able to do a lot of pocketing to reduce weight. I'll reduce the number of gears if it comes to it, but they don't weigh very much.

Thanks for your input, sort-of. I reviewed every swerve drivetrain I could find and made a huge checklist of features that they utilized. What I've tried to come up with is a collection of those things. I'm not sure if this is what you are noticing, but the first few pictures lack shafts, spacers, and sensors. I'm well aware that those are required for a working design.

Thanks everyone!

06-08-2014 18:00

asid61

Re: pic: 3216 Swerve

That weird custom pulley thing can be made on a 4-axis CNC. However, do you really need it? Is it not possible to just modify some Vexpro pulleys to do the same thing?

To hold the bevel gear in place axially along the shaft, you have a couple options:

1. A screw clamping down on a washer on the end of the bevel gear shaft. This is pretty easy, and you only need something like a 4-40 screw, Then use a spacer on the bak of the bevel gear.

2. E-clips or circlips. Both are good. I prefer e-clips because they are easier to mount, but they also need a deeper groove. Slap one on near the end of the shaft and you're good to go.

3. Roll pins. My least favorite option, as they are a PITA to get into a hole, but if you are up to the task go for it. Theya re very strong and hold position well.

Do not use set screws. There are still a lot of axial forces acting on the bevel gear due to their operation. Set screws can and will slip.