24-08-2012 13:43

MattC9

Re: pic: 2012 frc971 transmission

Are those custom gear's or are they ones from andymark?
Also, Where do you get these special pneumatic pistons, bimba? and how are they different from the standard ones?

24-08-2012 13:46

Akash Rastogi

Re: pic: 2012 frc971 transmission

I loved your simple solution to saving space this year, along with the pancake cylinders for shifting. Are there any tips or tricks you might have when making an "inverted" gearbox? Were there any issues you encountered that were unique to this type of setup? What would you do differently?

Thanks!

24-08-2012 15:11

Jared Russell

Re: pic: 2012 frc971 transmission

Is there any measurable benefit to the fan above the CIMs?

24-08-2012 15:35

Cory

Re: pic: 2012 frc971 transmission

The pistons are from Bimba. They are shorter than a standard piston (albeit wider). They also generally ouput less force, as the rod is smaller than the commonly used shifter pistons.

24-08-2012 15:41

SteveGarward

Re: pic: 2012 frc971 transmission

Do you have a part number for the pistons? I was looking at a couple of different ones a couple of weeks back. Just curious which ones people are using.

24-08-2012 17:16

R.C.

Re: pic: 2012 frc971 transmission

Jared,

They really helped on the practice bot. I was talking to Austin at Davis this year and was unsure about trying out fans. After Davis we put fans on the practice bot, it really helped us get longer practice time.

On the comp bot in 2 minute matches, totally unsure 971 would have to answer that one.

-RC

24-08-2012 17:48

cbale2000

Re: pic: 2012 frc971 transmission

I don't have any experience with them on CIMs, but I added one to the KOP air compressor on our demo bot and it dropped the running temperature and idle cooling time significantly. We had been having issues where the compressor would get so hot from running constantly it would melt the hose coming out of it (causing an "explosive" loss of pressure). Since adding the fan we have yet to see such a problem occur again, even after a 5+ hour demo that was outside on a hot day and in direct sunlight.


Somewhat related, on this years robot we used two FisherPrice (I think) motors to run our shooter system. We had been having an issue where after moderate use, the motors would heat up and then, the lower one in particular, would stall when trying to shoot a ball. The motor had been mounted to the robot using an aluminum collar we made for it (which covered about 60% of the motors surface area and was roughly 3/4" thick). To fix the problem, I took the collar off and cut small groves around it (like a computer heat-sync, and since putting it back on the robot we have yet to have the launcher stall on us again.


I could see a fan on a CIM being useful particularly in elimination rounds where you have a lot of matches close together with very short breaks between them. In years past our motors have gotten very hot (one year we even used a can of compressed air upside-down to cool down our drive motors) during finals, so an active cooling system could be a good way to alieviate such problems and maybe squeak out that little extra bit of performance during the most important matches.



TL/DR version: Temperature matters, cooler running motors will perform better.

24-08-2012 19:42

AdamHeard

Re: pic: 2012 frc971 transmission

I'm not seeing how this is single reduction. I can't get the numbers even close to those speeds, and in the picture it looks like the gearshaft is above the wheelshaft (or the wheels would hit the CIMs).

Is there another gear reduction external to this gearbox that directly drives the center wheel?

24-08-2012 20:08

James Kuszmaul

Re: pic: 2012 frc971 transmission

You are correct; A 15 tooth gear coming from he transmission drives a 50 tooth gear which is on the same axle as the wheel.

25-08-2012 15:29

kevincrispie

Re: pic: 2012 frc971 transmission

In the back of the picture, you might be able to see that some of the wheels had gears on them. These were the center wheels that interfaced with the transmissions.

27-08-2012 01:57

roystur44

Re: pic: 2012 frc971 transmission

The use of the final gear allows us to make the transmission removable while the belts and wheels can stay on the chassis all tight. Let's see a WCD do that.

We need to make the inside side plates a little stiffer by insert some bends into the plate. This will give us a flange to mount the plastic grease guards.

We need to insert alignment pins from the transmission to the chassis to ease the assembly and get the optimum gear spacing. We had to manually tune the alignment and if the screws became loose the transmission would have power loss.

27-08-2012 16:26

roystur44

Re: pic: 2012 frc971 transmission

Yes,

Fans help out on other motors too. Try running a pair of Fisher Price motors at high load and run them intermittently in forward and reverse. They don't like that and you will see how smoking hot they get. Attach a pair of small fans to direct continuous air flow over the motor cooling slots and casing and you will see a tremendous improvement of the power output over a time frame.

27-08-2012 17:15

Nate Laverdure

Re: pic: 2012 frc971 transmission

I love the space savings with this "inverted" setup, but in order to place the CIMs outboard of the gearbox, you risk a higher moment of inertia. This will have the effect of making the robot harder to turn.

Some quick calculations tells me that moving the CIMS outboard will increase the total moment of inertia by about 5% (for a "long" chassis) to 8% (for a "wide" chassis) for an otherwise equally-distributed robot. For a constant turning torque, this results in a similar percentage less angular acceleration.

Takeaway: for a snappy turning response, mount your heaviest items as close to the center of rotation as possible.

27-08-2012 17:49

Nuttyman54

Re: pic: 2012 frc971 transmission

This is a good point, and 5% is something that will be noticeable to most drivers. There are a few other factors as well that go into the decision for us:

1) With most 6wd dropped center setups, turning is usually really really snappy already, sometimes too much. Especially with a robot geared above 15fps, turning response can be almost uncontrollable without a lot of practice.

2) The drive code that 971 uses (along with 254 and maybe some other teams) is set up to do controlled radius turns in normal operation, rather than controlled turn rate, which mitigates the "snap" turn response characteristic of 6wd. This makes the robot much more controllable at high speeds (the 971 2012 robot was geared to 17fps no load). Further discussion on this style found here. Because of this, we try to avoid needing to turn in place when we drive, so the moment of inertia aspect isn't really an issue for us.

28-08-2012 00:46

MichaelBick

Re: pic: 2012 frc971 transmission

I'm wondering if you think that you would run this drivetrain again considering the following:

• weight gain for the extra stage rectified by the less weight of belts
• efficiency loss for the extra stage mildly rectified the the greater efficiency of belts
• more prone to over turns mildly rectified by better software
• more prone to over turns mildly rectified by more space on the inside of the chassis
• 1 more point of failure mildly rectified by less maintenance in general
• harder to implement encoders mildly rectified by less machining required
• harder to replace belts mildly rectified by easier to replace gearboxes
• more expensive(extra gears for extra stage and for encoders) rectified by less machining

Out of all of these things there are a few pretty nice benefits(that still outweigh, at least imo, any negatives associated with them): less machining and less maintenance.

However you lost some one of the most important benefits of a classic WCD to me: the prototype with old drivetrain parts without keeping a large stock of parts(ie belts).

Did you think the trade off was worth it? Do you think for next years you would go with a more conventional belt drive, or maybe even a classic WCD?

28-08-2012 00:48

AdamHeard

Re: pic: 2012 frc971 transmission

Most of these are aren't an issue. They don't have an additional stage of gearing compared to a west coast drive, they just have their shift stage before a reduction (versus a reduction, then shift stage).

28-08-2012 01:00

MichaelBick

Re: pic: 2012 frc971 transmission

I just relooked over the photos and it makes sense now. Nevermind

28-08-2012 09:52

Nuttyman54

Re: pic: 2012 frc971 transmission

I'll try to answer all this as best I can. Adam pointed out that we don't actually have an extra stage, we just moved the second stage external to the gearbox and onto the center wheel in a dead-axle setup, which in itself gives us some nice benefits as Roy mentioned.

One thing to keep in mind when you look at 971's designs and design choices with this drivetrain is that it is the result of many years of iteration, and we spent a lot of time trying to make it an integrated package with the team. Our decisions were not based solely on mechanical benefits, because we have some really great software mentors and have developed a driving style along with the physical robots over the years. This robot is the result of designing not just for the game, but for the team that would be using it, and playing to our own strengths. A lot of the tradeoffs and decisions make more sense in that context, such as the outboard motor effect on rotational inertia discussion above. Because we know that our drive style and code mitigates this factor, it benefits us more than it might some other team.

The team would definitely do this drivetrain again. It performed fantastically and we never really had any problems with it that I can remember. It was light, efficient and very easy to maintain.

It's also worth noting that 971 has not done a WCD since 2007. One of the hallmarks of WCD, and what makes it so easy to prototype on old drivetrains is the cantilevered setup and open layout which allows standard parts and easy swapping. In general, sheet metal construction encourages a more integrated approach to achieve an efficient design, but that makes it not great for prototyping. So in that regard, we don't consider it a "loss" because WCD has not been in consideration since the design didn't play to our fabrication strengths.

The 2012 design is the result of many years of various drivetrain iterations and innovations. Each is unique because of the game challenge, but there is a clear progression of eliminating problems while trying to maintain the improvements. One of the big driving forces behind the changes this year was the maintenance. On a traditional WCD, the cantilevered wheels are very easy to access and replace tread on because they are outboard of the frame. Due to our sheet-metal fabrication, cantilevered is not an optimal design so we switched to the dead-axle setup to allow drop-out wheels.

I'm not entirely sure what you mean by "conventional" belt drive, but we found this setup to be a nearly perfect driveline for us this year. We began working belts into our designs in 2011, with the elevator and roller claw both being belt-powered. Our experience with that gave us enough confidence to want to use it on the drive, and our experience this year means we will probably continue. As far as classic WCD, as long as 971 is building with sheet metal, I doubt we will ever run a "classic" WCD. It is far better suited to box tube construction techniques like 254, 1868 and 1323 use, and really isn't a good use of sheet metal resources. I don't think that our design is any better or worse performing than an equivalently designed WCD, but for 971 it's the right choice.

28-08-2012 09:52

DampRobot

Re: pic: 2012 frc971 transmission

MICHAELABICK brings up a good point. What lead you guys to use belts? I've done a few weighted design tables to compare them with chain, and they would only win with an extraneous "quite operation" or "coolness" category.

What benefits (other than the above, which shouldn't really be considered in a design) led you to use belts on your bot this year? As I see it, they add a lot of lead time and cost for parts that need to be ordered and can't really be reused in many situations (unlike chains and sprockets). I believe I read a report a few months ago that concluded that the efficiency benefits from belts over chains are marginal at least for FRC. If belts cost more, take longer to get, and don't perform much better than chain, was there some other factor that led you to chose them?

28-08-2012 10:15

Cory

Re: pic: 2012 frc971 transmission

Weight is an obvious reason. Pulleys are all aluminum and belt is vastly lighter than chain.

28-08-2012 11:46

PAR_WIG1350

Re: pic: 2012 frc971 transmission

Depending on the belts' construction, the issue of stretching can also be eliminated (or at least minimized to a negligible level) so you can design for exact center to center distance without the need for tentioners to compensate for the stretch. This reduces complexity and can save additional weight.

According to the Gates website, belts, especially GT2 belts, reduce backlash significantly which makes the robot more responsive and allows for more precise control in autonomous mode.

28-08-2012 11:57

roystur44

Re: pic: 2012 frc971 transmission

Belts have better damping characteristics under load. Try running a chain setup vs a belt setup over 12 inches to a high speed shooter wheel and you will see a big difference in the speed control, efficiency and the damping.

Belts are quite. There is less friction loss with belts which means higher motor efficiency. Chain tends to stretch over use belts do not. Lastly belts are easy to align they are more forgiving in long runs and when properly tensioned don’t require chain guides

28-08-2012 14:09

Brandon Holley

Re: pic: 2012 frc971 transmission

To go along with this- we ran our belts without tensioners at all this year, dead center to center distance and that was it. Zero issues all year.

You also don't have to 'make' chains to certain lengths, you can just buy the belt and design around it. This usually saves at least a few hours later in the build season. I know once we had all our components in, the assembly of our drive system took less than an hour.


Damp- would you mind sharing your W.O.T.? I'm just curious to see what metrics you value versus my own.

-Brando

29-08-2012 00:27

MichaelBick

Re: pic: 2012 frc971 transmission

After realizing that your gearbox had two reductions instead of three(that was really nice designing by the way, as expected from 971), it became clear why the belt drive had some really nice advantages over the chain. I really thought that you were running a three stage gearbox for a second, which really threw me off.

Thanks for the great explanation of your drivetrain. It was fantastic that you emphasized building to a team's strengths, and I think that is one of the most, if not the most, important lessons for all the new teams.

29-08-2012 10:53

DampRobot

Re: pic: 2012 frc971 transmission

Sure thing. I've always designed belt setups just like chain, with a tensioner/sliding endpoint, so this effected my WOT:


Category Weight Chain Belt
Availability: 4 4/16 3/12
Weight: 4 3/12 4/16
Manufac.
Speed: 3 4/12 4/12
Total: 11/40 11/40

The thread I was referring to in terms of efficiency can be found here: http://www.chiefdelphi.com/forums/showthread.php?t=76964&highlight=belt+chain+effici ency. I believe their methods were quite scientific and methodical, and they concluded that belt was 6% "faster" then chain over a given distance.

6% used to seem small compared to the difficulty of obtaining belts that can only be used for one application. I suppose that not having to do sliding tensioning blocks would be a plus for chain, but I'm really not that sure. Although teams certainly have been successful with it, center distance design with belt or chain has always seemed like a technique that could tend to cause problems when you actually try to put it together.

29-08-2012 15:14

AdamHeard

Re: pic: 2012 frc971 transmission

6% is substantial when teams are operating on fixed power.

The argument that the lead time of belts is a disadvantage isn't really the entire story there. Do teams really build their entire robots with no parts that have a lead time? We have several suppliers of belts, and we KNOW that we can get any given belt we need in reasonable time. We also make sure to design around belts that exist and are in stock, which is trivially more effort than designing around a chain spacing you know.

29-08-2012 17:22

scottandme

Re: pic: 2012 frc971 transmission

For those who are using/have used belts in their drivetrain, what belt/pulley combinations have worked well? I'm assuming everyone is using the 5mm GT2 profile, but I've seen a good amount of variance in belt width and pulley diameter.

We're currently working on a drivetrain based around 15mm wide 5mm GT2 belts, and 20T pulleys. The gates website has a lot of information regarding the rated load capacity of different profiles/belt/pulley combinations, and it looks like 20T pulleys should be sufficient, but some real world experience would be helpful.

29-08-2012 17:24

Jared Russell

Re: pic: 2012 frc971 transmission

Having a number of teeth that is evenly divisible by 3 lets you hold a pulley with a standard 3-jaw chuck more easily, just in case you want to do something to the bore.

29-08-2012 18:38

scottandme

Re: pic: 2012 frc971 transmission

The tentative CAD design for now has us running a 6WD Franken-WCD with the two 15mm belts running inside a 2"x2"x0.125" tube. Not really ideal from a maintenance standpoint, but I'm hoping the belts will reliable enough.

I'm also fond of the divisible by 3 trick, but I think the 18T is the minimum Gates recommends for the 5mm profile, and bigger pulleys have higher load capacity. The 20T is about the biggest pulley size that fits comfortably inside the side rail tube, accounting for the center drop. The next pulley divisible by 3 is the 24T, which is too big. Setting up the 4-jaw isn't too big of a hassle, but I would probably just make a fixture plate for 8 pulleys, probe the existing bores and have the mill drill the bores to 0.5" before broaching them.

29-08-2012 18:43

AdamHeard

Re: pic: 2012 frc971 transmission

You get a little more freedom if you aren't set on running the belts internal. That would let you pick a lighter profile tubing for your frame as well.

29-08-2012 20:37

scottandme

Re: pic: 2012 frc971 transmission

That's true, I think it'll be the pulley diameter needed that dictates that decision more than the weight gain/loss. 2x1x0.125 is 0.8085 lb/ft vs 1.036 lb/ft for 2x2x0.125, so it's only ~1.3 lbs lighter all said and done. Probably figure some weight added when you push the gearbox further inside the chassis to accommodate the ~1.25" of belt.

It's been hard to find enough detail from some CD research, but here are some I found.

1625 ran 2x1 frame with 9mm wide belt http://www.chiefdelphi.com/forums/sh...ad.php?t=86668

2791 looks like they used 2x1 with a single 9mm belt in 2011, but I think they mentioned ratcheting going full fwd to reverse. Also needs idlers to maintain proper wrap, I would rather avoid that. (http://www.chiefdelphi.com/forums/sh...ad.php?t=94701)

2791's 2012 chassis looks like 4x2 tubing with two 15mm belts. (http://www.chiefdelphi.com/forums/sh....php?t=101454). More weight, and requires >4" wheels, but solves the belt problem at least.

29-08-2012 21:39

Chris is me

Re: pic: 2012 frc971 transmission

We've run a very similar setup for two years, but we used 3x1.5" tube. The extra inch lets you get a bigger pulley in the tube and a little more room for installation, which makes the drive stronger and less likely to ratchet with quick direction changes.

Our drivetrain was literally maintenance free since we first set it up in week 4. Not a thing needed to be maintained. It was wonderful.

30-08-2012 09:01

Brandon Holley

Re: pic: 2012 frc971 transmission

I'd guess we're venturing into a thread takeover at this point...We utilized 5mm, 15mm width HTD belts with great success this year. We ran these on 24tooth pulleys.

-Brando

30-08-2012 09:12

Brian Selle

Re: pic: 2012 frc971 transmission

What width/size pulleys did 971 use? HTD or GT2? Are these 3.5" wheels?

30-08-2012 13:22

roystur44

Re: pic: 2012 frc971 transmission

The SDP-Si catalog will drive you crazy!!!

https://sdp-si.com/eStore/..%5Css%5CPDF%5C80502171.pdf

A 6A55M024DF0910

Yes custom 3.5" wheels. The thread slips into a slot and is tie wrapped around the axle. Quick and easy to replace threads

30-08-2012 13:59

Brian Selle

Re: pic: 2012 frc971 transmission

Thanks! I saw how your treads worked at worlds... very nice.

30-08-2012 17:51

AdamHeard

Re: pic: 2012 frc971 transmission

I'm really curious to know more about people's specific implementations with timing belts, please post about it in this thread;

http://www.chiefdelphi.com/forums/sh...04#post1183704

11-09-2012 08:15

greasemonkey

Re: pic: 2012 frc971 transmission

what is the reduction on that?

13-11-2012 01:11

AustinSchuh

Re: pic: 2012 frc971 transmission

One of the main reasons was because we could then loc-tite the pulleys on to the wheel assembly easily and get a strong enough bond to not need anything else. Chain would have been a bit more of a pain to do that with.

As stated in other places, we have had good experiences with timing belt in 2010, and wanted to get rid of more chain.

We are looking at making a couple of minor tweaks to this design to make it easier to swap wheels, move the encoder to the center wheel/last reduction, and maybe make it lighter. This design served us pretty well this last year.