Timing belt in drive success

[AdamHeard]

Splitting off of this thread here;
http://www.chiefdelphi.com/forums/sh...hreadid=107998

I'm curious to know the following (please, complete data sets so it's useful) from teams who have run timing belt in drive

-Was it 1:1?
-Toothcount on pulley
-Profile and Pitch
-Belt Width
-Wheel Diameter and Type (roughtop, colson, etc...)
-Gear ratio before the pulley (high and low gear)
-tensioners?
-If exact c-c, did you add/subtract from this number? Also, what type of machine was the part fabricated on.

Once some information is posted, I'll tabulate it for comparison.

973 is prototyping belts for a wcd this fall (after exclusively using belts over chain in the 2012 season), and will update with our results.

IF you could post directly in this google doc that'd be great!
https://docs.google.com/spreadsheet/...2J4M 2c#gid=0

[scottandme]

Great idea:

Helpful resources I've been using up to this point in planning:

GT2 Design Manual - http://www.gates.com/catalogs/file_d...cation_id=2999

Light Power and Precision Design Manual - http://www.gates.com/catalogs/file_d...cation_id=2999

The main catalog index - http://www.gates.com/catalogs/index....cation_id=2999

The torque/HP tables have been especially helpful, and can probably go a ways in explaining ratcheting (assuming proper tension/spacing).

The Gates FIRST page also has some good basic references/calculators - http://robotics.gatesprograms.com/first

[protoserge]

FRC836 (The RoboBees) used a Gates timing belt for drive in 2012. It was fairly reliable, considering it was a tensioner/idler system (we had one belt go off pulley three time during the season in competition due to ball interference). Ideally, you should design the system to not have this "feature".

The biggest issue with timing belt drive is supply. We had optimized the drive (pulleys and belt lengths) only to find that Gates and their supplier were out of stock. It was a frantic back-and-forth telephone ordering process to get a solution that worked.

The above post has a lot of great resources. I would highly recommend using the off-season to develop a size chart for belt length with variables of drivetrain dimension and the torque/speed requirement (determines pulley diameter/teeth) to make the ordering process as painless as possible come build season.


Specs (placeholder for later)
Rear-biased, centrally mounted AndyMark SuperShifters
60 tooth pulley (6 wheels all had this pulley, rear axle drove center axle)
92 tooth pulley (main drive to rear axle)
25 tooth pulley at AM SuperShifter

This is off memory, but here is a general layout of the drivetrain (I'll have to get a photo).


- — - — - — - — - — - —

Rear-----AM---------Front
|90|-----25--------------
|60|---------60
-------------60--------60

[ratdude747]

I've seen toothed belts both work well and completely fail.

The time I did see it work well was on 2783's 2009 Unicorn drive... The drive belts were between the CIMs and the crab modules (which had a geardown).

The time I saw it fail was on 1747's 2010 drive... they went between supershifters and 8" plaction/plastic Omnis. The belts skipped teeth like crazy and eventually shredded. They were replaced with 35 Chain which solved the problem.

What it came down to IMHO was that toothed belts do not torque well... They are better at higher speeds, making then better for interfacing motors to transmissions, not transmissions to wheels.

[R.C.]

Originally Posted by ratdude747

What it came down to IMHO was that toothed belts do not torque well... They are better at higher speeds, making then better for interfacing motors to transmissions, not transmissions to wheels.

Disagree,

If you look at the gates spec sheet you'll see that belts can handle some really high loads. Can easily handle FRC loads but you have to size appropriately.

What size belts were you using? What pitch? Width? Pulley count?

-RC

[ratdude747]

Originally Posted by R.C.

Disagree,

If you look at the gates spec sheet you'll see that belts can handle some really high loads. Can easily handle FRC loads but you have to size appropriately.

What size belts were you using? What pitch? Width? Pulley count?

-RC

IIRC it used all KOP sized belts and Pulleys. The ones with the really shallow teeth.

I still think that chains are better for high torque situations...

[AdamHeard]

Originally Posted by ratdude747

IIRC it used all KOP sized belts and Pulleys. The ones with the really shallow teeth.

I still think that chains are better for high torque situations...

This is unfounded opinion though. What is better? I can spec a chain and belt to handle the same power output no problem. This kind of opinion on chief could potentially misinform less experienced users.

We used Gates gt2 belts exclusively on our 2012 robot on nearly every system (ranging from tiny tiny 2mm to big 5mm) without issue. The gt2 profile is FAR superior to generical XL/MXL belts people are probably used to, and are better than it'd predecessor from gates (HTD) which comes in the kop

Gates provides all the necessary equations to determine if the belt is strong enough. If the belt fails, you did something wrong; either it was under/over tensioned or it was simply too small to use (both are verifiable items).

[trilogy2826]

Originally Posted by AdamHeard

We used Gates gt2 belts exclusively on our 2012 robot on nearly every system

Does anyone have experience manufacturing the GT2 sprockets? Gates specifically warns against using their CAD to make physical sprockets. The available sprockets seem inefficient from a weight perspective. We have the tools to do this, but without being able to confirm the tooth profile, it is a risk. Either way, I will make some up and let you know how they work out.

[AdamHeard]

Originally Posted by trilogy2826

Does anyone have experience manufacturing the GT2 sprockets? Gates specifically warns against using their CAD to make physical sprockets. The available sprockets seem inefficient from a weight perspective. We have the tools to do this, but without being able to confirm the tooth profile, it is a risk. Either way, I will make some up and let you know how they work out.

We are machining some 24T 5mm gt2 9mm width pulleys currently that we will be using on a prototype. I'll update with results.

We cut 2 test items on our router that felt really nice on the belt by hand, and the full set is being cut on a production cnc so I can only a high quality.

We cut off of the sdp-si CAD.

[s_forbes]

Originally Posted by AdamHeard

We are machining some 24T 5mm gt2 9mm width pulleys currently that we will be using on a prototype. I'll update with results.

We cut 2 test items on our router that felt really nice on the belt by hand, and the full set is being cut on a production cnc so I can only a high quality.

We cut off of the sdp-si CAD.

I will be interested to see how that turns out.

Our belt system was sketchy last year, but sufficient enough to last through a regional and champs with no failures. Lots of lessons learned, primarily that the tension required in the belts necessitates a stiff frame (our fiberglass material is good stuff, but flexes way too much). Especially with a serpentine path and cantilevered idlers.

Interesting to note, we used 5mm pitch GT2 belt profile coupled with HTD sprockets. Not optimal, but functional enough to work well if the rest of the system is designed properly. We did this for cost purposes, the HTD sprockets were given to us as a sponsorship from Gates. Had we used the proper profile for the pulleys the drivetrain may have performed better, but it worked great at competition.

An entertaining video from when we were testing the first iteration of our idler system. We fixed this with a better idler design.
http://www.youtube.com/watch?v=8YQk0fjhk-o

[IndySam]

Originally Posted by AdamHeard

We are machining some 24T 5mm gt2 9mm width pulleys currently that we will be using on a prototype. I'll update with results.

We cut 2 test items on our router that felt really nice on the belt by hand, and the full set is being cut on a production cnc so I can only a high quality.

We cut off of the sdp-si CAD.

Hmmm, we may have to try and waterjet some as an experiment. Being able to cut the profile and the hex might be a huge savings in time and cost.

Wonder if we could cut them out of 1/4 stock and stack them.

[trilogy2826]

Originally Posted by AdamHeard

We cut off of the sdp-si CAD.

Thanks for the point in the right direction. I was assuming (why do I keep doing that) that the Gates proprietary profile for GT-2 would not stray past their walls. I did an overlay of the SDP/SI 24T and the Gates 24T and they are indeed different in tooth geometry. Not much, but just enough to heed Gates warning about not using the CAD to machine.

We will be running a batch using low-voltage EDM, which will (I believe) give the most accurate resolution and accuracy. I am hoping the smoothness spec can still be met with this technique. We will probably be teflon coating these. so it should not matter. If you want to trade an EDM for a milled sprocket sometime, let me know and we can compare/contrast.


Also, if you don't go down the Shaker sprocket route and want to add flanges, what is your technique? Our current method is to laser a flange, press in the flange angle guide, and bolt it to the sprocket. If there is a way to do this without post machining the sprocket, that would be handy.

Originally Posted by IndySam

Hmmm, we may have to try and waterjet some as an experiment. Being able to cut the profile and the hex might be a huge savings in time and cost.

Wonder if we could cut them out of 1/4 stock and stack them.

We tried the following techniques in the 2011 and 2012 offseason to manufacture sprockets and gears:

1. Laminate plates riveted, glued or bolted together to a consistent thickness: Laser and waterjet 0.04, 0.08, 0.125 and 0.25 6061 AL and Steel. Result: Any thickness over 0.08 resulted in a kerf that ended up damaging the belt. 0.04 and 0.08 cut accurately, especially in steel, but the multiple laminations required for a given thickness were difficult to align, even with a complex process and a custom jig. Moreover, the laminations presented a "roughness" that drastically decreased the life of the belt. Overall, not recommended.

For what it's worth, this technique has worked very well for us in gear manufacture from small to large (thanks 148 for the inspiration)

2. EDM the sprocket profile to the desired thickness. Works great with gears and I will be trying them on GT-2 belts in the next couple weeks

I would like to do a cost/benefit analysis of make VS buy and machine Vs EDM... eventually.

[AdamHeard]

Jake, it must be awful nice to be able to get the finish from the EDM.

I'm hoping the milled profile works for us, I can't imagine it won't with how low our runtimes are (even on practice bot) compared to industry. I wouldn't be surprised if it leads to a bit of belt wear. I'd be willing to swap some down the road.

Our dream is to run 5mm pitch gt2 9mm wide in 24T with exact c-c; per the math this is marginal, so our expectations is that our prototype will fail. We would then run with tensioners.

We're machining our pulleys in one shot into a large block (as attached). We then will hold these on a mandrel to lathe the rear side to length and diameter at size. At this point a flange is generated. For the other flange we will bolt on a .032" plate (easy for us to router) with #4-40s.

987 had a very cool process for their large pulley that we plan to copy for large and less critical pulleys in the future. They picked the drillsize closest to the proper radius for the root of the teeth, and drilled all the teeth. They then used a mill cutter capable of an undercut to remove the top of this material all the way around leaving a flange.

We used the same process as you for cutting plate gears this season, in 24, 20 and 16DP. They initially look nasty, but quickly wore smooth. We LOVED it, and probably cut over 100 gears. I'm sure you guys did as well.




RC, for 2mm and 3mm pitch gt2 we primarily used 6mm wide. I really need to clean up Encore's CAD to a quality I can post so teams can see how timing belt is really easy to effectively integrate. Utilizing this smaller belts (which are still KILLER strong if you use them right) we were able to package much larger reductions in smaller packages than possible with #25 chain.

[Chris is me]

Originally Posted by trilogy2826

Does anyone have experience manufacturing the GT2 sprockets? Gates specifically warns against using their CAD to make physical sprockets. The available sprockets seem inefficient from a weight perspective. We have the tools to do this, but without being able to confirm the tooth profile, it is a risk. Either way, I will make some up and let you know how they work out.

We bought pulley stock and machined it down at home. Not too hard to work with. You'll want a number of teeth divisible by three if you can in order to make it easier to hold in a three jaw chuck, but it machines easily and you can easily remove the hub or any other features you might not want.

This year, since we do belts-in-a-tube, we just had a single piece of pulley stock both belts ran on. Worked great.

[steelerborn]

A lot of people have a bias toward chains, and like to give belts a lot of negatives they don't really have. Just do some research and you will be surprised all of the different applications belts were used in.

I would still use a tension system only because it makes mounting much easier. I used sdp-si for all my designs my only negative is sometimes they are out of stock for belts sprockets and can take quite some time for you to get your parts, and that would be horrible if it was in the middle of the build season . Just got to be sure that you have a good selection of sprockets in stock or modify your design for what is in stock.

Just out of curiosity what width belts does every one use, I see alot of 3/8 inch belt out there. But I think you could probably go smaller without the fear of breaking a belt.