View Full Version : belt drive VS chain drive
VPpanther
19-02-2015, 20:54
Hi
we are building a robot for a special project. the importance here is durability, we are hoping to build this for it to have minimal maintenance (we will be handing it over to people who don't know their way around robots). It will be a full fledged FRC robot. Our main questions, emphasizing durability:
1. chain or belt drive?
2. Colson high performance wheels or just the regular wheels
3. is there a specific type of drive train we should go for?
piersklein
19-02-2015, 20:57
Belt drive: belts don't stretch and I've only ever seen one break
Colsons: Much, much, much grippier and more durable than standard KOP wheels
6 wheel drop center: reliable drive system you shouldn't have much trouble with
6 wheel drop center: reliable drive system you shouldn't have much trouble with
Before going that route...
Is there any particular application for this robot? Terrain you expect it to encounter?
And I will note that I've seen teams have a lot of trouble with 6WD drop center. Not in the 6WD drop center part, mind you, in the execution of said 6WD drop center.
Andrew Lawrence
19-02-2015, 21:13
You can make anything work well if you put enough work into it. If your team has the means, I would suggest trying everything you can. See for yourself what works well for your team and what doesn't. Try and make a modular design that will allow you to swap out parts to see what will work better for you. For the most part, there really isn't a universal "better" design choice for teams.* You can be successful with anything if you try hard enough.
*While there is no "end all be all" best solution, there are some that are generally more effective than others given the circumstances of FRC and the style of competition. Of course, those are all case dependent.
VPpanther
19-02-2015, 21:14
thank you for the quick responses
my first guess would have been that chain would be more durable because its metal, could you enlighten me on that?
and I'm not sure if we could do a drop center chassis because the rocking back and forth would severely impact what we're trying to do. The robot would probably only have to face level surfaces most of the time with the odd bump now and again (its actually for a sponsor to take to tradeshows and events to advertise). Would it be bad to just have all 6 colson wheels touching the floor at ounce? (i was leaning toward what 610 did last year but on a larger scale: http://prntscr.com/67dvyj )
We have a very tight budget so we sadly don't have the opportunity to test solutions before implementing them
and I'm not sure if we could do a drop center chassis because the rocking back and forth would severely impact what we're trying to do. Would it be bad to just have all 6 colson wheels touching the floor at ounce?
What ARE you trying to do, in general terms?
What is the expected floor?
What sort of upper elements do you anticipate having--precision or "meh, whatever, it's close enough" in terms of motion?
Without knowing those, it's not going to be very easy at all to give anything like a "correct" answer, not that one exists. Knowing those, we can steer you around the pitfalls and away from any "incorrect" answers. This is especially important if you've only got one shot.
I would probably go with you don't want 6 colsons on the floor simultaneously, but that would also depend on wheelbase vs trackwidth--the higher the trackwidth/wheelbase ratio, the more "stick" you can get away with in your wheels.
Tom Line
19-02-2015, 21:21
Belt drive: belts don't stretch and I've only ever seen one break
Colsons: Much, much, much grippier and more durable than standard KOP wheels
6 wheel drop center: reliable drive system you shouldn't have much trouble with
Do you have data showing a colson wheel has a higher coefficient of friction than a traction wheel?
VPpanther
19-02-2015, 21:56
What ARE you trying to do, in general terms?
What is the expected floor?
What sort of upper elements do you anticipate having--precision or "meh, whatever, it's close enough" in terms of motion?
Well the robot is for a sponsor, they will take it to tradeshows and events to advertise their products. so the robot will most likely be facing clean level surfaces with the odd bump every now and again. upper elements wise it will have a humoniod type of structure with a head,shoulders, and arms so leaning toward precision. right now our dimensions are looking to be 34'' by 40'' and about 7-9 feet tall (the upper portion will be very light consisting of mostly sheet metal, so it will be very bottom heavy)
That's a bit better. 6WD colsons will probably work (and if it doesn't, you might be able to find some Lunacy wheels on AndyMark and use those--they're slick enough). That wide/long of a robot, you can probably get away with "flat" configuration, though I'd actually suggest using "mixed wheels", with at least one end being smoother/less "sticky" than the rest.
A couple of other items to bear in mind: I'd go single-speed transmissions, geared for a "reasonable" speed (say about 10 ft/second or less), but with two motors on each (in case of one failing). We're dealing with foot traffic here, so precision driving could be important. Also, some form of safety system should be on there, say a drivebase lockout (again because we're dealing with lots of people).
I'd agree on the superstucture being more "precision" than not; after all you don't want it rocking onto some potential customer's head while under the control of your sponsor.
VPpanther
19-02-2015, 22:35
oh okay, thanks for all the insight. I'll relay your suggestions to the team. much appreciated!
my first guess would have been that chain would be more durable because its metal, could you enlighten me on that?
Belts generally have steel or fiber glass tension members, which are pretty strong and keep them from stretching.
Quieter and lighter than chains, too.
I'd go with belt drive. It doesn't sound like there is going to be a whole lot of abuse going on. Maybe look into going with an andymark am14u. Easy to build and seems more than stout enough for what you guys are doing. That and no worry about needing hub adapters or anything.
Well the robot is for a sponsor, they will take it to tradeshows and events to advertise their products. so the robot will most likely be facing clean level surfaces with the odd bump every now and again.
I know this may sound rather strange, but kiwi may actually be the best answer for this set of requirements, and will bring a bit of cool factor as well.
upper elements wise it will have a humoniod type of structure with a head,shoulders, and arms so leaning toward precision. right now our dimensions are looking to be 34'' by 40'' and about 7-9 feet tall (the upper portion will be very light consisting of mostly sheet metal, so it will be very bottom heavy)
Honestly, i'd shoot for plastic rather than sheet metal if you man manage it. Not only is it lighter, but it's friendlier/less intimidating. For all the negative things that Douglas Adams had to say about Sirius Cybernetics, I have to admit that "Your plastic pal who's fun to play with" sounds a whole lot better than "Your sheet metal pal who's fun to play with".
Belt drive: belts don't stretch and I've only ever seen one break
For the record, chains don't stretch unless exposed to many times their rated load. The holes in the large links do expand a bit, which tends to increase effective chain pitch, and produces an effect which looks like stretch.
Also for the record, timing belts do break. This is why most recent automotive maintenance schedules recommend replacement after about 90,000 miles. Given that autos are normally driven at something close to 60 miles per hour, this corresponds to approximately 1,500 hours of operation.
The big point that will probably decide what is best for your current application: belts are much quieter!
Scott Kozutsky
19-02-2015, 23:22
It's all about implementation. We had belts on our versachassis WCD last year and we switched to chain because the belts were breaking (they were waaay out of spec. load so that makes sense). The equivalent pitch diameter and width of 25 chain was much stronger and performed fine (24t 9mm belt vs 18t 25 chain).
My experience with chain and belt:
Belt is lighter, do not stretch, are quieter, and more efficient than chain.
Chains requires less precision, are much smaller for the same strength, are much easier to source (especially up here in Canada), can be easily modified to different lengths, have many more available sprocket sizes, you can easily make any sprocket you can't source if you have a CNC mill, and chains can be easily bolted to for a function.
We like the compactness and general ease of use of chain so we decided to use it over belt this year.
Sam_Mills
19-02-2015, 23:29
my first guess would have been that chain would be more durable because
Steel is more ductile than the fiberglass or kevlar that you will find in a lot of timing belts. Which is "more durable" or "stronger" depends on material properties which is an entire science in and of itself.
Just know that chain will stretch with use, and is harder to replace for someone who has no experience if you are just handing the drive off to a separate group. More information about the specific use of the bot would be helpful if you want the best possible answers.
As for wheels, Colsons all the way. They wear slower than Hi-Grips and have a higher friction coefficient with carpet. If the robot isn't competing you may want even higher durometer wheels than Colsons though, so that they wear even slower. Colsons can last 100+ matches on carpet, but if its on pavement at a demo, you will eat through an 0.125" drop before you know it.
As far as general drive trains go, I would recommend designing a WCD (cantaleivered wheels, drop center, 6 wheels, shifting gearboxes). We made a nice one in the offseason. For this year's bot, however, we simply applied the concepts learned from WCD (slots, tensioning, bearing blocks) and made a standard 4WD that has independently controllable wheels for switching to mecanum or "regular". The WCD gave us a lot of experience for the effort.
roystur44
23-02-2015, 13:50
Belt has a much better damping effect than chain. Try spinning a chain system and a belt system at a high rate of speed.
Belt is easier to keep aligned over a long distance between axles.
and I'm not sure if we could do a drop center chassis because the rocking back and forth would severely impact what we're trying to do. The robot would probably only have to face level surfaces most of the time with the odd bump now and again (its actually for a sponsor to take to tradeshows and events to advertise). Would it be bad to just have all 6 colson wheels touching the floor at ounce?
Based on this, I'm not entirely sure you understand the complete purpose of drop center. Without drop center, using Colsons, and given the dimensions of the robot you specified earlier, your robot will have an awfully hard time turning. Drop center increases the ratio of width of wheelbase to length of wheelbase, decreasing wheel scrub while turning. A 1/8" drop is pretty much unnoticeable, but if you really can't have any rocking, consider using omni wheels.
Without drop center, using Colsons, and given the dimensions of the robot you specified earlier, your robot will have an awfully hard time turning. Drop center increases the ratio of width of wheelbase to length of wheelbase, decreasing wheel scrub while turning.
Technically, you are correct in the reasons. However, you need to rethink the robot dimensions, specifically the length/width ratio (assuming 34" wide, 40" long).
They're 34" x 40". 1.17 length/width. A typical FRC robot that's difficult to turn has tended to be a 1.35 length/width ratio (28"x38"), and before that a 1.2 ratio. They'll have an easier time than most FRC robots, particularly given that the actual wheelbase will be shorter than 40", probably more like 36" which will give nearly a 1:1. If the ratio of length/width (actually, wheelbase/trackwidth) goes under 1, turning suddenly gets easier without a need for a dropped center. Ask Ether for the physics on how that works sometime, I'm a touch rusty.
There's a reason you rarely see wide-bots using drop-center--it doesn't really help their turning, and it adds to their overall "oops, I fell over again" factor.
Scott Kozutsky
23-02-2015, 21:48
Technically, you are correct in the reasons. However, you need to rethink the robot dimensions, specifically the length/width ratio (assuming 34" wide, 40" long).
They're 34" x 40". 1.17 length/width. A typical FRC robot that's difficult to turn has tended to be a 1.35 length/width ratio (28"x38"), and before that a 1.2 ratio. They'll have an easier time than most FRC robots, particularly given that the actual wheelbase will be shorter than 40", probably more like 36" which will give nearly a 1:1. If the ratio of length/width (actually, wheelbase/trackwidth) goes under 1, turning suddenly gets easier without a need for a dropped center. Ask Ether for the physics on how that works sometime, I'm a touch rusty.
There's a reason you rarely see wide-bots using drop-center--it doesn't really help their turning, and it adds to their overall "oops, I fell over again" factor.
Outside dimensions =/= wheelbase. If you're running a west coast drive you've likely got a slightly larger track width and shorter wheelbase than an otherwise equivalent sheet metal drivetrain for example. Other factors like wheel size and center of gravity are major factors in both turning and tippyness, mainly because of how they affect the wheelbase.
From my experience on 1310 (2013), 4wd colsons even in a wide config can cause turning problems. Keep in mind that our high gear was too fast for a 4 cim drive that year but turning was still an issue. (our wheelbase must have ended up at like 2:1 wide but I can't remember the actual dimensions unfortunately)
I have yet to hear a team that used 4wd all traction tank say that they liked how it drove.
Most competitive wide robots I know of used 6wd drop center. 254, 2056 in 2014; 254, 1114, 469, 33 in 2013.
I'd really like to see where you got those stats. I haven't heard of many teams running 4wd long, let alone teams doing it effectively. At least without omni or some other non-tank solution.
You haven't been in FRC for 12+ years. Trust me, 4WD used to be the most common drivetrain around, maybe with the exception of 2WD. This was before about 2005, when the 6WD drop center really started gaining ground; 2004 was when the WCD really hit the FRC world.
Scott Kozutsky
23-02-2015, 22:07
You haven't been in FRC for 12+ years. Trust me, 4WD used to be the most common drivetrain around, maybe with the exception of 2WD. This was before about 2005, when the 6WD drop center really started gaining ground; 2004 was when the WCD really hit the FRC world.
That's fair but similarily these things evolved away from that for a reason. I've been told that casters were once very popular but I haven't seen one compete at all.
Sorry about rewording my comment. I just figured i could get my point across better with different wording and examples but that was after I already pressed submit. :ahh:
Chris Fultz
23-02-2015, 22:16
Check out this -
http://www.chiefdelphi.com/media/papers/2216
and this -
http://www.chiefdelphi.com/media/papers/2246
Check out this -
http://www.chiefdelphi.com/media/papers/2216
and this -
http://www.chiefdelphi.com/media/papers/2246
I saw that before, but I never found a test for 9mm belt, at least not on the load ratings of it. I don't like belts because thay are so wide.
That's fair but similarily these things evolved away from that for a reason. I've been told that casters were once very popular but I haven't seen one compete at all.
Sorry about rewording my comment. I just figured i could get my point across better with different wording and examples but that was after I already pressed submit. :ahh:
No problems there. I think I can counter your examples though. It might help your case to use teams from 2012 and prior; 2013 and 2014 were Frame Perimeter sizing rules rather than Fit in the Box sizing rules, which former tended to result in square-ish robots instead of longer rectangles like happened before the Frame Perimeter came to be used.
You see, there's a big difference between a 28" long robot that's 38" wide and a 28" square or some such similar size robot, something about a longer lever arm to turn the robot. I know I saw one really tippy one, back when the 28" by 38" box was first used in 2005, but I can't recall the drivetrain. (I do know they had wheely bars--saved them more than once! If anybody can recall, it was 393 competing at Sacramento, I think. No, they aren't on TBA--but I distinctly remember that regional having a bunch of non-West Coast teams.)
I have actually seen a few caster-bots competing, and I've actually seen one on Einstein, one of the last competitive ones. :yikes: Folks have figured out that these things called omni wheels are better than casters, and if you don't have omnis, you really want a peg of HDPE or something like that before you reach for the casters. The HDPE peg might get snagged, but it won't try to control your robot for you!
Scott Kozutsky
23-02-2015, 23:54
It might help your case to use teams from 2012 and prior
I have no real experience prior to 2010. Both 2010 and 2011 had no real reason to be wide so there weren't lots of wide bots then. The best wide bots I can think of from 2012 all had more than 4 wheels (though usually 8, probably for the bump). 33, 610, 341, 118, 67.
Since I didn't go to champs that year I don't really know some of the other good widebots though.
I understand that 4wd wide can work but I just think it's a suboptimal solution, especially with very high traction wheels like Colsons.
vBulletin® v3.6.4, Copyright ©2000-2017, Jelsoft Enterprises Ltd.