The last two years I have been the lead of my team’s design team. I am graduating and it is now my job to make sure that I pass on all of the knowledge that I have acquired. While going over through what I have learned I have been trying to site where I learned it and why it is important.
During the time I have been leading the design we have been using center to center chain runs on our drive trains. I remember reading that one team (2363?) posted that you should add .012" for #35 and .016" for #25 to the center to center distance to pre-load the chain. Over those two years we had zero issues doing that to our drive train chains (both years were #25 between 18 tooth sprockets) and have had zero chain failures. As of recently I have been wondering if the pre-loading of the chain is helpful, or causing it to wear out quicker. What are the advantages and disadvantages of this practice?
Additionally what risks does loose chain present in drive trains?
I had a long post explaining a bunch of the concepts you’ve touched on, but instead I’ll leave you this document and recommend you read it, understand it as best you can, and pass it on to the students that will be filling your role moving forward. It’s hugely helpful. The gates design guides are as well if you’re ever looking at belts.
The biggest risk of loose chain in a drive train or any mechanism for that matter is it becoming so loose that it comes off the sprocket. This becomes dangerous in situations where you are using either a double sprocket or two plate sprockets side by side as the loose chain can come off and lodge itself in a position that locks the chain run.
Depending on how key the chain run is, it coming off can either render a mechanism useless or impair function partially or severely. This is one of reasons teams have switched over to direct driving one wheel off of their drivebase gearboxes as running chain from your gearbox to your wheels means if one of them becomes so loose it falls off, you lose your drivebase. In other situations it doesn’t matter.
As an example, before our Einstein match against the Daly alliance, as our driver and I lifted our robot off the cart I reached past the lift point and found completely slacked chain on one side. It was too late to fix so I rotated the chain until it came off both wheels. Since our setup was traction in the back directly off our gearboxes & omnis in the front, losing this chain didn’t impair our function too much. Our driver noticed some difference but it wasn’t hard to overcome. With such a long run of chain we’ve had to adjust the versablocks the front wheel sits on usually once per event by a slight amount. The longer your chain run, the more “stretch” you will see over the course of it’s use. You see this occur as each link of chain wears down ever so slightly and the pins loosen up. Additionally your sprockets will wear down slightly which creates this slack teams see overtime.
Having a method for tensioning your chains is important. Either by adding a calculated center distance like you have used, a floating adjustable tensioner, cam tensioner like most West Coast Drive setups, or slotted motor mount for smaller mechanisms. Keep in mind that a calculated center to center chain run is only good if you can hold those tolerances while machining it. Don’t attempt this for the first time in season without checking your work. ::rtm:: Last season we attempted this with #35 chain and we had a nasty time at our opening two events with many of the issues described above. By our third event each wheel had a floating tensioner but this wasn’t an ideal setup and added unequal friction on both sides of our drivebase but it was necessary at the time. This is what drove us to us to use versablocks this season so we had a more reliable setup. Einstein was our first failure of the year from an incomplete pit check.
Ensuring you have proper alignment between your sprockets is extremely important. Sprockets that aren’t in line with each other will throw loose chain in a heartbeat under load.
I wouldn’t be as worried about the life of the chain, so much as its performance. FRC robots have relatively short lifespans compared to nearly every other machine on the market.
I personally don’t have the data to say whether or not pre-tensioned chain actually wears quicker. But all chain stretches (because of wear in the roller bushings), so not having the pre-tension leads to saggy chain after run-in. Adding the pre-tension may cause it to stretch a bit more (or maybe not), but it ends up canceling out the stretch giving you a pretty decent chain run. I know many teams have run pre-tensioned chain, and I don’t see that stopping any time soon.
The major danger to loose chain is skipping teeth. Once it gets beyond that point, it can actually jump off the sprocket. That needs very loose chain, though. As lower levels of stretch, you’ll run into problems with lots of backlash, which will make programming autonomous routines and closing the loop on drivetrain velocity harder.
Thanks for the advice. We do have a CNC mill that has the ability to hold +/- .0005 which has made center to center runs possible. I am interested in chain wear and slack because of the lack of tensioning in the system.
As of now we have been replacing the chain after two district events. This is a preventative measures to keep the chain from skipping links.
Totally agree service life stated in manufacturer specs conform to industrial applications and are likely serious overkill for FRC.
Per the spec from diamond above you’ll find the following:
-Chain should have some slack at mid span on the slack side. (We don’t really have a slack side given our application and reversal of loading in drive trains)
-Chain stretch occurs disproportionately at the beginning and end of service life.
-Chain not subjected to pre-tensioning (in this case referring to a process to achieve the early pin/bushing wear before installation) does wear quicker that chain processed that way prior to use.
Ultimately an adjustable static tensioner is the easiest way to combat the problem of slack. Spring loaded tensioners are generally best to be avoided, because they don’t respond well to reversal of loads and they add extra tension to the chain, which accelerates pin/bushing wear.
Companies do sell chain that’s been put through an industrial process to do a controlled break in of pins and bushings. This might be a good option, as if it’s installed at the proper spacing without pretension you’d likely experience less stretch over the service life on your robot. I’m not aware of any team running roller chain to a wear based failure mode. Generally it’s a tensile failure from a shock load or some other less characteristic catastrophic failure mode.
Hopefully Paul sees this thread and can shed more light but in his presentation he states that for #25 adding .018 is correct and for #35 adding .012 is correct. He also states that this is not per x amount of links (his example in the presentation has 62 link run), he states that the tolerance stackup maxes out at around a foot he recalls. He did say that on really long #35 runs you may want to jump it up to .018.
I’d personally like to have him confirm the information about the tolerance stackup maxing out just to make sure I understood him in the video correctly.
The testing done by 2363 resulted in similar numbers for the #25 chain, with .019-.020. They also support with their data that the length of the chain run does not effect the tension tolerance.
As an aside I will point out that there are probably application exceptions. For example in 2016 we had our chain runs outside the tube on the wheel side. We actually ran our C-C distances .018 under.
We never had an issue with our chain in competition, and we believe that the chain being loose was better as any hard hits from the defenses allowed the chain to better deflect out of the way due to being loose then breaking. We did have a chain link get ripped apart once while practicing that year, but we believe it was due to going over the Cheval-de-frise at an angle and having the pointed corner of one of the ramps catch on the chain.
In general however most teams I’ve talked to that run direct C-C chain lengths have had great success using the .018 value. We have also used it on our standard drive trains including this year and have had no issues with the chain all season long.
We used the .018" value this season and the chain we ordered from VP absolutely did not fit properly. I’m not sure exactly what the problem was there, but we ordered chain from AM, clearly a visually different supplier than we received from VP, and it worked on the same center distances. I’m curious what went wrong, but I’m still not clear on it. That said, we didn’t have to mess with it all season. Never broke a chain, never threw a chain, never shortened chain, etc.