chain transmission without tensioning

I am now working on CADing a 6WD model. I want to use chain transmission without tensioning. From your experience, is it possible, given a high capability of CNC machining for the drive base channels?
Also, I was wandering what tolerance I have while designing it. Do I have to make the distance between the sprockets exactly the expected length? A little bit less? Is 1 mm deviation, for example, enough to make the chain fall, or make it be over-stretched?
I am particularly interested in using a 25 chain, but any answer that deals with 35 chain or belt drive will also be very helpful to me (other members of the team are now designing belt drives).

25 chain is very unforgiving for misalignment, spacing, and flexing of the structure. We were constantly dealing with chain jumps on our 2012 robot that used 25 chain.

35 chain is much more tolerant to each of those items, but significantly heavier. Our 2011 robot didn’t have traditional tensioners, but slotted holes for the external wheel mounts that allowed us to slide the wheel until the chain was “tight” but not stretched. We never had to adjust the entire competition season.

You shouldn’t design the center distance to be slightly larger than the expected chain length, because assembly will be nearly impossible and will put high tension in your chain. Slightly less distance will allow the chain to sag and will cause some amount of hysteresis (different response forward and backward) in your drive system. Again, 25 chain will be more likely to jump a tooth or fall off in this case compared to 35 chain, both due to the roller center distance, and the width of the teeth.

Unlike timing belts, roller chain tends to grow slightly in length as it “breaks in”. This is due to normal wear in the hundreds of axial joints present in the chain. Due to this, it is really not possible to machine the “perfect” center-to-center distance between sprockets. For short chains, you can arrive at a spacing that will work, and simply be somewhat more slack later in life than it was initially. Short chains can accomodate some slack without whipping or tracking probems. For most chains of typical length however, some tensioning method is needed to accomodate variations and growth in chain length. This can be adjustable C to C length, or an idler or tensioning glide that deflects excess slack in the chain.

We do this every year with 25, and it works really well. We machine the robot sides to the exact center to center distance for the sprocket, and it works well. If you chain does stretch, just throw an idler sprocket in there.

Is it a cantilevered drive system? If not, you can add an aluminum stand off with a piece of pvc/teflon/nylon/delrin around it. You want the ID of the plastic to be slightly larger than the OD of the stand off so it spins freely. Off set teh stand off and route the chain up and around it. Slot the holes if you need to.

In 2004 we ran #35 chain. Center distance between the wheels was 1/8" longer than nominal. We have a chain puller so we used that to pull the chain tight enough to put in the master link. Never touched it again. The chain stretched enough in the first few hours to take the strain off the bearings and by the end of the season there was a notably sag.

We run #25 chain on theoretical center distance without tensioners all the time. Alignment is key. The chain wants to jump when the sprockets aren’t axially aligned properly.

Placing these holes can be done with a manual mill. No particularly high tolerance process is required. ±0.005" will work all day long.

Example Chassis with untensioned #25 chain runs.

FRC118 ran this setup for 2013 with zero maintenance issues or chain failures.

Can’t you spring-load a sprocket to do automatic tensioning? (or am I missing the point here?)

You’re missing it–the point is that they don’t want to use ANY tensioning if at all possible.

That said, I’ve seen folks just put a loose sprocket into their chain runs that need tensioning–best one I saw was a rubber one in the drive chain; ran back and forth as needed to get a good tension. Though I do second the Delrin recommendation. If you happen to be running the chain over structural members, stretch the chain over a block of Delrin on one of them, and let the chain dig in a bit until it finds its happy zone.