Tread and brownouts

[mrnoble]

We are discovering that tank tread leads to significant systemic friction issues; not with the floor, but within the system. 12.3V batteries dip to 10V when the treads are free spinning, and down into brownout zone when on the floor and mildly pushing. Anyone else experienced this? What steps have you taken to alleviate the problem? We are currently working through the system (including gearboxes, which in our case are two sets of single reduction clamping boxes, 14:70 reduction, with the 70's meshing the two CIMs per side).

[mrnoble]

Other significant things to point out: we are running Brecoflex, and the pulleys are cantilevered, WCD style.

[Electronica1]

How many pulleys are you driving per side? Also, what is the diameter of the drive pulley?

[IronicDeadBird]

4 per side, roughly 4.4 inch pulleys trying to get caders.

[Electronica1]

Originally Posted by IronicDeadBird

4 per side, roughly 4 inch pulleys trying to get caders.

So, from what I understand, you only want one drive pulley per side in a tank tread system so they don't fight each other. You should be fine with a 4 inch pulley. How much tension do you have on the belts?

[mrnoble]

We are only driving one pulley per side; the others are "belted" to each other via the tread (no additional belts or chain). We are tensioning through a typical WCD setup, with a nautilus cam tensioner. Fully tensioned, the treads have about 1/2" of deflection along a 12" gap with about 50 lbs force applied at the midpoint. I haven't calculated the pounds force.

[cbale2000]

Originally Posted by mrnoble

running Brecoflex

Originally Posted by mrnoble

the pulleys are cantilevered, WCD style.

In my opinion, and having used treads (Brecoflex specifically) on a number of robots, this is asking for trouble. In theory you can make cantilevered pulleys work, but you have to have some VERY ridged shafts. The problem with doing this on tread drive is that any sort of flex in the shafts or change in the relative position of the pulleys can drastically increase the friction of the system, which can, amongst other things, cause increased current draw on the motors. That 1/2" of deflection may not seem like a lot, but what's happening is the contours on the back side of the tread is grabbing unevenly on the pulleys, causing higher friction than normal; this will also, very likely reduce the lifespan of the shafts/bearings/pulleys in your drive. Additionally, having the pulleys cantilevered leaves you more vulnerable to having a tread to pop/break off your drive.

Now, that said, your current gear reduction may also be a factor here, when my team used to use tank treads, we would gear the system to run at, maybe, 10fps tops when using 4 CIMs. In 2006 we got away with a bit higher speeds by adding a pair of Mini Bike motors ("Big CIMs") to the drive, but even then you should almost never plan to be one of the faster robots on the field while running tank treads. Lowering the reduction on the gearbox/belts may help offset the efficiency loss enough to save some battery life.

[mrnoble]

Actually what you said reinforces my theory on what is happening. I told some of the students last night that needle bearings might work better than ball bearings for the loads on the front and rear pulleys. We have a couple of backup options, one being to create an outer plate to take out the cantilever factor; the other being to run wheels instead of the treads. We will discuss our options today. Thanks for the advice, I wish we'd known then what we know now.

[Gdeaver]

What Brecoflex belt material are you using? We tested the PVC material and found to much friction with carpet. Very high draw while turning. We have 6 idlers with ball bearing between 2 plates. Friction loss and current draw is very low unloaded.

[jwfoss]

Two things I learned back in 2010.
- Check your "contact patch" if its too long you may have excessive scrub
- Increase the gear reduction to get more torque to overcome scrub

Can you please post the configuration of your drivetrain? I suspect that you have a large contact patch between your pulleys that contact the ground. You may need to create a "drop center" like most teams do with a wheeled drivetrain.

[mrnoble]

We are currently using the PVC belts, though the problem isn't scrub; we have a drop center, .1". I'm pretty sure the real problem is the cantilevered axles.

[Gdeaver]

Last week we tested the pvc belts. Way to much grip. Very high current draw while turning. That's with a non cantilevered design.We have gone with the polyurethane with a very slight 60 degree bevel on the sides. Is working very well.

[jwfoss]

Originally Posted by mrnoble

We are currently using the PVC belts, though the problem isn't scrub; we have a drop center, .1". I'm pretty sure the real problem is the cantilevered axles.

.100" drop is less than what we typically run on an 8wd with colson wheels. Please post up your drive configuration with dimensions so we can properly help you. Scrub is absolutely a factor in your problems, be it due to some deflection in the shafts, the c-c distance between your contact wheels, your gear ratio, or your drop dimension choice.

[mrnoble]

Thanks, I will post our design today.

[Brandon Holley]

Originally Posted by cbale2000

In my opinion, and having used treads (Brecoflex specifically) on a number of robots, this is asking for trouble. In theory you can make cantilevered pulleys work, but you have to have some VERY ridged shafts. The problem with doing this on tread drive is that any sort of flex in the shafts or change in the relative position of the pulleys can drastically increase the friction of the system, which can, amongst other things, cause increased current draw on the motors. That 1/2" of deflection may not seem like a lot, but what's happening is the contours on the back side of the tread is grabbing unevenly on the pulleys, causing higher friction than normal; this will also, very likely reduce the lifespan of the shafts/bearings/pulleys in your drive. Additionally, having the pulleys cantilevered leaves you more vulnerable to having a tread to pop/break off your drive.

Now, that said, your current gear reduction may also be a factor here, when my team used to use tank treads, we would gear the system to run at, maybe, 10fps tops when using 4 CIMs. In 2006 we got away with a bit higher speeds by adding a pair of Mini Bike motors ("Big CIMs") to the drive, but even then you should almost never plan to be one of the faster robots on the field while running tank treads. Lowering the reduction on the gearbox/belts may help offset the efficiency loss enough to save some battery life.

Originally Posted by mrnoble

Actually what you said reinforces my theory on what is happening. I told some of the students last night that needle bearings might work better than ball bearings for the loads on the front and rear pulleys. We have a couple of backup options, one being to create an outer plate to take out the cantilever factor; the other being to run wheels instead of the treads. We will discuss our options today. Thanks for the advice, I wish we'd known then what we know now.

Along a similar thread as the above posts...


With a typical WCD, you are usually tensioning chains/belts, so there is tension in that system as well. However you're also usually doing that within the comfy confines of the 2 bearings (chain in tube) or on the inside of the tube (far away from the long cantilevers of the wheels).

My concern with the setup you're describing is that you NEED solid belt tension for the system to work, however that tension is wreaking havoc on your drive efficiency. By not being able to keep the ends of your shafts all parallel to each other (with an outer plate), you may be seeing a fair amount of deflection thats causing some binding.

The key symptom I zeroed in on from OPs description is that even running free on a bench you're seeing relatively significant voltage sag, to me indicating you're drawing a large amount of current, even in an unloaded state.

Have you tried running on a bench with less or almost no tension?

-Brando