Rhino track failier?

[Brandon Holley]

The glossiness and types of fractures I can tell from these pictures indicate the material may be glass-filled. Definitely would make the pulleys super stiff, but definitely suseptible to impact failures, as the glass in the resin makes the plastic surprisingly brittle.

Additionally, while the spec'd total tolerance may be 0.0005" (0.0127mm for those of us versed in those units), this is a basically an unmoldable range. It would be tough for an injection mold to hold that steady over many thousands of parts, even things like the HVAC of the factory or running at night vs running at day could have that effect. I envision few suppliers would sign up for that tight of a tolerance.

That being said, molded press fits have been common in AM and VP parts for years without issue- so this may not be part of the failure at all. Just wanted to add some realism to the discussion regarding tolerance and injection molding.

-Brando

[Kevin Sevcik]

Originally Posted by Brandon Holley

The glossiness and types of fractures I can tell from these pictures indicate the material may be glass-filled. Definitely would make the pulleys super stiff, but definitely suseptible to impact failures, as the glass in the resin makes the plastic surprisingly brittle.

AM specs the material as black polycarb. That's consistent with their other black plastic parts, none of which seem glass filled.

[Andy Baker]

All,

Thank you to the Ginger for starting this thread. During this past weekend, we have received emails from three teams concerned about the same issue. I won't list the team names, as they may consider it confidential information.

This post is a review what we are doing. I will post updates in this thread regarding how things are progressing.

Customer Service:
If anyone has cracked hubs for these Rhino Track Drive modules, please send a note to sales@andymark.com. We will send out* replacement pulley halves. It will help us if you take pictures of the pulley fracture and tell us which pulley it is. Please refer to the pulleys numbered as such in this picture:



In house testing:
We have a test mule here at AndyMark that has been tested many times. Our initial tests (before kickoff) did not show cracking. We drove this mule over the wooden defenses after kickoff and it seemed to do fine. As it drove more last week, seemed to have more resistance to driving the tread. Today, when we drove it, the 120a breaker tripped. We took it apart and found that there were pulley fractures on both sides, on pulleys 1 and 4. Pulleys 2 and 3 did not have fractures. They had a few hairline cracks, but nothing was wrong with pulleys 2 and 3 to affect performance. Here is a picture of our pulleys we took off today:



This picture shows that the front (#1) and back (#4) pulleys were fractured. Since the middle two pulleys fared much better, we think this fracturing is due to the higher forces on the front and back pulleys with regard to belt tension. We are building up the test mule with new pulleys in order to make a test today. We will be testing the chassis with a slightly looser belt tension to see if we can find a belt tension sweet spot between the belt not falling and the pulleys breaking.

Pulley improvements:
The bearing fit is good, in our opinion. It's a light press fit with no taper. The radius at the bottom of the bearing hole is smaller than the radius on the bearing. Since we are able to recreate this failure here, I don't suspect that a customer was inadvertently using thread locking material (which is always a no-no with polycarbonate).

Material: This material is black polycarbonate with no filler. We have tried fiberglass fill years ago with omni wheels, and we learned that while strength is increased with fiberglass, impact resistance is significantly reduced. We could try to make this part out of some sort of Nylon or a co-polymer using Nylon, but that would require an entirely new mold and would take much time to try.

Mold tool improvements: We can take away metal from the mold to add plastic to the pulley. We will look at improving the mold in 3 ways:

1. Adding a rib in the pocket between the mounting holes
2. Raising the counter-bored level for the screw head so that the top of the head is slightly under the face of the pulley (we believe that we can do this without changing the screw length)
3. Increase the radius on the ribs supporting the inner circle bigger

Corrective Action:
There is enough evidence here to justify that we need to improve this mold. Once we get a new part designed, I'll post a pic of it on this thread with the additions highlighted. Since our molder is here in Kokomo and works with us very closely, I am confident that we can make a mold improvement and have parts available within 8-10 days. Thank heavens this part wasn't made overseas.

* - While you may need cracked pulleys sent to you very soon, you may also wish to wait until we get this mold changed and have new, stronger pulleys. Please stay tuned for timing regarding when we can do this.

I am warning our ops and customer service folks that we may be sending out many replacement pulleys soon.

It will be an interesting week. Please stay patient with us as we solve this issue.

Sincerely,
Andy B.

[ChuckDickerson]

Andy, thank you for the prompt attention to this developing issue. I'm sure many teams are going to be watching this situation as it develops.

One additional data point request for Andy and others that have seen the failures. What was the total weight of your robot? Are these weighted 120lb machines or are they failing with lighter basic rolling chassis as well?

[team222badbrad]

We have yet to test our VexRhino (custom drive using AM pulleys and belts).

We will let you know if we have any failures.

[Richard Wallace]

Andy, I think Rhino customers really appreciate the open approach AM is taking to address this. I agree that a mold improvement is likely to solve the problem.
---------

Another approach, more in the quick-fix direction, might be to distribute the tension and shock loads on the pulley over a larger area. This might be accomplished by inserting a tight fitting sleeve (maybe aluminum tube) through the pulley bore (0.96 inch), and then sleeve type bearings (maybe oil-lite bronze) into the ends of the sleeve to accept the 3/8 inch bolt spindle.

I have ordered some pulleys to try this, along with aluminum tube and bronze bearings. Tube and bearing part numbers from McMaster-Carr are 9056K28 and 6338K465, respectively. I plan to turn the tube O.D. for a tight fit to the pulley bore, then bore (drill) it to a tight fit for the 5/8" bearing O.D., and face to 2.175" long. After a light press to fit the tube into the pulley, the bearings can be inserted and then the assembly should replace a damaged one. I don't know how much extra friction the sleeve bearings will add, relative to the 1416ZZ ball bearings -- current draw will tell that tale. If this approach works it might be used to repair damaged pulleys.

I am just an electrical engineer fiddling with mechanical things based on FRC experience. I would appreciate critique of this idea, especially from those with stronger Mechanical Engineering background and/or more FRC design experience.

[Andy Baker]

Thank you all for your understanding, support, and patience.

Originally Posted by Richard Wallace

Another approach, more in the quick-fix direction, might be to distribute the tension and shock loads on the pulley over a larger area. This might be accomplished by inserting a tight fitting sleeve (maybe aluminum tube) through the pulley bore (0.96 inch), and then sleeve type bearings (maybe oil-lite bronze) into the ends of the sleeve to accept the 3/8 inch bolt spindle.

I have ordered some pulleys to try this, along with aluminum tube and bronze bearings. Tube and bearing part numbers from McMaster-Carr are 9056K28 and 6338K465, respectively. I plan to turn the tube O.D. for a tight fit to the pulley bore, then bore (drill) it to a tight fit for the 5/8" bearing O.D., and face to 2.175" long. After a light press to fit the tube into the pulley, the bearings can be inserted and then the assembly should replace a damaged one. I don't know how much extra friction the sleeve bearings will add, relative to the 1416ZZ ball bearings -- current draw will tell that tale. If this approach works it might be used to repair damaged pulleys.

I am just an electrical engineer fiddling with mechanical things based on FRC experience. I would appreciate critique of this idea, especially from those with stronger Mechanical Engineering background and/or more FRC design experience.

Richard, this is brilliant! Thank you for posting the idea. We may try something similar here in our tests. You're an ME in disguise!

Regarding the mold changes, we are implementing 4 changes to the pulley.



Due to the speed of things happening, if folks want to post input here on what else to do to this Rev4, it may or may not get implemented.

Andy B.

[Kevin Sevcik]

Originally Posted by Andy Baker

Due to the speed of things happening, if folks want to post input here on what else to do to this Rev4, it may or may not get implemented.

Andy B.

Rev 4 improvements sound good, I think. Possibilities for the hopefully unnecessary Rev 5:

• If the cracks are starting at that thin section cleared for the bolt head, make it thicker by switching to #8 screws or flat-head screws. There would obviously be potential sourcing issues for either.
• If Richard's idea pans out, modify the mold to fit standard sleeve bearings all the way through. McMaster has 3/4" OD, 3/8" ID, 1" long sleeve bearings. That should thicken the wall plenty, but might transfer the failure to shock-loading of the oilite bearing.

Which reminds me to tell Richard to thoroughly shock-test his solution. I've seen some oilite bearings shatter on some of our subsystems before. There's a gray colored iron enhanced version that's more shock resistant that you might want to try. Your McMaster part number would be 2938T39, which is actually 0.20 cheaper per than your original bearing.

[Richard Wallace]

Originally Posted by Kevin Sevcik

Which reminds me to tell Richard to thoroughly shock-test his solution. I've seen some oilite bearings shatter on some of our subsystems before. There's a gray colored iron enhanced version that's more shock resistant that you might want to try. Your McMaster part number would be 2938T39, which is actually 0.20 cheaper per than your original bearing.

Thanks, Kevin!

One question: the bearing you cited is rated for higher loads, but at much lower RPM compared with the one I was looking at. Actually neither is rated for maximum wheel speed in the Rhino drive, but the 6338K465 is closer, rated for 700 lb at 250 RPM, while the high load bearing you cited is rated for 1500 lb at 80 RPM.

Are the RPM ratings critical here? I understand that the material composition of the high load bearing makes it less susceptible to shock.

[Kevin Sevcik]

Originally Posted by Richard Wallace

Thanks, Kevin!

One question: the bearing you cited is rated for higher loads, but at much lower RPM compared with the one I was looking at. Actually neither is rated for maximum wheel speed in the Rhino drive, but the 6338K465 is closer, rated for 700 lb at 250 RPM, while the high load bearing you cited is rated for 1500 lb at 80 RPM.

Are the RPM ratings critical here? I understand that the material composition of the high load bearing makes it less susceptible to shock.

I think McMaster's ratings are highly conservative. Based on PV and V max numbers I can pull off the web, the max SFM should be 225, which translates into about 2300 RPM. Not that you could handle a full robot load at that speed. What RPM max do those Rhino Drives run at? Give me that and I can give you a max load.

[Richard Wallace]

Originally Posted by Kevin Sevcik

What RPM max do those Rhino Drives run at? Give me that and I can give you a max load.

Max Rhino drive RPM is the team's choice. AM options for gearing with the TB Mini (easiest to mount on a Rhino module) range from 12.75:1 down to 5.95:1. Since the Rhino timing pulleys are 1:1 (42 teeth on drive and driven), track pulley speeds will be in the range 415 - 890 RPM. My team plans to use a ratio between these extremes, so about 670 RPM max is the target for us.

[Kevin Sevcik]

Originally Posted by Richard Wallace

Max Rhino drive RPM is the team's choice. AM options for gearing with the TB Mini (easiest to mount on a Rhino module) range from 12.75:1 down to 5.95:1. Since the Rhino timing pulleys are 1:1 (42 teeth on drive and driven), track pulley speeds will be in the range 415 - 890 RPM. My team plans to use a ratio between these extremes, so about 670 RPM max is the target for us.

PV of the SAE 863 is 35000. V @ 900 RPM is 88.5 SFM. So P is 395.5 psi. Projected area for one of those bearings is .375 sq in. So max load on one bearing is 148.3 lbs. You're using two, so max load per pulley is about 300 lbs. That's combined load from tension and weight, but I think you should be fine regardless.

[Matt C]

Originally Posted by Richard Wallace

I have ordered some pulleys to try this, along with aluminum tube and bronze bearings. Tube and bearing part numbers from McMaster-Carr are 9056K28 and 6338K465, respectively. I plan to turn the tube O.D. for a tight fit to the pulley bore, then bore (drill) it to a tight fit for the 5/8" bearing O.D., and face to 2.175" long. After a light press to fit the tube into the pulley, the bearings can be inserted and then the assembly should replace a damaged one. I don't know how much extra friction the sleeve bearings will add, relative to the 1416ZZ ball bearings -- current draw will tell that tale. If this approach works it might be used to repair damaged pulleys.

The only issue that I see, looking at the drawing for the pulleys is that the center bore is relatively loosely held at .95-.97. You may want to open the pulley bore slightly to prevent the need for matched sets of tubes and pulleys to get the desired light press fit.

[camtunkpa]

Andy the revision looks pretty good. Are there going to be longer bolts included in the upgrade or do you feel the current bolts will be long enough even with the bolt recess change?

[Andy Baker]

Originally Posted by camtunkpa

Andy the revision looks pretty good. Are there going to be longer bolts included in the upgrade or do you feel the current bolts will be long enough even with the bolt recess change?

Good question, Cliff.

We also discussed this issue. The distance of thread engagement was 0.485 with the current version. With the new version (Rev4), the engagement will be lessened by 0.125, resulting in 0.36. Since there is a taper on the end of the screw, the last 0.15" (approximately) can't be relied on with regard to screw thread engagement. So, we now have about 0.21" of engagement. Since it's a 10-24 screw (with 24 threads per inch), that means we now have 5 threads of engagement at each screw. With 6 screws all having this same engagement, we believe that this will be enough. However, we will make a test and report back if this is good or not.

Andy B.