Banebot 56mm gearbox - double D related

[Joe_Johnson]

There are threads here, here and here that are all discussing the same issue. Will the double D joint between the last carrier and the output shaft survive?

There are sub-questions:
Will the 12:1 1 CIM joint fail? If so what can be done to fix it?
Will the 16:1 1 CIM joint fail? If so what can be done to fix it?
Will the 12:1 2 CIM joint fail? If so what can be done to fix it?
Will the 16:1 2 CIM joint fail? If so what can be done to fix it?

We can debate who is to blame and how could this have happened all day long. It is not going to get us the answers to the question, nor is it going to get us to a solution.

Here is the state of things:

Data Point 1: Hardness test of shaft and carrier:Output shaft: RC 23, RC 23.3, RC 23.2
Carrier Plate with D: RA 45.9, RA 46.9, RA 45.8

Rockwell C to Tensile Yield:
RC 22 – 115 Ksi
RC 23 – 117 Ksi
RC 24 – 119 Ksi
RC 25 – 123 Ksi
RC 26 – 125 Ksi

Rockwell A to Tensile Yeild:
RA 45.0 – 62 Ksi
RA 45.5 – 63 Ksi
RA 46.0 – 64 Ksi
RA 46.5 – 65 Ksi
RA 47.0 – 66 Ksi
RA 47.5 – 67 Ksi
RA 48.0 – 68 Ksi
Data Point 2: Calculations & Date = Prediction
Using calculations + above hardness tests of the joint, I predicted that with the 12:1 gearbox should fail at about 3.1N-m input into the gearbox. Since 3.1N-m = 28in-lbs = roughly 1.5X CIM_stall_torque, alarm bells began ringing.

Data Point 3: static failure test
Using a test set up and a torque wrench, I was able to determine that the join failed with a single loading of between 3X and 5X CIM_stall_torque. This is better than predicted but not that encouraging considering that the “failure” was basically a spungy type event, allowing for actual failure to begin at lower than 3X. See picture.

Data Point 4: Cycling Failure test
Using the same test set up, I was able to show that the joint will fail at 2X CIM_stall_torque with repeated loading. See Picture.

Data Point 5: Failure in the field
This still did not mean that teams would see failures in the field because we don’t know the dyanmic loading that we expect to actually see during real world conditions. I typically use 2x as the dynamic load factor, but that is not a law of physics it is just a rule of thumb. So… …we needed confirmation.

Team 166 uploaded a picture that confirmed this problem will be seen by teams. This is after 3 hours of driving. Team 166 did not say what ratio & how many CIMs they used, but I suspect it was 16:1 with 1 CIM. Will someone from 166 confirm this please.

Where do we go from here?
Now, what do we have to do going forward? I believe that we can get a quick 2X in the output by simply getting harder carrier plates (RC 23 or higher). I believe that will almost certainly get us to the 12:1 1-CIM case as well as the 16:1 1-CIM case.

What it does to the 2 CIM cases, I can’t predict. My best educated guess is that for most teams the this will address the problem for the 2-CIM cases too, but I have to be honest, the picture from 166 has me worried. If they were in fact using the 16:1 1-CIM set up, then going to a 2X hardness will be exactly matched by a 2X in loading due to the 2-CIMs.

We need more testing and we need more thought on solutions.

More Information
For your information, Banebots is well aware of this. They are working hard with FIRST, their suppliers, and yours truly to get a good solution out there to teams. Everyone is sorry. Beating us all up will not help. Calling them will only take time away from fixing the problem. We are working on it.

Joe J.

[MrForbes]

In the pictures of the failed plates, it looks like the shaft is not fully engaged with the plate. In addition to hardening the plate, it might be worthwhile to look into some type of mod to the shaft to get full engagement.

[Jonathan_Norris]

May I add that this same problem is happening with the 42mm gearbox’s, which many teams planed to use with the FP motor to power an arm. It will fail on you quicker than the 56mm drive gearbox’s. If I was on or mentoring a team planing on using these gearbox’s I would tell them to grab a pair of single speed Andymark gearbox’s which for the same price accept 2 CIMs or 1 Bike CIM and 1 CIM.

[Nitroxextreme]

I have been following the banebot transmission failure threads for some time now and now I feel teams need to take action.

We are currently running 2 CIMs on each gearbox…so we definately need to get parts made. At the moment are transmissions are on the robot and not easily accessable, but we will get to them to replace the Double D plate. We have yet to have an issue…but are going to take precautions now.

Does anyone have a CAD drawing of the piece that needs to be made?

It is my hope that teams which have the resources…get many of these plates made to help out other teams who are not as lucky.

I would love to not have to worry about it and wait for BaneBots, but their response time (at least with me) was not acceptable. Teams need to start production now so the plates can be replaced BEFORE regionals.

[dapub]

Is this a problem only with a 2 motor setup? Should we be concerned if we are just using one CIM per gearbox?

[cbpetrovic]

Dr. Joe.

We are using 1 CIM w/ 12:1 ratio.

An ME here would like to know how you calculated the stress. Below is his observation from the photos as well as conversations with me.

“The initial interface is 2 lines of contact (essentially zero area) between the shaft and the D hole. It is only after material begins to deform (ever so slightly) that area contact is made. I am curious what he uses for stress area and how he determines it.”

C. B. Petrovic

[sanddrag]

Is the double D shaft pressed into that plate, or it just sits in there? Howabout the planet pins? This part that fails is just a 2D geometry plate? Is it easy to swap out? We have a lasercutting sponsor. There is a chance we may be able to set up a large production run of lasercut plates available for a small cost. Is laser accurate enough (draft angle)? What kind of material would it take to make this plate never break? How would it have to be treated/hardened? Joe, talk to me. I want this to be a good year for FIRST. A robot that drives without worry is the key to success.

[Joe_Johnson]

@dapub

Is this a problem only with a 2 motor setup? Should we be concerned if we are just using one CIM per gearbox?

From the data we have to this point, it looks like it could be a problem over time for even the single motor CIM drives if driven hard back and forth for many cycles. I think it will take significant cycles to fail but based on the experience of Team 166, I think it will fail.

The good news is that when it fails it doesn’t seem to take the rest of the gearbox down with it. It is not ideal but it is something.

Joe J.

[Joe_Johnson]

@cbpetrovic

Dr. Joe.

We are using 1 CIM w/ 12:1 ratio.

An ME here would like to know how you calculated the stress. Below is his observation from the photos as well as conversations with me.

“The initial interface is 2 lines of contact (essentially zero area) between the shaft and the D hole. It is only after material begins to deform (ever so slightly) that area contact is made. I am curious what he uses for stress area and how he determines it.”

C. B. Petrovic

I assumed a linear stress distribution normal to the joint surface starting at the center line and increasing to a peak at the corner of the D. I then integrated the torque to get the sum (on both sided of the D) which had to resist the torque on the joint. I found an engineering rule of thumb that allowed me to estimate failure of the joint by saying the material will start moving at 1.5X the tensile yeild stress. Once the material begins to move, the stress distribution changes but the joint is losing the battle. So I used the torque that made my linear stress distribution have a stress in the corner greater than 1.5X Stess_Tensile_Yeild

I hope that helps.

Joe J.

[flightofone]

Would a homebrew hardening of the plate be of use in helping this problem? Heating it to red-hot and tossing it into a bucket of cold water seems like a fun experiment. Into potentially explosive oil could be even more exciting (just kidding, this could/would be very dangerous). http://www.efunda.com/processes/heat_treat/hardening/direct.cfm

[Cascade]

Is there any possibility FIRST would let us use last year’s transmissions as a solution?

This is so bad I can’t even understand how this mistake could happen. We have been planning for a year and getting ready for this year’s competition and we may be sunk before we even take the field. Why were these things not tested before the switch was made?

boooo:(

[cbpetrovic]

This plate looks like it could be manufactured out of 4140 pre hard (Rc 28 - 32) fairly easily. 4140 Rc 28 - 32 should give Sy of around 120 - 130 ksi. A wire EDM could do all features with high precision. Comments?

Could you provide drawing for us to use?

thanks

Ron Reich 166

[Joe_Johnson]

I have been talking with a lot of folks. A plan is starting to take shape.

I have 2 teams that have agreed to do some additional testing for us (thanks to #1279 & #1618).

By the way, I have confirmed that FIRST had these gearboxes almost 9 months ago and they say they did extensive testing on real robots with both the 1-CIM and 2-CIM set up. It is still possible that the one confirmed field failure is a fluke of some kind and not representative of the loading we expect to see in the field.

We will work hard to understand the extent of the problem and to come up with a way to address the issues.

Joe J.

[sanddrag]

Is there any room to make the plate thicker? Or, what if we go to hex instead? Forgive me, I haven’t actually looked inside the thing yet.

[haroony341]

team 2234 at Episcopal Academy had originally planned to use Banebot’s dual CIM gearbox attached to the 56mm planetary gearbox but we no decided to scrap the idea, and go with AndyMark’s dual speed transmission. more than just the double-D connection, the larger sized gears in the planetary box were worn down almost instantly. we had only done a few tests with no load at all and the gearbox just locked up. we found that the one of the gear’s tooth was bent. we chipped it off as a temporary solution. but a few days later, after just one failed load test, it locked up instantly, and each of the gears had something wrong with it.

hopefully we’ll have better luck with the new transmissions

[RaMoore]

We have tried the homebrew hardening technique and this is looking promising as at least an improvement.

How we hardened/quenched the material:
Used two propane torches to get the steel to a red hot level (approx. 550 deg. C) and then submerged in oil (motor oil in this case). The oil rapidly cools the metal (better than water) and makes a bunch of smoke (burning oil?). The metal is submerged until cool enough to almost handle. If quenched in a small volume of oil, change oil to cool oil for next cycle. The material is then reheated to glowing red hot and then re-submerged. Process was repeated three times and then tested.

Testing procedure:
motor was clamped down and half of a torque coupling was placed on the end (two jaw, pictured). A c-clamp was clamped onto the table in between the two jaw teeth and gave approximately 90 deg. of travel before locking up. The motor was then powered 100 times back and forth into full stall. Motor shaft rotates 90 degrees and then is slammed into stall when the jaws lock up. (This really isn’t good on the motor nor the trans. but that’s what we want right?) The test was performed in 20 cycle intervals (since the leads got to hot to hold) until 100 cycles were achieved. Pending results show little distortion in the cut aways (pictured). Future tests are pending.

[Rich_Kressly]

@haroony341

team 2234 at Episcopal Academy had originally planned to use Banebot’s dual CIM gearbox attached to the 56mm planetary gearbox but we no decided to scrap the idea, and go with AndyMark’s dual speed transmission. more than just the double-D connection, the larger sized gears in the planetary box were worn down almost instantly. we had only done a few tests with no load at all and the gearbox just locked up. we found that the one of the gear’s tooth was bent. we chipped it off as a temporary solution. but a few days later, after just one failed load test, it locked up instantly, and each of the gears had something wrong with it.

hopefully we’ll have better luck with the new transmissions

I happened to be there to witness this one myself. I stopped in to visit 2234 to discuss shipping and a few other things with them, but instead I found a group of rookies huddled around a frozen gearbox after nearly no-load testing. Here you have a rookie that’s done its homework, has an experience FIRSTer working with them and spent more than three days trying to get something to work that was listed as a “proven solution” in the tips and best practices guide. I saw the bent planet teeth myself that were hitting the mounting hole housings between stages and I also saw a central gear move laterally about 1/8" during a bench test. Mind you when the team first got these items they spent time filing burrs off of gears just to get it all to run the first time. I’m concerned that some of this product may also be pushing the limit of acceptable tolerances.

I’m not trying to bash anyone, and I’m very glad Dr. Joe is working so hard on this. I’m sure his work will prove to be a huge benefit down the road. But, standing there with this crew, what they didn’t have was time and neither an experienced mentor from another veteran team or myself could guarantee that it would run well even if they supported the shaft better, and had plates to sure up both sides of the box (which another team would have had to make for them), and they found a way to offer more motor support.

I know everyone involved is working hard on a solution and I understand that new vendors are important for a lot of reasons. However, as a rookie team mentor last year all I can say is that I’m so glad that the kit gearbox we had in 2006 was so robust. I’m equally thrilled right now with my team’s decision to go with AM single speeds.

Right now, headed into week 4, teams need to know they’ll run reliably and right now, in my way of thinking, that means using another product if a team can afford it.

However, I eagerly await Joe’s final word and banebots’ ultimate fix for these issues. Yet, the fruits of this labor will probably need to wait until next year for many teams. The clock is ticking and unfortunately it looks like teams will have to dig into their pockets this year for something they thought was already in the kit.

I’m trying to be kind here, but I’m concerned for many teams.

[Mike_Schroeder]

@Rich Kressly

Here you have a rookie that’s done its homework, has an experience FIRSTer working with them and spent more than three days trying to get something to work that was listed as a “proven solution” in the tips and best practices guide. I saw the bent planet teeth myself that were hitting the mounting hole housings between stages and I also saw a central gear move laterally about 1/8" during a bench test. Mind you when the team first got these itemd they spent time filing burrs off of gears just to get it all to run the first time. I’m concerned that some of this product may also be pushing the limit of acceptable tolerances.

I’m not trying to bash anyone, and I’m very glad Dr. Joe is working so hard on this. I’m sure his work will prove to be a huge benefit down the road. But, standing there with this crew, what they didn’t have was time and neither an experienced mentor from another veteran team or myself could guarantee that it would run well even if they supported the shaft better, and had plates to sure up both sides of the box (which another team would have had to make for them), and they found a way to offer more motor support.

I know everyone involved is working hard on a solution and I understand that new vendors are important for a lot of reasons. However, as a rookie team mentor last year all I can say is that I’m so glad that the kit gearbox we had in 2006 was so robust. I’m equally thrilled right now with the team’s decision to go with AM single speeds.

Kres, I agree, I don’t want to bash anyone, but i do know that something went wrong here. coming from a team with little to no engineering support at least mechanically i have got to say that we aren’t taking the chances of waiting for a fix, and are already getting the money together for AM single speed, and i would like to thank bane bots for their donations, and also hope that this doesn’t end their fledgling entry into FIRST

–Big Mike

[MrForbes]

@squirrel

In the pictures of the failed plates, it looks like the shaft is not fully engaged with the plate. In addition to hardening the plate, it might be worthwhile to look into some type of mod to the shaft to get full engagement.

I brought home one of the teams unused 56mm transmissions, and my sons and I did a kitchen table dissection. We discovered that the shaft does fully engage the plate under some conditions, but we also discovered that there is about 1/32" end play in the planetary geartrain stack. So, the end play seems to be a bit excessive, and allows the shaft to disengage about one quarter of the way from the plate.

Instead of a shaft modification, perhaps a select fit washer to control end play might help.

[haroony341]

we put a washer behind the gear that connects the 2 CIMs together. it did help a little bit but the 4:1 gears were constantly grinding against the dual CIM adapter where the screws hold the 2 together. and adding enough washers to keep from grinding would have pushed the gear to far forward it wouldnt be connected to the CIM gears, or the would barely be connected and cause just as many problems.