Banebot 56mm gearbox - double D related

[Andy Baker]

I had a chance to read up on this issue last night and also talk to Joe this morning. I'm coming in late to this party, but am offering help or advice. Here are a couple of things I see...

1. Just how bad is this problem? What is the % of failure rate here?

2. What is the best fix? It's great to see that is what is focused on in this thread. Best means quickest and easiest for teams, not just the best exact design (for instance, the crossed-dowel fix is a nice design, but difficult to implement since the shaft is hard).

3. Specs for a fix are needed quickly. I'm very interested to see what would CB Petrovic comes up with the 4140 material solution. If I had to pick the "best" fix, that may be it... to re-create a bunch of plates out of 4140.

which brings us to this...

4. IF someone can get a fix (prints and a CAD file) done quickly, there could be ways to mass-produce new parts to get out to teams. I would suggest that a well-toleranced print be made, along with an acurrate CAD file (.stp version). Then, someone could go to http://www.mfg.com and post this as an RFQ. They could ask for quotes to be returned within 1-2 days, and finished parts to be required within one week. We at AndyMark have used www.mfg.com as a fabrication resource of multiple parts and have had good experiences 90% of the time. The bid could be awarded to multiple suppliers, each providing 1-2,000 parts. My guess would be that this part would cost anywhere between $8-$14 if someone was making 1,000 in this short lead time situation.

Once parts were made, boxes of them could be sent out to regions, and then teams could drive a couple of hours to get their part from a central location.

Andy B.

[MrForbes]

That sounds like a plan.....

Some things to consider: The 12:1 and 16:1 output planet carriers are different, because the planet gears are a different size, so the pins are in a different location. Would the replacements be available for both the KOP 12:1 and the upgrade 16:1 transmissions? or would teams have to settle for using 12:1 only?

If you're not familiar with the transmission design, it is a two stage planetary, with a 4:1 first stage, and a 3:1 second stage. An upgrade 4:1 second stage is available from Banebots, and apparently some teams have decided to use it, and apparently it has the same problem with planet carrier plate failure under some conditions. The stages are mostly interchangeable, except that the pinion for the CIM motor is only made for the 4:1 stage (as far as I know), and also I believe it is the only one that will fit in the motor mounting end of the transmission.

Should the replacement plate be made significantly softer than the shaft, to act as a sacrificial part?

Would it include the planet gear pins, or would the teams have to press them out of the old plate, and into the new plate?

Is anyone going to consider the end play issue now, or wait until it causes problems in regionals?

[Rosiebotboss]

Thanks Andy for taking the high road in your posting.

For the "newbies" that are keeping up with this thread: this is what FIRST is about, a perfect example of gracious professionalism. Someone recognized a problem and posted it here. Instead of finger pointing (yes, there was some in the beginning but that quickly subsided), the dialog quickly turned to "Ok, what now? Let's come up with a solution." That posting mobilized an army of engineers, students, teachers and teams working together for the common good.

Back to Andy.....Andy, thank you for taking the high road here. You very easily could have said, "Hey, teams!! AndyMark has an alternative design here. We can provide......." You can guess what the rest of the sales pitch would be. But, no. Andy has offered a solution instead of a sales pitch. THAT is GP in action!!

[Jadium]

Hi,

I know this will not help all of the teams, but since we were already machining our own shafts for a direct drive application we decided to hex the end of the shaft and the carrier plates. The parts will be made by Thursday and will hopefully be tested out Thursday night for a couple hours. We will be making the plates out of 4140.

I don't have access to modeling software until Thursday, so the only drawing I can post is a quick sketch, I will also post the IGES files we sent out for machining. The files will only be useful for the carrier pin hole dimensions as we have a hex in the plate instead of the double d. Also please note that the dimensions in this file were calculated using the SINE Law by measuring the pin diameter and the outer dimension between two pins of the pentagon, so if anyone can double check the dimensions...

By Thursday I will be able to post some results of the the strength of the modified shaft and carrier plate and ensure that the parts as dimensioned on this sketch and in the IGES files mate into the existing transmission.

[Joe Johnson]

Originally Posted by Jadium

Hi,

I know this will not help all of the teams, but since we were already machining our own shafts for a direct drive application we decided to hex the end of the shaft and the carrier plates. The parts will be made by Thursday and will hopefully be tested out Thursday night for a couple hours. We will be making the plates out of 4140.

I don't have access to modeling software until Thursday, so the only drawing I can post is a quick sketch, I will also post the IGES files we sent out for machining. The files will only be useful for the carrier pin hole dimensions as we have a hex in the plate instead of the double d. Also please note that the dimensions in this file were calculated using the SINE Law by measuring the pin diameter and the outer dimension between two pins of the pentagon, so if anyone can double check the dimensions...

By Thursday I will be able to post some results of the the strength of the modified shaft and carrier plate and ensure that the parts as dimensioned on this sketch and in the IGES files mate into the existing transmission.

I have had good success with STEP files. IGES is fine but circles are often not circles, arcs are not arcs, etc. It is a pain sometimes -- especially if you have to modify the part or if you have to measure things.

For what it is worth.

Joe J.

[Joe Johnson]

Originally Posted by Andy Baker

snip
4. IF someone can get a fix (prints and a CAD file) done quickly, there could be ways to mass-produce new parts to get out to teams. I would suggest that a well-toleranced print be made, along with an acurrate CAD file (.stp version).
snip
Andy B.

Andy,
I will have a STEP file and associated print with tolerances made by one of the best engineers I know (he's a Delphi guy that I could never quite get involved in FIRST... ...maybe some day).

I am also getting the shaft drawn up in CAD too.

Here is my thought process. I am looking for options here. We don't really have a bound on the problem yet. I believe it is likely that an RC 23 carrier plate will be the solution for teams with 1-CIM. But there are teams that have counted on using the 2-CIM with 16:1 ratio. Also, hoping for a good test outcome is not a plan -- it is just a hope. So, I am looking for options in case the tests show that an RC23 carrier is not enough. In this case, there is no more we can do with the carrier alone because we will just push the failure to the shaft. So... ...I am starting to think about the next step in case we have to take that step.


2 questions for folks with the right knowledge base:
#1 should we make the joint square to increase the surface area (and thereby lower the stress)? I have found an 11/32" square broach (8.73mm). Going from a 9mm D to a 8.73mm square will lower the stress to 46% of the current value (an increase in the failure torque of 2.15 (50% of the reduction is from 4 sides taking torque rather than 2 and the extra 4% reduction is that by going to the smaller distance between the flats, you get a larger load surface).

#2 I am thinking that we should have a target hardness of RC 40. This is a Tensile yield of 180Ksi (1250 Mpa). Is this too brittle? I am thinking it is probably ok. My reasoning is that I know that Forkbolts for car door latches are hardened to RC30-38. If RC38 is good enough to take the impact loading from a car crash, I think that RC38 is not too far. It is a small step from 38 to 40. Please share your educated gut feelings with us (provided you have an educated gut).

Joe J

[Tazlikesrobots]

It is great to know that there great minds working on the solution. We are a team of very limited resources (financially and know how) and we rely heavily on the items provided in the KOP. The thought of a filed gearbox will most certainly means a pedal powered robot for us, but on board human players are now allowed Hopefully a solution will be found soon!

On behalf of our team, I want to say thanks in advance to everyone working on this problem! It is posts like this that give teams a "heads up", practical advice on a workaround or an alternate solution. This is just one of the reasons the FIRST community is in a class all its own.

[Springman]

Originally Posted by Joe Johnson

Andy,
I will have a STEP file and associated print with tolerances made by one of the best engineers I know (he's a Delphi guy that I could never quite get involved in FIRST... ...maybe some day).

I am also getting the shaft drawn up in CAD too.

Here is my thought process. I am looking for options here. We don't really have a bound on the problem yet. I believe it is likely that an RC 23 carrier plate will be the solution for teams with 1-CIM. But there are teams that have counted on using the 2-CIM with 16:1 ratio. Also, hoping for a good test outcome is not a plan -- it is just a hope. So, I am looking for options in case the tests show that an RC23 carrier is not enough. In this case, there is no more we can do with the carrier alone because we will just push the failure to the shaft. So... ...I am starting to think about the next step in case we have to take that step.


2 questions for folks with the right knowledge base:
#1 should we make the joint square to increase the surface area (and thereby lower the stress)? I have found an 11/32" square broach (8.73mm). Going from a 9mm D to a 8.73mm square will lower the stress to 46% of the current value (an increase in the failure torque of 2.15 (50% of the reduction is from 4 sides taking torque rather than 2 and the extra 4% reduction is that by going to the smaller distance between the flats, you get a larger load surface).

#2 I am thinking that we should have a target hardness of RC 40. This is a Tensile yield of 180Ksi (1250 Mpa). Is this too brittle? I am thinking it is probably ok. My reasoning is that I know that Forkbolts for car door latches are hardened to RC30-38. If RC38 is good enough to take the impact loading from a car crash, I think that RC38 is not too far. It is a small step from 38 to 40. Please share your educated gut feelings with us (provided you have an educated gut).

Joe J

In reference to the hardening question, I have spoken to our local heat treating facility and they recommend a hardness of RC40. The implication of this is that you should probably start by using standard 4140 (vs. prehard @ approx. RC23) and have the parts heat treated professionally after machining. Speaking from my own experience, 4140 will become too brittle (break vs. deform) for this application at RC>45.

[eugenebrooks]

Originally Posted by Joe Johnson

#2 I am thinking that we should have a target hardness of RC 40. This is a Tensile yield of 180Ksi (1250 Mpa). Is this too brittle? I am thinking it is probably ok. My reasoning is that I know that Forkbolts for car door latches are hardened to RC30-38. If RC38 is good enough to take the impact loading from a car crash, I think that RC38 is not too far. It is a small step from 38 to 40. Please share your educated gut feelings with us (provided you have an educated gut).

Joe J

Joe,

A target harness approaching RC 40 should be fine. We have run
welded axles, 4130 flanges welded with 4130 rod onto 4130 axles
for several seasons. We harden these and then temper at 800F.
Tempered at 900F, a little softer, the RC is 36 and the tensile yield
strength is 161,000 psi. We temper at 800, slightly off the charts, so
to speak, but safely above the brittle zone, to get just a little more
strength. We have bent these axles in competition with direct robot
impact on a cantelevered wheel setup, but have never broken one.

Hardening the carrier plates to a value approaching RC 40 is probably
just right. Using 4140, or even better 4340, is best for this application.
We are using 4130 because this is the material we have on hand.

Here is a link to useful heat treatment info:
http://www.aerospacemetals.com/steelalloys.html


Eugene

[Joe Johnson]

Originally Posted by eugenebrooks

Joe,

A target harness approaching RC 40 should be fine. We have run
welded axles, 4130 flanges welded with 4130 rod onto 4130 axles
for several seasons. We harden these and then temper at 800F.
Tempered at 900F, a little softer, the RC is 36 and the tensile yield
strength is 161,000 psi. We temper at 800, slightly off the charts, so
to speak, but safely above the brittle zone, to get just a little more
strength. We have bent these axles in competition with direct robot
impact on a cantelevered wheel setup, but have never broken one.

Hardening the carrier plates to a value approaching RC 40 is probably
just right. Using 4140, or even better 4340, is best for this application.
We are using 4130 because this is the material we have on hand.

Here is a link to useful heat treatment info:
http://www.aerospacemetals.com/steelalloys.html


Eugene

Great data. Thanks.

I have just spoken with a materials guy at Delphi. He has recommended 1040, hardened (he advises oil quench since the parts are relatively thin) then tempered to RC 40-42.

This is from a source I will call "usually reliable" but I am really out of my comfort zone on this one.

Everyone with expertise in this area please feel free to comment.

Joe J.

[capnrmorgan]

We are 2nd year team that is planning on using the 56 Trans in direct drive. I am new to posting as well. One possible solution we came up with is to reduce end play by adding a shim between the motor and the drive gear. This pushes the whole mechanism forward so ultimately the drive shaft is set "into" the Double D. We don't have a drive train to test yet. If other teams could make comment or possibly test this solution we be most grateful. We feel the end play is allowing the failure to occur. Also my technical adviser suggests crucible packed with charcoal, bone, or other carbonized material and forge heated for 8 hrs or more. We are concerned about warping the plate. We are opting to not heat treat and have andymark trannies just in case we have to redo the drive train.

[MrForbes]

Originally Posted by capnrmorgan

Also my technical adviser suggests crucible packed with charcoal, bone, or other carbonized material and forge heated for 8 hrs or more.

I always had a feeling metallurgy was one of the dark arts.....

As for the shim idea, it sounds reasonable to me at first glance, but I have not had a chance to take a close look at the fit of the motor and gear. I think it would be relatively easy to figure out the shim thickness required, although you'd have to pull the gear off the motor shaft to install it, or change it if it's wrong. Also putting the shim there would be best from the viewpoint of having full gear tooth engagement. Adding the shim in the second stage planetary would allow partial sun gear/planet gear contact.

[ZZII 527]

An update on the heat-treating front: Yesterday, I tempered the 42mm plate, which seems to be some type of tool steel, and managed to get a significant increase in hardness, although my first attempt was too brittle. I will try again when I get another sample, but in the mean time I tested the 56mm plates too. Unfortunately, they are definitely NOT tool steel and will not harden much without adding carbon. Some data:

56mm carrier before heat treatment: HRA 46.1, 46.5, 46.5
56mm carrier, 950C for 15 minutes, oil quenched: HRA 49.3, 49.9, 49.4
56mm carrier, 950C for 15 minutes, water quenched: HRA 57.1, 57.9, 58.5 (about a C16)

It's a slight improvement, but not the C40 we are looking for or even the C23 of the shaft. Case hardening is still an option, but not an easy one. I suspect the solution will be an entirely new plate made of something harder and with a tighter fit for the shaft. I will probably shift my focus to this avenue now, although hardening is still an easy solution for the tool-steel 42mm carrier.

Good luck all!

[MrForbes]

Originally Posted by ZZII 527

I suspect the solution will be an entirely new plate made of something harder and with a tighter fit for the shaft.

I was pondering making the shaft to plate joint a press fit. I think it would be relatively easy to get the size for round part of the hole figured out, but I wonder what the tolerance is on the flats on the shaft. If they vary by more than a thousandth of an inch, it would be challenging to get a reliable fit.

It would be really nice from the viewpoint of the wear we are seeing....I know in automotive transmissions, once splines start to get loose, they wear out quickly.

[Joe Johnson]

Originally Posted by squirrel

I was pondering making the shaft to plate joint a press fit. I think it would be relatively easy to get the size for round part of the hole figured out, but I wonder what the tolerance is on the flats on the shaft. If they vary by more than a thousandth of an inch, it would be challenging to get a reliable fit.

It would be really nice from the viewpoint of the wear we are seeing....I know in automotive transmissions, once splines start to get loose, they wear out quickly.

I think that we have pretty much put to bed the idea that the carriers can be easily hardened (by teams or even by a "swap service" of some kind).

I think the urgency of getting the mat'l analysis is lessened.

As to the tight fit, this is going to help some but my theoretical analysis/calculations do not show a significant improvent in this case (nothing like what I expect we need).

Here is my current thinking of most likely patches:
#1 harder carriers (= to the shaft hardness) address all the 1-CIM cases (16:1 & 12:1)
#2 harder (perhaps RC 40-45) carriers plus redesigned joint (e.g. square hole) plus harder (again perhaps RC 40-45) shafts (with mating joint) address all the 2-CIM cases (IF, and this is a big IF, tests show 2-CIMs stress the joint beyond what is done in #1 -- note that motor torque is not the only thing that determines the torque this joint sees, friction is another and it may limit the torque that the joint sees to something close to the 1-CIM number)

Stay tuned.

Joe J.