Caution EVO Shifter Users: Dog Gear Failure

During our last match at the GoS Week 0 event in Pittsburgh two days ago, the dog gear of our EVO shifter cracked into pieces. A team member posted pictures to FRC Reddit, and I want to provide some more details for the benefit of teams using EVO shifters. Here’s a picture of the failed dog gear.

This is just our ‘quick look’ analysis. It’s possible we’re wrong and something fluke-y happened that won’t be an issue for other teams. Also, we’re not trying to bash Andymark. Though disappointed, we have no ill will toward them. It’s a new product line, and we understood the risks of early adoption. On the customer service side, they have been responsive and helpful.

WHAT HAPPENED
The robot was crossing the field in high gear, and our driver anticipated a collision with an opposing team at moderate speed. He commanded the shift to low gear a split second before the collision. He knew something had failed in the drive train shortly after. Our regulated pressure was ~50 psi, though it’s possible this shift was lower (we used an off-board compressor setup). We don’t believe our pressure was any lower than 30 psi. Our low gear is ‘Ratio A’ (default), so even at low pressure it was shifting into the spring assisted direction. After the match we located the majority of the cracked dog gear.

WHY IT FAILED
There were multiple crack locations, two of which we consider the possible initiation point of the failure. Regardless of which occurred first, they failed in rapid succession…

  1. The inside corner of the the tooth that contains the counter bore for the locking screw (picture). The wall thickness in that location is ~3x thinner than Andymark’s previous dog gears (picture). There is now a shoulder screw in this bore, rather than a cap screw, and the head might be intended to pick up stress. The nominal diametrical gap between the bolt head and the bore wall is 0.005” (from the CAD model). This has potentially high variance since standard COTS shoulder screws don’t control the head diameter as tight as the shoulder diameter. In any case, even nominally this is too much gap to be expected to pick up any substantial load before the bolt’s shoulder and threads do. That load is transmitted through the thin wall section of the dog tooth, leading to high stress and failure.

  2. The shoulder screw where it steps to the 4-40 threads. This is where the bolt sheared. Previous generations of AM shifters have had bolt shearing problems, but the new shoulder screw was meant to address that (per Andymark). Unfortunately, with the threads still #4-40, if they pick up stress before the shoulder, the failure is the same. This is possible if the bore on the dog interfacing with the bolt shoulder is oversized, or a failure starts elsewhere, deforming the bore (See #1).
    **
    WHAT OUR TEAM IS DOING**
    This failure required a deep disassembly of our robot (we’re using the kit frame, so the output shaft is also our center wheel shaft). That on top of the previous EVO shaft concentricity issue recall, which also required a deep disassembly to replace the gearbox. This has made us nervous about using the EVOs in competition should the dog gear or some other yet unknown issue pop up. After weighing the pros and cons, our students chose to return to the kit single speed gearboxes. But, I wouldn’t expect any teams to do the same based on one failure of one dog gear on one team. So…

WHAT WE RECOMMEND TEAMS USING THE EVOS DO
Basically, follow all the warnings and cautions Andymark had for the previous generations of shifters, which were supposed to be fixed with the EVOs. The real bummer is that on previous generations, when the locking pin failed, it was a disappointment but not catastrophic, as the $2 bolt could be replaced between matches with no disassembly (as long as the threads weren’t deformed in the dog). Now, a failure causes not only the shearing of the bolt, but the cracking of the $22 dog gear which can’t be replaced without disassembling the gearbox, and that you very likely might not have a spare on hand for. A major problem if it happens in the thick of competition.

Reduce your shifting pressure to lower stress on the shoulder bolt. This increases shift times but you really don’t want a broken dog gear, especially if using the kit frame.

Consider replacing the shoulder bolt every few matches. The dimensions are the same as McMaster 92012A505. Per Andymark via e-mail … “The shoulder screw is the same physical dimensions as the 92012A505 part number, but is a custom screw made of a carbon steel and not the stainless steel that the McMaster screw is made of.” Although 92012A505 is actually alloy steel, not stainless, I think the point was they used a different material. So it’s up to individual teams whether you believe it’s more important to have the specific carbon steel Andymark designed with or use replacements to limit fatigue cycles, since no AM part number for just the screw exists.

Avoid shifting “on the fly”. Especially while engaged in high torque (pushing) situations. Shifting under torque increases the shift transient time and puts that thin walled corner under a higher stress for a longer time. Doing after-the-fact research we found a few people on various forums questioning dog shifting “on the fly”, but Andymark never recommended against doing so.

LESSONS LEARNED FOR THE FUTURE

Don’t use shifters as ‘drop in’ replacements to the kit toughboxes if using a kit frame. We had grown so accustomed to ‘assemble it and forget it’ with the kit gearboxes that making gearbox replaceability a priority wasn’t considered. EVO shifters, as they are now, still have too many reliability issues and replacing gearboxes face mounted to the kit frame requires too much disassembly for them to be compatible, in our opinion.

Do a ‘high torque test’ early. We’re thankful we got one at the Week 0. Pushing is limited Thursdays of regionals, too. It’s a nightmare scenario for this problem to be found Friday morning, when defense actually starts to happen.

Probably just use ball shifters. The ‘drop in ability’ of the EVOs to the kit frame and the ability to easily reconfigure gear ratios in the future sold us on them. But, in hindsight, we should have purchased ball shifters from Vexpro or WCP and made the necessary adaptations to the kit frame, or use a custom frame.

REPLY FROM ANDYMARK
“In response to your team’s EVO Shifter failure, we are currently reevaluating the design of the dog and shoulder screw. We did test the ‘shift down to low gear at high speed, on the fly’ scenario through the summer on a full weight robot using an AM14U2 with 6" wheels, however we did not test shifting down immediately before a robot impact. We did use this test mule for moderate speed impact testing for a different project over the summer, and found we were actually shearing aluminum output shafts in that test instead of cracking the dog. Additionally, our test mule had a different ratio-set than the EVO Shifters we currently sell. Finally, our test mule had 4 tooth dogs with a slightly shorter counter bore and a longer shoulder screw.“

EDIT: Here’s Andymark’s official fix.

This is the dog, not the dog gear. Came here wondering how you possibly sheared a dog gear, turns out you didn’t.

I’m not surprised they failed in this way. IMO all dogs should be secured with roll pins. The convenience of the screw is not worth the lowered cross section of the dog around the head or the weakness of the bolt at the threads.

If you shear the screw the dog is toast as you can’t get the broken off part out of the threads. At least with a sheared roll pin you can reuse the dog.

Plus as you’re seeing your dog is now much weaker due to the reduced cross section at the head, particularly with how big that counterbore is.

Andymark calls them dog gears. I’m discussing an Andymark product, so I did too.

To be clear, when did this actually happen? I know we were the robot that collided with you and directly after that, you pushed us straight across the field. We are using EVOs too in the exact same ratio as you, also with 6" wheels. You seemed to have no problem pushing us, and I was pushing back, but due to not having a fresh battery, our robot just browned out the entire time. If this had happened as soon as we collided, you would not have been able to push us. Correct? Also, where on the field did you find the pieces? We collided by the red loading zone.

Btw. You scared the crap out of me at Pittsburgh when you came up to us and said, and I quote: “Are you using EVO’s? Because we found these on the field.” as you handed me the pieces of the dog.
I thought they were from our robot and I had several mini heart attacks!

Thanks for this detailed account! I will be careful in the future so this hopefully does not happen to us…

It was after that pushing, later when it was kind of a run around the field. The broken pieces were right under the robot (blue side I think), as it couldn’t do anything but spin with half the drive train knocked out. I would rate that initial collision as more than moderate, and I let the drive team know it was way too aggressive for a scrimmage. We joke the dog gear break was karma. But, good point- the initial collision could have been the start of the failure, and the few more shifts during the match could have completed the break.

As to the scare- I’m really sorry! I should have chosen my intro more carefully, but I was in the middle of my own less-than-mini heart attack…