Unusual NEO motor failure and teardown

We had a new NEO motor fail with only a few minutes of heavy driving after being run on a robot. One of the windings completely burned up, but it’s not clear what started it. Taking the motor apart did reveal some possible manufacturing issues that could contribute to failures.

Photo album with analysis here: https://imgur.com/a/rNgr2qo

5 Likes

Hey Jack,
We need a bit more information from you to fully figure out why this specific motor failed. It is possible that this one had a manufacturing defect, but we want to make sure we understand the condition that it failed in.

Was your team running Smart Current limiting or any closed loop control on your motors? What is the gearbox configuration that these were in (number of motors in each, gear ratio, and approximate weight of robot).

I also wanted to address a couple of things from your photo gallery to help you (and others) what you are actually seeing during your tear down.

  • The case is aluminum diecast - the coating is a black plating process (not a plastic coating)
  • Since this is an out-runner inside a case (for safety) The heat dissipation goes through the front face of the motor where the coils are coupled. The outer case will also carry some of this heat from the front plate. In out-runner motors this is how heat transfer normally works since all the heat comes from the coils and not the magnets.
  • Sizes - this is a custom motor, and we went with sizes that allowed for the performance we were hoping to achieve. The thickness and diameter of the lamination stack each have different impacts on the performance of the motor. (so yes, they are non-standard sizes)
  • The number of wraps has a huge impact on motor (and any inductor) performance. Using a bundle of smaller wires allows for significantly higher wraps in the same space than a single coil wire. This is very common with Bldc motors
  • Our lamination on the pack are insulated through their lamination process, we didn’t add the paper insulator as it was not required. In our experience the failure mode is coil wire to coil wire not coil wire to lamination, so we choose not to add additional cost
  • The soldering of the white wire does seem to be a cold solder-joint and of all the things you point out I would say if the failure is related to something manufacturing, this is likely the case.
  • The epoxy that is used to hold the wire coils together is applied and then put in a vacuum chamber which pulls it through the entire coil pack. There will always be some build up based on the direction the coils sit in the vacuum but there is a coating throughout the coils. There is a big difference in holding the wires together vs cutting them apart. Additionally as the windings are continuous, epoxy on a single side provides support to the entire coil.

Also just for fun…here is a video of the winding process for NEO.

20 Likes

Don’t try that at home! Nice to see the manufacturing process.

I just toured a BLDC factory two weeks ago, and that is a really nice looking winding machine you guys have.

Is the epoxy curing vac chamber also heated? Heating the chamber might decrease the cure time and reduce the “settling” effect due to gravity. You could also rotate the stators while they cure, but that’s a good amount of capital to get the fixturing.

We have a ~120lb robot with a 13lb battery and 2 NEOs per gearbox with an 8.71:1 low gear and a 18.81:1 high gear and 6" wheels, for a free speed of 17.06ft/s in high gear and 7.89ft/s in low gear. We had been driving it around a field at near top speed for 5-10 minutes when it started to smell. All of the other motors were warm, as expected.

We were not using closed loop or any smart features in the firmware and we used the public void set​(double speed) method to set the output of the motor controllers. The Sparks were running firmware 1.0.65.

When looking at the current consumption of the this motor and the other motor in its gearbox as recorded by the PDP, they were both drawing similar average amounts of current, but there was a unusually large difference in the current draw between the motors at some times.

One thing that concerns me now is how easily that blue wire broke off of the hall sensor board. Looking back at the picture, the blue, black, and green wires, and possibly others, were just soldered to the top of the board instead of inserted through, so I’m wondering if this solder joint had already failed before I opened it. I didn’t put much stress on the wires when removing the board (I pried up the side across from the wires with a screwdriver) and none of the other solder joints broke.

To clarify my comment about the case coating, there appears to be a transparent lining or a thick layer of clearcoat inside of the outer case.

Another thing I noticed, it’s only really visible in the left side of this photo:


but one or two pieces of wire seemed to have separated from the top side of the stator on the burnt phase. They did not appear damaged and it was not clear if they came off in the process of removing the rotor or if they were like that beforehand.