We will be using a Snow blower with hex output. We need to get it’s position however a pulley will be on the shaft. Therefore, we do not have a lot of space on the shaft.
We have mag srx encoders and first choice ones but we are not sure how to mount it to the shaft of the motor, as the shaft is not resting on any bearing or anything. Encoders, pots, and current/pose estimation were the solutions we thought of respectively.
About pose Estimation through current, we are concerned about if it would be precise enough. Have any of you tried it? If so, do you have any tips for it?
How would your current pose estimation work? If it would be more current = more distance than the problem is lets say you have an issue and your drive has more friction, then it’s going to read more current so your code will think it’s gone further, when in reality it’s the opposite. If it’s less current = more distance then, well, that won’t work.
We had a similar situation with a Bosch seat motor and AndyMark’s hex output. We were mounting it through a 2x1, and ended up with only a small amount of shaft sticking out.
We were able to mount a Rev through-bore encoder on standoffs, interfacing about 1/4 of the way into the hex bore. We added in the 1/4" round adapter they include in the bag to fill the rest of the bore, and bolted into the shaft with a 1/4-20. The bolt was definitely a press fit into the adapter (it basically threaded its way in), which is intended.
So far, it’s working great! Using it for position feedback over ~180 degrees of travel.
To be completely honest… I think my entire team completely forgot about the integrated encoder when we were putting it together - I didn’t even think of it until we put the wiring harness in later. It was a fun little project, though!
Given an applied voltage and current, along with parameters to describe the motor, you can build a model to guess the output shaft speed. Integrate this to get rotational position.
However.
I struggle to imagine that a 20ms sample rate of motor-controller measured current and voltage will be produce even a mildly useful guess. Multiple orders of magnitude worse than even the worst encoder. Then again, this is gut feel, with exactly zero data to back it up.
We can’t get our hands on those beautiful rev encoders for this season sadly because of the reasons stated above However, do you have any alternative solutions to use mag srx encoder to mount it to end of an 1/2" hex shaft?
That is the major problem we thought of. However we will be using it to turn a pulley so,we thought the friction difference might not be that crucial. Maybe we can learn about other people’s experiences with it if anybody ever tried it
You’d first need something to retain what is on the shaft. If you need the end of the shaft for that, it influences what you can do. Probably best case is if you have a enough room and a shaft collar to hold it on. That would free up the shaft end.
You next need a way to mount the magnet to the shaft so that it is centered on the shaft. The magnet for mag encoders is the long skinny one, and there is two possibility, either an external mount (which is longer) or drilling the shaft to fit and pressing in (using adhesive to hold in place). Neither is ideal, but either could work. You might be able to print an adapter from the hex to a magnet holder, for example.
Last you need a solid mount that places the mag encoder within range of the magnet. Depending on surrounding supports this could be easy or tough. You might be able to capture the mounting holes to build a support up from the motor, but that is some distance. We had to build mounts like this last year and I hated how much trouble it seemed to locate the right spot in space.
I’d say finally, there are some printable cases for through hex encoder using the mag encoder. Truthfully that maybe the easiest if you have that ability, and then you just have to keep the body from rotating with the motor (without concerning about sensor spacing).
The seat motor hall effect is maybe more trouble than it is worth. The output has to be conditioned to be detected on the RoboRIO, for one. Also, you only have a single channel, so you have to develop some logic to know which way the motor is turning (we queried the input voltage sign to figure that out).
You might be able to live without knowing which way the motor is turning in certain applications, but still those are the main issues with the seat motor from the sensor standpoint. You also have the issue stripping out the output, which we did twice last year, but that is also can be done differently than we did.
That won’t work when you’re in the reversal process, or if there is an external load. You could be applying reverse voltage but rotating forward, and v/v.
However, do you have any alternative solutions to use mag srx encoder to mount it to end of an 1/2" hex shaft?