PDA

View Full Version : dynamixel issue demagnetize/magnetize?



cod65
12-14-2018, 07:14 PM
Weird thing Ive noticed with one of my MX64. I noticed rather stair-steppy jagged performance on one of two similar motors.
If I move it with my hand it feels like a charge/torque is holding it in place until it starts moving. But this happens when it is completely disconnected from anything. It feels like when the motor is still it is energized to some degree, but when I keep moving the motor it wont 'stick' and moves freely. It's as if a charge is trapped inside and building when still. The only thing I can think of is that a recent surge from a loose wire, which fried another motor, may have somehow magnetized the motor coil? Is that even possible ? Any ideas what is going on ? I'm gonna try at Robotis as well, but thought there might be insight here. I'm gonna re-flash the firmware for starters, but it honestly feels like an electromechanical resistance.

jwatte
12-14-2018, 07:49 PM
This happened to me too. The windings in the servo overheated, and the motor now has a bit of stickyness because of it. The only fix is to send it to Robotis for motor replacement. The overall cost of that operation was less than half the cost of the servo, so that wasn't as bad as buying a new one ...
Note that you may be tempted to keep running the servo with the stickiness. Unfortunately, this means that the motor will use more current when holding still if it's in one of the damaged positions, which will build up more heat, and may lead to total failure later. Of course, the fix for that is STILL to send the motor to Robotis for replacement, so maybe you won't be risking so much in doing that -- assuming that what your motor holds, is not precious.

cod65
12-14-2018, 07:53 PM
gah. that's two down. These things are way too sensitive, would never make it to Mars :(

jwatte
12-14-2018, 11:21 PM
They don't like being stalled for long times. Or, if you use 4S, for any perceptible time ... 3S is safer.
Also, they don't like reverse polarity power. Just sayin' :-)

Finally, you can shear the output spline if you hit some mounted beam with enough force, because the gearbox+motor is pretty strong. (Or at least that's what I've heard, from a friend....)

So, there's a reason the Mars rovers cost umpteen gazillion dollars, even when they just carry a fancy drill and a flashlight ;-)

cod65
12-15-2018, 02:17 PM
what is 4s and 3s ?

KurtEck
12-15-2018, 02:21 PM
what is 4s and 3s ?
how many lipo cells that are connected in series. So 3s is the 11.1v battery and 4s is 14.8. Both obviously higher at full. charge

cod65
12-15-2018, 02:39 PM
Oh I'm using a 12v power supply, not batteries

jwatte
12-16-2018, 11:23 AM
I'm using a 12v power supply, not batteries

Are you using the Robotis SMPS ?

Various power supplies may be poorly regulated to spikes in load, and swing the output voltage wildly; especially when movement stops and the motor dumps load back into the power bus. I don't know that I've lost any Dynamixels to this, though -- mostly more powerful motors will give me this problem unless I parallel the power supply with a very big capacitor or a stabilizing battery.

tician
12-16-2018, 12:33 PM
Various power supplies may be poorly regulated to spikes in load, and swing the output voltage wildly; especially when movement stops and the motor dumps load back into the power bus. I don't know that I've lost any Dynamixels to this, though -- mostly more powerful motors will give me this problem unless I parallel the power supply with a very big capacitor or a stabilizing battery.

The vast majority of power supplies can regulate in only one direction (input to output) by limiting how much energy gets transferred from the input to the output. If anything pushes energy back onto the output, then those power supplies can do fuck all to keep the voltage in its target range because they cannot transfer any of that energy back to the input (all they can do is stop transferring energy from the input to the output). Big capacitors and/or batteries give the back-fed energy a place to go until the load increases again to drop the output back into the power supplies operating range where it might begin again in transferring energy from the input to the output. Bidirectional energy transfer is not something usually found outside of industrial motor controllers (certain VFDs with inverter on input to pull double duty as active rectifier and back-fed mains inverter) and uninterruptible power supplies. Using a complete half-bridge on many switching converter designs allows the supply to potentially reverse the direction of the energy transfer and/or dump excess energy through some of the switches, but that functionality is rarely present in any mass-produced controller ICs.

jwatte
12-16-2018, 08:29 PM
Most VFDs these days have hook-ups for brake resistors, which end up taking the load of overvoltage.
Many battery-driven motor controllers have a diode path back to the battery for "regenerative braking" (or set their bridges up to do that.)
Finally, fixed-voltage outputs may have some kind of overvoltage protection, either as a TVS / Zener diode to conduct on overvoltage, or as a crowbar circuit to clampt the voltage, or simply as a shutdown mechanism (which isn't super helpful, really, but counts as "over-voltage protection.")
Anyway, this is the reason you want to make sure you have good capacity or buffer "float" battery on the output of a regular power supply (not designed for driving motors.)
The Robotis 12V SMPS is allegedly somewhat better than your default Chinese wall wart, although I don't know which of these mechanisms it uses.
Anyway, this kind of problem is more likely to break the electronics, rather than make the windings overheat, so it's something to worry about, but not the cause of the original symptoms described.