Shop OBEX P1 Docs P2 Docs Learn Events
CR Servo Moddin' Surprise — Parallax Forums

CR Servo Moddin' Surprise

ercoerco Posts: 20,256
edited 2011-10-25 08:01 in Robotics
I spent the weekend modifying several different servos for continuous rotation, and I was surprised that the final results varied considerably. I have done the hack many times, physically cutting the output gear & stops as required to rotate 360, physically removing the feedback pot, and replacing it with a small trimpot to null at 1.5 ms pulsewidth. That mod always works great on my big dumb Vigor VS-11 quarter scale servos. This weekend, I made the same mod to 3 standard-size servos a Blue Bird BMS-706MG (low-profile ball bearing) , a Futaba S3004 (one ball bearing), and a $4 China Ebay clone of a Futaba S3003 (bushing). The BlueBird and real Futaba worked, but were really too sensitive to control with any repeatable speed control accuracy, and there was hardly any deadband at the stop position. They went from full speed forward to full speed reverse over a very narrow range of pulsewidths. Surprisingly, the Chinese clone worked best for fine motor control, providing repeatable,wide-range speed control over a wider range of pulsewidths and having a wider deadband at 1.5 ms. For a robot, it's quite critical for the servo to stop 100% when you want it to. There's always some maddening drift and renulling to do, especially with a narrow deadband.

The higher accuracy of the precision servos worked against me here. Of course, the cheap servo which worked better electronically had a lot more mechanical slop. Much coarser gears (bigger teeth) on the output gear. And curiously, the real S3004 and the clone S3003 rotated in opposite directions for any given control signal.

I also tried putting the cheap electronics in the real Futaba servo, but it oscillated at certain spots and was unusable. Two very different motors that wouldn't interchange.

In just the right "app", a cheap servo can work better.

http://www.blue-bird-model.com/en/servos/bms-706mg.html
http://www.ebay.com/itm/400247378442 EBay servos

Comments

  • Martin_HMartin_H Posts: 4,051
    edited 2011-10-25 08:01
    Interesting. A large dead band and coarse accuracy works against a servo when used as intended, but for continuous rotation it sounds like a plus.

    My CBA robot uses GWS servos modified for continuous rotation. I've been really pleased with them because they rarely need renulling. Another pair of CR servos I have seems to have no dead band and creeps when I tell them to stop.
Sign In or Register to comment.