Anyone know if there is a Digital Continuous Rotation Servo?
Application is my ASB13 'Bot, so HS5485 or even HS5645 servos would be appropriate. Can these servos be modified for continuous rotation?
Currently using Parallax (Futaba), GWS S03N, and HSR-1425CR servos (different 'bots).
Getting two R/C servos of the “analog” type to track is not as easy as I’d like (I know, add encoders).
Other alternative (tests to follow) is to simply ignore the electronics in a pair of GWS servos (I fried one!), and use a pair of H-bridges (SN754410) and PWM to drive them (experiment under way).
Thoughts?
Alan KM6VV
Edit:
Hmm, I just looked at a competitor’s website, and it looks like these two digital servos are available in Continuous for +$20. Is this just from programming, or does the servo need to be modified (pull a pin?)? Does LM offer these in continuous? Please?
The drifting of continous rotation servos is probably due to the low quality of the internal pots or resistors that may be used in place of a pot. Using a 15 turn trim pot might help stabalize the drift.
The pot shouldn’t drift that much for the same ambient temperature, and fixed resistors can “fix” that. I suspect it’s the comparator/ “pulse stretcher” analog circuitry that’s used.
The drift is not due to low quality components, but rather the fact that many of the components are sensitive to their values changing with the temperature. The design was never intended to be as precise as we want, so it’s not a design flaw or low quality components. Changing to a 15 turn pot will not have any effect on the temperature sensitive components inside the servo. The digital 5485 would be easy to convert to CR. You have to separate the pot from the final gear. This can be done by removing an insert in the final gear so the two no longer touch, or it can be removed from the case and adjusted to ensure it stops with 1500uS pulses, or finally you can replace it with 5k resistors used as a voltage divider. There is a mechanical stop on the final gear that has to be removed as well. After that it’s ready to go. I don’t know how much of an improvement would be seen though.
So how do the digital servos work? Do they measure the pulse width digitally (uP), and then digitally control the DC motor with a PID (PD) loop? If so, then I’d expect an improvement. only the DC motor remains that could be slightly different between servos. (assuming the uP time base is accurate, internally they are R/C, no XTAL, I think).
Can you comment on this “setup” charge? I’ve had these R/C servos apart (Futaba, GWS anyway) before, and would expect to be able to do the continuous mods, as you’ve suggested.
What all does the programmer (HPP-21 or your HMI?) do? Do we need it?
The write-up on digital vs analog servos is interesting, but it points up some issues with using the digital servo as a continous rotation servo. The main problem is displayed by the us/torque graph. It appears that the digital servo would only have a ~30us band to go from a stopped condition to full speed rotation, where the standard servo has ~100us. It appears that variable speed control would be ~3x better with the standard servo. It also says that the digital servos apparently have much the same pots and such as an analog servo, so the same drift issues may be present.
I was mainly counting on more accurate PWM in to PWM out. Digital circuitry should do that. The “tighter” torque band would certainly not help.
I just wrote a “trim” function that should have allowed me to trim the speed of my two servos at various % uS outputs between 1500 (null) and 2500. There was practically no ramp-up, the speed went practically immediately to max just past 1500. This is with the Parallax CR’s. I have some Hitec and GWS servos to try as well.
The best bet is probably just the raw motors in the servos, an H-bridge and full PWM. We’ll see.
