Servo Issues

[EAGrahamJr]

Attempting to use the ServoDevice on GPIO 18 and I'm not having a lot of luck. If I set up the TOWERPRO_SG90 trim, the servo basically doesn't move at all. If I set up the same pin with PwmOutputDevice, I can get the servo to move (using setValue), but has basically two settings:

PwmOutputDevice device = new PwmOutputDevice(18);
device.setValue(.049f);  // swings to ~ 30 degrees
device.setValue(.05f);   // swings to ~ 170 degrees

This, on the other hand, does not engage the servo at all:

ServoDevice device = ServoDevice.Builder(18).setTrim(TOWERPRO_SG90).build();
device.setAngle(30f);

Am I missing something or is this really more an artifact of the Software PWM implementation? (I really don't like Python and was hoping diozero was my "Get Out of Jail" card.)

[mattjlewis]

I'll look into this ASAP

[EAGrahamJr]

Oh, there's no hurry! Ah, forgot - this is on a Raspberry Pi 3B+

On the other hand, I did find the very nice PCA9685 device already implemented, so it's probably not that big a deal. 😏

Thanks for a very sane (and comprehensive) library!

[mattjlewis]

FYI I've just pushed an update that includes a few fixes in this area. There's a new set of unit tests for servo devices via native servo control (e.g. Picon Zero, Arduino), as well as hardware and software PWM.
Are you able to pull and do a local build? I might be able to test with real servo devices this week, otherwise it won't be until after Xmas. I made some notes here for enabling hardware PWM on the Raspberry Pi - this is detected automatically and software PWM is used as a last resort.

[EAGrahamJr]

Yes, I can build locally 😃 - I can loop back around and try out the servo this week. The notes on the GPIO pins are really helpful.

[EAGrahamJr]

Servo fix verified for both hardware (18) and software (23) PWM. Both produce full sweeps from 0 to "real close to" 180 degrees.

[mattjlewis]

That's brilliant, thank you for testing. How jittery was the software PWM controlled servo?

[EAGrahamJr]

They both had moments of jitter (which was oddly not reproducible unless I ran the exact same sequence), but both were overall pretty steady. I was using the angle setting, so I would imagine that rounding errors probably contributed somewhat.