Greetings. I’ve got a pair of the servo amps we’re trying to use in a science olympiad robot (I’m the coach). The problem we’re seeing is that we get either a relatively narrow range of motion, or practically no motion at all. It seems like the usable range of the feedback pot is very small.
I’ve tried both 5k and 10k pots. We’re using an Arduino to send the control pulse. We started with the specified 1100-1500-1900 pulse timing as specified on the product page (and that has yielded best results), but the range of motion is amazingly small.
We’re using standard Radio Shack pots, approx 300* mechanical range, with maybe 280* range of usable analog sweep. The pots are coupled directly to the shafts of two different joints. In one case, we’re only able to get ~ 100 degrees of travel, and on the other, we get basically no usable movement, although if I position the shaft/pot just-so, I can get the servo to follow the joystick over a very narrow range.
I’ve tried varying the signal timing to outside the 1100-1500-1900 range, and the amp doesn’t seem to like that at all. I’m looking at the Arduino serial monitor, so I can see that I’m getting good analog input, ranging from 0-1023, and that is being scaled to a 0-179 input to the standard Arduino servo library.
Any other ideas on what to try? If I can’t get a good range of motion, I may have to sacrifice another analog input and monitor the feedback pot directly in the Arduino, and use the servo amp (or another h-bridge) to just drive the motor without direct feedback, but that defeats the whole point of purchasing the servo amp.
Thanks.
PS - a rapid response would be appreciated - this is making me look bad for my Student Olympians!!
I just had a conversation with Devon here recently about radio shack pots. Apparently they are all audio taper, not linear. I believe the issue is the pots. Can you verify the pots are audio taper?
They’re always mixed together in the bins for sure, but linear pots are available. The five pots in our multi-axis joystick are linear for sure, I can see how the Arduino a/d is reading them. I will pull the feedback pot and check out manually though.
Another thought: does the amp “slow” the motor as it gets close to the set point? The motor that’s giving the most trouble is significantly under-volted due to the limits of the amp. It runs great at 6v, but the load might be such that it wouln’t necessarily run well if the amp was rolling off the effective voltage…
OK, I’ve confirmed that my pot is linear. Lifted two legs so the servo amp wasn’t in-circuit and measured resistance as I manually moved the mechanism/pot through the range - 4.7k to 0k
The centerline (2.5k) is not on the mechanical center of the pot’s sweep range, but that’s because the pot tops out value-wise about 20* before the end of the mechanical sweep. Even with that, I’d expect the servo amp to behave somewhat normally but with an offset.
I have also tried adding a 1k and 2k resistor in series with the pot wiper, hoping that the servo amp wasn’t able to deal with the extremes of the voltage divider for some reason (knowing this would cost me V/degree resolution, but just trying to get something working). No improvement with that.
Also mocked up another motor into the system - one that should be happier running at 6V (or less), and it exhibited the same performance. Compared to putting 6V directly on the motor terminals (for both motors), I also see more “peppy” movement in both cases, suggesting that the servo amp isn’t putting out full VCC (6v) under any of my test scenarios. Ugh.
Help. This is our last build weekend before Science Olympiad. Kids need to start practicing for the main event!!
OK, due to time constraints we gave up on using the servo amps for now, but I’ll keep this going since I’d like to use them for SOMETHING, now that I own them. For now, I’ve reconfigured things to free up a couple of analog inputs on the Arduino and we’re running the feedback pots directly there and driving the motors with bigger h-bridges. My servo algorithm probably isn’t as good as what the h-bridge could do, but it’s working with only a bit of twitchiness now.
Changing to a better quality pot (Honeywell) where the range of usable resistance was spread over a wider sweep (even though the mechanical sweep range was the same) did improve things a little.
I think this mostly comes down to a mechanical overload, though. I’m guessing that as the error (between commanded and actual position) decreases, the bridge is decreasing the duty cycle because it seems like the voltage being delivered to the motor is dropping off. There’s enough mechanical load in the system that at less than full voltage, things won’t turn as well (or at all). I think the load was just too high, overall, as more than once the h-bridge seemed to be getting hot while moving things around.
