Home built Makelangelo has jitter issues
- This topic has 11 replies, 2 voices, and was last updated 7 years, 2 months ago by .
-
Posts
-
I have built my own Makelangelo and WOW! this rocks.
However I have had some issues (as any DIY project will). I am hoping you may enlighten me on the final issue I am having.
I used the 3d Print https://www.thingiverse.com/thing:798076 a arduino and motor shield. I have two sets of electronics down to the steppers and servos. When I apply 12v via my bench top power supply the system acts very weird and overheats the shield but when I take it down to 5v all works great but for one thing. Transit moves on the gondola are nice and smooth but when it is drawing and the servo is down the steppers get jittery. Here is what I have come up with on testing:
1. Disconnecting the servo does not get rid of the jitter on both units
2. Power over 6v to the shield makes the machine almost epileptic and the shield gets very hot on both setups
3. Changing the speed of the pen in the settings to 500 and everywhere in between 6500 and 500 makes no difference
4. Swapping out the steppers made no difference
5. The jitter is more pronounced on things like the Darth Vader DXF I made via your instructions
6. When I do a test fiboacci lager curves are ok but the tighter the curve the more jitter.
7. I have changed the code to reflect the 200 steps of a standard stepperI am not sure where to go from here to try and smooth out the motion. Any suggestions?
If needed I can make some videos to share to show the issue both with jitter an also with applying 12v (which seems to be suggested in your instructions)Thank you and amazing work!
Can you show a picture of your machine, please? Especially a closeup of the electronics.
It looks to me like the 5v servo power source is wired to the 12v motor power source. I’ve only experienced bad things doing that. mixing voltages usually kills the magic rock.
if you ever code something that "feels like a hack but it works," just remember that a CPU is literally a rock that we tricked into thinking
— daisyowl (@daisyowl) March 15, 2017
The jitter I’m seeing looks pretty normal at low speed. the steppers are pausing just long enough to translate the shake down the belt to the pen holder. One solution is to use steppers with a higher step count (we use 0.9deg steppers, or 400-per-turn). Another would be to dampen the vibration on the belt. Our units roll the belt around a peg on the bottom inside corner of the motor. This also keeps the belt at a fixed spot: rolling around the pulley creates a difference between the mathematical model in the robot’s brain and the actual model, also known as i-wanted-a-circle-and-got-an-egg effect.
This is going to get technical, but it’s relevant.
The motor shield talks with i2c to the arduino. this is great for adding lots of devices but it bottlenecks communication. There is no way with the AMS boards to say “ok, everyone, here are the target steps and direction to reach in N steps, now i will send pulses and you step at the right time.” instead I have to instruct each motor individually. Even a properly powered board has an i2c delay every step. That might only be 50us but it adds up really fast.
For that reason when I was looking for a better board I went with the RUMBA. It can “bit bang” the steps with a much smaller delay between steps (faster, smoother). It can also be upcycled to make a 3d printer (I made a prusa i3) because that’s responsible.
Board without drivers: https://www.marginallyclever.com/product/rumba-board-no-drivers/
You’ll need one driver per motor: https://www.marginallyclever.com/product/a4988-pololu-driver-single/
You might also want our 400-step motors: https://www.marginallyclever.com/product/stepper-motor-nema17-12v0-3a-400-steps-per-turn/
- You must be logged in to reply to this topic.
