Micro Robot Arm Precision Test

What are some good tests to apply to this robot arm? We want to figure out whether or not the thing is working the way we think it is working. For example, we think there are no missed steps, but is this really true? The robot arm operates open loop. There are no rotary encoders involved. The microcontroller keeps track of the current positions of all the joints by counting the pulses that are getting sent to the stepper driver chips. The motors might slip and lose a step. And even if the motors are moving reliably, is the microcontroller counting the steps correctly? There might be a bug in the software.

We need data. So let's check how precise the robot arm is able to move. In the video below, the robot arm moves in a fixed manner. We mark 2 positions; One to check mainly for errors along the Y-Axis, another position for errors along the X-Axis.

The results show that the position used to check deviations along the Y-Axis got lower over time (the robot arm was moving for 5 minutes in total), while the error along the X-Axis seemed small. Since the error is in the same direction as the earth's gravitational pull, this might indicate that there indeed were some missed steps. But one could also argue that the precision is within margins considering the gears backlash is about 2.5°. It also seemed like the error was converging to some certain amount.

More time had to be invested to pin down the cause/causes of the error seen in this video. Since it is moving fairly accurate I call it a day. The next project's waiting!

Buy the 12V 28BYJ-48 stepper motors used in this project on Amazon (Affiliated link):

Read more here:

[PiefacePete]

I have just watched a video that may have the answer; The output shaft is connected to the output gear by a friction clutch!

It's quite possible you are not missing steps at all. There is a view of the output shaft and gear here;
I think it would be possible to lock the gear and shaft together without too much drama.

[merxellus]

I wasn't ready for that speed and accuracy

[voyager]

I really like that you also explain the math you have used to solve the real problem and use it into the project.I believe this is real engineering thank you for explaining all these stuff and keep up good work!

[bobbyp]

0:54 my dog when he smells another dog on me

[solosailor]

Nice! Your software is the most impressive tech, yet the machinations are mesmerizing to watch. Algorithms do not video well, alas...

[ianm]

Hi I am a sophomore in mechanical engineer and now I am realy interest with robotic

I want to as, what is key poin if I want to build a precision and smooth robot ? as a begineer is it right the mechanical design doesnt play significant role than the electrical/code sir?

[LeoCastroFot]

The problem is not with the stepper motors or gears, probably, but with the structure of the arm and its response to the natural expansion of the material and its connections added to the force of gravity.

[jaled]

Hello I congratulate you, excellent project, I have a question, which of these stepper motors did you use?28byj-48 - 12VDC 32-Step 1/16 Gearing or 28byj-48 - 12VDC 32-Step 1/64 Gearing. I ask you because I only get 1/16 Gearin

[michaegi]

This is really impressive. Last year I build a special advent calendar for my nephew. I used the same motors. The connection beween wood and the motors didn't survive the kids test. I really wonder how to connect these tiny axes. The only product I can find are the toothbelt connectors. Isn't there a connector like these for sevor motors available?
How are the axes conntected to the wood for your robot arm?

[felixman]

hey there, if you were to use trinamic step sticks/ any trinamic driver with "stealthchop" and "coolstep", you can be sure that no steps are lost because trinamic can read the back emf from the stepper to determine when a step is lost, or even the tourque, or load angle. AMAZING STUFF you might find super useful.

[grathado]

Would you be ok if I created 3d printed files for your robot design and released it with credit to you?

[gnorty]

If I had to bet on it, I would say the errors are coming from the elbow joint missing steps. it's doing most of the work in raising the arm against gravity at reasonable extension. the shoulder joint is moving much more gently when the arm is extended, so less torque on it. I don't think the wrist will be a problem moving that small load.

A great test though. Perhaps you could do a similar test isolating the movement to a single axis and measuring the movement on a curve. It would be particularly interesting to see how the shoulder joint performs with the arm at different extensions.

How accurate are the measurements between the pivot points? I am curious to know how much difference a millimeter or 2 difference in the effective lengths affects accuracy.

[victorzamacona]

What kind of engineering do you learn to be able up create these things

[Flare]

I have looked at a lot of your code and other videos. Do you think it's possible to use 6 steppers and 6 encoders set up like your last servo projects? Following an encoder arm?

[kennethkindrich]

How do you get the stepper motors to be so I try to make them faster on my arduino but they only move slow and vibrate

[charimuvilla]

Shouldn't it reach the exact same position each time? What is the reason for the slight inaccuracy? Missed steps?

[henriktoth]

Can you write here the programming CODE please?

[area]

Do these steppers really move this fast? I've always seen them at like half that speed in all the videos online.

[mikealbert]

Looks great. Can this be used on a SCARA style arm?

[Gambiarte]

If the motor does not lose steps it will always stop in the same place.