Build Thread- 6 axis Robotic Arm project

[henry_phd]

I've been deeply interested in robotic arms lately and have been researching and design a 6 axis one from scratch.

I have purchased a Labvolt Armdroid that uses 6 stepper motors for 5 DOF + gripper and it's not bad but more of a toy than a tool (other words adult toy) so want to build my own.

I'd like to see if anyone is interested out there to help.

My expertise and capabilities:
-math and programming, I will write most code in Matlab
-electronics layout (can mill my own boards and design the circuit)
-3d CAD drawing (Solidworks)
-CAM machining, almost anything 3 axis we can draw I can program with CAMworks to machine in about 5-10mins

Weaknesses that would like help:
-motor control (stepper voltages, currents, drives, etc)
-gearing
-material selection

Goals:
-FREE full open source 3d model is solidworks that can be machined/built easily from the assembly file
-free code to drive it
-almost every part will be able to be made easily on a 3 axis CNC machine with minimum tooling or bought from master.com
-mostly made of 1/4" AL plate to reduce cost
-easy to assemble/take apart
-anything that can be controlled in software rather than hardware will be
-2 to 4 ft reach at full extension
-6 to 10 lbs at full extension
-6 DOF

Intentions:
-minimize use of gears to reduce weight
-use stepper motors (cheaper and less gear requirements) and microstep them, 400 steps/rev, prob NEMA 23 - 34
-accuracy .001-.002" repeatable
-absolute encoder on all axes (stepper motor for easier control but need absolute encoder to get solid repeatable accuracy)
-code to translate the spherical coordinates to XYZ moves
-speed to be around 200-360 degrees/sec
-perhaps similar to CRS Robotics F3 ($60,000 robot) but bigger
-material cost under $2000 (realisticly it's not going to be super cheap, motors, drives and encoders aren't super cheap)

Rather than just building it and leaving everyone else on their own, I am building a detailed model for each part. Every plate, every hole, every screw will be added into the 3d solidworks model. You can freely move each axis/part to see how the set screw holds the motor bracket in place and view the range of motion, calculate the weight, etc.

I did some prelim drawing and made up a quick wrist design. I will post the solidworks assembly file to show the quality of the drawing. You can hide each plate and machine them individually.

Later on I will even include CAMWorks CAM data so you can just select the machine you want to it will pop out the set up sheets and G-Code to build your own robot.

End results:
-robot arm can pick up a tool and carve out a sculpture in 3d with many undercuts
-robot arm can hold things/weld/etc
-robot arm can pick up a rum bottle, pour it into a glass and shake you a martini

[henry_phd]

My first steps are power and motor selection.

Let's assume the arm is going to do 4ft and hold 20bs (say 10 + 10 pounds of arm) so I need 80 ft-lbs at the base.

I want 360 degree/sec so around 60rpm.

Power = torque * rpm = 80 ft-lbs *2*pi*60rpm /33000 so close to 1 HP (745W) at the base!

That's a lot of juice.

Most stepper motors are rated with a voltage and current. My understanding is that the voltage doesn't mean anything because you can drive it at much higher voltage as long as you meet the current requirements.

Anyway, let's look at something like the NEMA 23 PH266M which can be bought for around $20 on ebay and has 400steps/rev.

It's rated at 6V and 1.2A so does that mean it's only a 6W motor?

[lerman]

I think that you will probably need to use servo motors instead of steppers.

Yup. Those little steppers just don't have much power.

Ken

[Adamj12b]

This sounds like something I can help with. I am very interested in robotics as well. I am an electronics engineering student and do this kind of stuff in my spare time.

Sense I was in high school, I have been designing a robotic arm that was very large. I first started in Autocad R14, then moved to Solidworks 2007 Student Edition. Best decision I ever made.

The arm uses servo motors linked to the different parts with #25 chain. My goals are for it to list 100 lbs, and be repeatable. I have estimated the cost of my arm to be around $1900 USD. But, Due to funding, it has never made it past Solidworks. I keep editing parts and fixing things, until the day comes where I can afford to build it.

Ok, I guess you would like to see it now? Can do.

There is still parts not on the arm yet, such as gears and chain, or the gripper. I haven't drawn the gripper yet, because I am stuck on how to draw it in solidworks. All the solidworks that I know, I taught myself.

I have 2 milling machines. One cnc knee mill, and one cnc benchtop mill. So I am able to make all the parts when it comes time.

Ok here is the picture.

As for other help, I can do alot of electronic work, as well as programming.

Look forward to seeing where this goes.

-Adam

[TOTALLYRC]

Originally Posted by henry_phd My first steps are power and motor selection. Let's assume the arm is going to do 4ft and hold 20bs (say 10 + 10 pounds of arm) so I need 80 ft-lbs at the base. I want 360 degree/sec so around 60rpm. Power = torque * rpm = 80 ft-lbs *2*pi*60rpm /33000 so close to 1 HP (745W) at the base! That's a lot of juice. Most stepper motors are rated with a voltage and current. My understanding is that the voltage doesn't mean anything because you can drive it at much higher voltage as long as you meet the current requirements. Anyway, let's look at something like the NEMA 23 PH266M which can be bought for around $20 on ebay and has 400steps/rev. It's rated at 6V and 1.2A so does that mean it's only a 6W motor?

As guess lets say that the max usable voltage is 40v based on sqrt of the mh *32=max v
40 x 1.2 = 48 watts, still not a lot of power.

80v and 7 amps is the max a g203v will handle. 560 watts. You are already in servo territory for the base motor. The other motions could get away with stepper, depending on fast you want them to go.

360 degrees per second is commercial speeds. Be very careful.

Mike

[z-darabian]

Originally Posted by TOTALLYRC As guess lets say that the max usable voltage is 40v based on sqrt of the mh *32=max v 40 x 1.2 = 48 watts, still not a lot of power. 80v and 7 amps is the max a g203v will handle. 560 watts. You are already in servo territory for the base motor. The other motions could get away with stepper, depending on fast you want them to go. 360 degrees per second is commercial speeds. Be very careful. Mike

Hi , I have a project like this , I'm a robotics engineering uni. student . I've got a team to build this . i am a designer in my group in solidworks (cad & cam) i want to determine power , torque , speed , acceleration & every parameter about motor but i don't many thing in SW. - why u say 360 deg/sec is a commercial speed . pls help me !!! if u have information about it
tanx a lot every one !

forgive me ! i'm not fluently in En because i'm Iranian & my language is Fa.

[Ray Ingersoll]

I've been experminting with a hydraulic driven robotics I have found using
low cost air cylinders with windshield washer pumps make sure you have
an alumnium housing with stainless steel rod I can pickup the pumps at
the junk yard for about $.50 the air cylinders are $10.00 I cannot hold
back the cylinder the pumps can be controlled from the computer with
an H bridge you can use a 50000 k ten turn pot then use the 5 k for 1 turn
of the pot which interfaces to the game port of the computer. cheap and
plenty of power.

Ray Ingersoll

[mactec54]

Hi henry_phd
It seem this will be a great build I would like to see one built as a milling machine just mount the robot arm & have a rotating table in front of it, with parts being cut,as each part was cut the table would rotate to the next etc, I know they already have robots that do this, but not a small one that you could build as a hobby

[oxford]

Any updates on the project?

[Ray Ingersoll]

The problem with the windshield washer motors is they let the fluid thru I've been
trying to find a cheap valve to keep the cylinder locked in position anyone have any idea,s welcome the input.

[keebler303]

Ray

You are looking for a check valve I think? Lets flow go to the cylinder but not return through the pump. You would have to have an additional valve or something to let the fluid leave the cylinder.

Matt

[Quebecois_Sti]

it will be hard to have an accuracy of .001-.002" repeatable with stepper motor, you need servo motor. To carve out correctly in material you will need a stiff, not moving arm.

I don't think it's gonna be possible at 2000$, instead you should check for used 6 axis robot.

but don't forget, nothing is impossible!

good luck in your project.