Alex Brown's Scorbot robot arm project

      I've been thinking about building a robot arm for a long time.  I even brought in a prototype of a fairly strong (able to lift a six pack) shoulder joint a couple years ago.  But, haven't really gotten anywhere. 

     While visiting John Boisvert's robot building house in San Diego a couple months ago, I saw several Scorbot arms.   I offered to buy one, but John apparently never lets go of any good robot stuff :-)   But, Steve Vorres showed me that one was available on ebay for $200.  This seemed like a good price for a semi-professional robot arm.  I believe it was designed as sort of an educational robot for teaching industrial robotics techniques.  But, by our standards, it looks pretty good.  It has a 2 foot reach and can pick up about 2 1/2 pounds at full reach.

    By the way, please excuse if this page gets messy or confusing.  I'm just getting used to this method of building a page.  Right now I'm looking for how to insert an image......

Hey, it worked!

You can see it is a substantial size arm with 6 Pittman motors.  It is 5 DOF plus the gripper.  It comes with a controller box which can be seen in the left background.  But back to the beginning.

It was advertised "as is".  At first I tried bidding on one that was advertised as all working.  Well it went for well over $300.  So, I figured this stuff is low technology and I can probably fix most anything that isn't working.  So I bought the "as is" unit for the "buy now" price of $200.  (probably a good price since a later sale of a clearly not working Scorbot (gripper motor missing) went for $250.)

Well it came after a few weeks in a box that was basically falling apart the box was about 1/3 full of foam peanuts and the arm and controller were just floating around in this.  If the box had ever been turned over, the parts would have crashed on the ground...and probably fallen out.  Amazingly, there was no apparent damage...hooray for UPS!  Of course SOMEBODY beat up the box ;-)

The seller told me in a presale communication that he didn't know of anything wrong with the arm.  When it came it had two notes taped to it:  "Gripper doesn't work" and "Encoder cracked, needs to be replaced". 

Well, it turned out that the gripper motor didn't run.  I took it apart and found that one of the brushes had fallen apart.  I couldn't find any souces for replacement Pittman parts, so I found that the brush assembly from a newer Pittman which I had would work, although it was quite different.  It took a few tries and learning about properly shimmiing the endplay of the motor shaft.

The encoder on one of the 6 motors was actually missing.  It turns out an outfit called Intellitek still supports the Scorbot (did I mention this was build 26 years ago?).  I ordered a replacement encoder disk and a couple spares.  Well, when they got here, I found that they were quite different from mine.  And basically wouldn't work unless I changed the LED - photodetector board which was $75.  So, I learned to make a new encoder disk (fortunately, it was only a 6 segment disk) and it worked.  I actually had to make two.  The original disks are made of plastic and after 26 years are getting pretty brittle, so I broke another one.

This actually got most things working.  I later found that one of the microswitches which reports the "home" position of each arm wasn't responding.  Naturally it turned out to be a broken wire in the middle of a long wire bunder inside a braided sheath.  Took some effort, but I fixed that also. 

Now everything worked except the original tutorial software which was designed to run on something the vintage of an IBM PC-XT.   I tried to run it on a newer PC, but the software was apparently set up to use control signals as well as the usual RS232 signal wires which generally aren't supported any more.  Well, I have a PC-XT which hadn't run in 10 or 15 years.  And it actually had a serial card.  Well after a days work, I got it running, but the software didn't work on it either.

But, I really didn't need the software.  The controller box has basic motor control software that CAN be run by simple RS-232 communication from any PC.  So, I now am able to control everything that seems controllable.

Now, my real goal is to use this as an educational tool to learn how I want to build a real robot arm for Rocky.  This arm is quite large and weighs 28 pounds.  Even for Rocky it is bigger than I'd like...although I still might try to mount it on him.

So, what have I learned so far? 

Well, one thing is that the arm has a 24 inch reach and looks like it can't really reach very far.  Well, I measured my arm from shoulder joint to fingertips, and it's only 27 inches.  So, the robot arm can reach almost as far.  Except that a human can lean his body forward to reach around 48 inches across a table and robot will be reaching more like in the low 20s.

The Scorbot has all the motors mounted in the base so the arm doesn't have to lift them along with the payload.  Forces are transferred to the end of the arm using belts which are relatively light.  Another advantage to this method is that moving one joint doesn't necessarily affect the orientation of joints farther out.  If the gripper is sticking straight forward and you change the angles of the shoulder and/or elbow, the gripper remains pointed straight out. 

All the motors are backdriveable.  What this means is that when the arm is holding up a heavy load, the motors are continuously drawing a lot of current.  In a future design, I'd like to use a non-backdriveable system like worm gears or jack screws to hold the load without motor torque.

The electronics to drive the motors are all in that controller box, which runs off 115 vac.  To install this arm on my robot I'm going to have to design a replacement for the controller. 

to be continued....