Originally wanted the arm for the TOM Tri-track, but decided to make a stand-alone robotic arm / excavator for my 7 yr old son. This is the result. Electronics: Cypress Psoc 1 CY8C29466 3210 kit under the hood. A serial ttl connection to the psoc made adding a bluetooth HC-06 simple for pairing with an android phone as a remote control. A 2 axis joystick controls swivel and boom operation, while a single stick moves the arm and single stick operates the bucket. This design was not well fitted due to inexperience and many reinforcing screws were added, but functioned as desired and it worked on Christmas Day as intended. Moving forward, I would change the servo in the wooden base for a geared stepper motor like a nema 17 or 22 for swivel movement. The boom servo on top of the swivel has trouble because of the weight of the arm. I would probably put a worm drive on that with a stepper instead of the servo. Great learning experience for me. Sorry the files are somewhat incomplete. This may qualify as a hacked version rather than a straight remix.
Before modifications. Servo unable to lift boom. Arm base unstable
After modifications. Spring helps servo and adjusted pivot points helps with servo load. Additional side slide holds arm down for swivel motion
UPDATE Nov 2018 - new base mount and the option to use stepper for boom lifting and the option to use servo or stepper for swivel
Boom vertical movement without servo: With 2 of the small chain stepper derailleur sprockets, one screwed to the side of the boom and one on a 28-BYJ stepper mounted with the shaft horizontal, you can use a bike chain to lift the boom (Thanks to Ledjo : https://www.thingiverse.com/thing:337033 - I filled in the center, added a 5 mm hole with 2 flat sides at 3mm apart to allow the chain ring to slide ride on to the servo shaft). This motion relies on the boom weight to move the boom down with gravity, so there is probably a better design here to pull the boom down.
Both ExcavatorBase II and III use a bolt to hold the base to the ExPlate2. If a lock nut is used, then you set it once and it holds. This is much better than the lurpy 4 screw hack with the bushing for rotation when everything was mounted on top of the servo.
For 2 stepper motors on the excavator base: the servospin gear was fixed to the base plate (ExPlate2) that bolts down where the old base plate mounted. When paired with the ExcavatorBaseII, attach the ExTurnGear to a 28-BYJ stepper motor mounted in the vertical shaft position to swivel the arm.
For 1 stepper and 1 servo, use ExcavatorBaseIII and insert a servo upside down in the rectangle hole and put the servospin gear on it. I tried this and had issues with the gear-servo mesh not holding very long - disclaimer, it wasn't printed in super high res and only with PLA, so perhaps with high res and PETG, it may be more structurally sound.
Thanks to (https://www.thingiverse.com/rainerunsinn -> https://www.thingiverse.com/thing:1674077) for the 28BYJ lego mount that I cut up to put on the base as the stepper motor cover.
ExcavatorBaseII in position. Uploaded version has additional material under vertical stepper to better place the large gear so it doesn't drag. Bike chain is a place-holder for now since moving stepper with chain loop will do nothing. Chain must be pinned on boom and the other end must be tensioned around the sprocket, but not reconnected to the boom.
Opinion on mechanics and electronics:
My opinion: I don't like the inexpensive servo durability having stripped out a few now. I'm not buying high end servos for this, just using leftovers, so there are solutions with servos that could work. I have better leftover steppers though they are slower. The trick is the second microcontroller. I had to add one that interprets the servo signal from the first microcontroller and output the stepper signal. The first one reads the analog controls and talks to the bluetooth radio. The intent is to have a 115200 bps serial communication so an HC06 can be used to tie it to an old android phone or kindle fire to play with or use the manual controls.
Random thought: It may be that there is just a lack of current for the stepper and servo motors.
I chose not to use arduino for the microcontroller because I have not learned how to change the PWM frequency for more than one stepper at a time, just the duty cycle (It must be possible or 3D printers would not use them - I'd like to make it simple so I'll look into it. A nano that will run 3 servos and a stepper or 2 servos and 2 steppers...). A PSOC4 chip allows me to change the duty cycle, the period, or the frequency on the fly, so it can run 2 steppers at independent frequencies while still interpreting a pwm servo signal. Once i get it all sorted out I'll post the code.