I reload ammo to save money and get more accurate loads at the shooting range. It is a fun hobby but so is electronics and automation. I had the idea to take some of the repetitive tasks like measuring out powder and automate them.
The robot will have to detect the specific short tone given for a good load to initiate a pickup the dish, swing over to bullet tray, and dump into a funnel on a bullet casing. The robot will also have to detect the specific long tone given for a bad load to initiate a pickup of the dish and a dump back into the top of the powder scale. After either of these actions the robot will place the dish back on the scale. If the last action before placing the dish back on the scale was a dump into a bullet then the robot will move the funnel to the next bullet and wait for either a short beep or a long beep before deciding what to do next. This will loop until all bullets have been filled.
I’m not sure if I need a robot arm design or something else. I want to be able to program it in xyz coordinates of the end effector if possible. This means that on a move the dish will not tip because it will hold it level no matter what joints are moved. I don’t know what software to use. I didn’t want to purchase RIOS because I couldn’t see a demo. I don’t want to buy something until i know it will work the way I want to have it work.
Any discussion or suggestions would be appreciated. I’m still struggling on navigating the forum as this is only my second post. Please suggest some good topics for me to read that would help me on my project.
That brings back some memories! We measured loads by hand. Can it press the bullets into the brass as well?
You have a pick-and-place task. An X-Y table would be most useful. Very similar to testing ICs.
I’d consider a “lifting ring” that picks up the pan, and swings it over to a “dumping station”. And a pick-and-place to fetch a cartridge from the tray, put it on the dumping station, and then return it to it’s place on the tray.
The BasicAtomPro on a BotBoard II or an ARC-32 can detect the tones, and operate the lifting ring and the dump station, and run the stepper motors. For the XY, I’d go with a stepper drive and lead screws. R/C servos don’t have the range, and the accuracy needed to retrieve and replace the cartridge would be suspect.
For the XY, look at low cost wood router plans. The driver card can be quite simple. A driver card for the EggBot, and similar sized steppers with long leadscrews could work, for example.
Pressing the brass is done manually. I just want the filled casings sitting in a tray waiting for a bullet.
I’d like your idea except I would not like to have to lift the casings. I would like to attempt to fill them while they are in the bolted down tray. A simple reloading funnel would be transfered to each bullet to fill it. I also do not understand how a lifting ring would work if it has to tip the dish with the spout end down, wouldn’t it rotate out of index, wouldn’t it also fall out of the ring when tipped.
Do RC servos really have bad resolution and accuracy? I’m looking for about 1mm in xy space of resolution. It would also have to have ramping and accelleration capability for smooth movements. I hate spilled powder. I’m also afraid of using steppers in an open loop design as I do not want to fall out of index.
Thank you for your suggestions, I will look up XY tables for CNC to see if I can use anything from that.
It is possible to make a setup that will automate this process using hobby servos. But I agree not as a 3DOF arm. Make a single horizontal movement arm with a rotation on the gripper that can move the powder tray from the dispenser to a stationary funnel. Under the funnel would be a servo controlled rotating platter that has the casings spaced along the circumference. This would be very simple to set up and well within the servos resolution capability.
Yeah, I thought about it some more (on the drive to work), and one could certainly move the tray of casings via a simple X-Y table. The funnel would be fixed.
The ring was to to be able to lift from the pan from the scale, and transfer to a rotate-able “dump” platform above the funnel. An appropriately sized “gripper” could grasp the pan, swing it rho-theta (arm) over the funnel, and rotate the wrist to dump. Simple!
R/C servos aren’t meant for precision positioning (certainly not 1mm), and have limited range. Steppers easily easily get down to that. And open-loop is repeatable. If you’re concerned, just make a simple home “bump” to a pair of limit switched to calibrate. You can even simply run the XY carriage slowly into a convenient 0,0 mechanical stop, then zero your position counters. You can do all the ramping you want with them, not hard!
A pair of 1/4"-20 threaded rods and 50-100 oz/in stepper motors could do it for you.
But how would it go about dumping back the powder if it accidentally dispensed too much. It would need 16 inches of vertical travel.
If I went with a drilled circle pattern in a Polyethylene Block then I could mount it to the index table, the robot could be on a single horizontal axis and stop when a pair of limit switches are tripped before it is dumped into the bullet on the index table. I could use a continuous rotation servo to control horiontal speed, acceleration, deceleration upon hitting the first limit switch, and then it would creep the remaining distance to the final limit switch. There would be similar limit switches at the other end of the horizontal slide. I’d probably purchase the rotating gripper and attach it to a vertical slide that could go up and down 16 inches because to dump back a bad load it would have to reach the top of the scale container.
Another idea for the horiontal axis I had would be to modify a standard servo so that the potentiometer turned very slow over 100 turns of the spline shaft. I think this would allow it to stop at any turn of the shaft. 0 turns position would be obtained by feeding the servo 1.25ms pulse width, 50 turns would be 1.5ms, and 100 turns would be 1.75ms. I could potentially find any position in between then, right? Has anybody tried this? Is there a servo that already has the pot slowed down.
I was just prefering the robotic arm at first because this robot is what I am in charge of programming when I am at work.
It isn’t the actual robot I program but it is that same model, running on the same conveyor system, and even palletizing the same type of product as the one I am in charge of maintaining at work. Not to get too far off topic. I still might try to buy a used lab robot arm for this project, I love industrial servos and robot systems.
Kris
You could make a rotary cartridge holder, you’ll just have to load it first. You mentioned working in the cartridge case, which would be XY access.
Go one row further, past the end of the cartridge case, and dump excess powder there.
You can indeed swing an arm between two fixed stops, with limit switches if you like. Even use just a simple geared DC motor.
You need two rotational positions of the grip, horizontal to pick up the pan, and then another some 180 degrees to dump it into the funnel. Again, two stops, two limit switches, and a DC motor if you wish. The gripper rotation and arm swing don’t require a great deal of precision, R/C servos should be fine. Gripper again has two positions, closed on pan, and open. Could even be spring close, and open.
100 rotations of a splined shaft with an R/C servo will eat up all of your resolution. Tilt/Pan R/C servos might have a 5:1 or perhaps 10:1 gear ratio. I’ve worked with them. Adequate for camera, probably not at 100:1. Try it! Run a little 10-32 threaded rod through a nut (or pair of nuts for backlash comp), and see if you can hit your 10 x 5 (cartridges in case?) marks.
The circular shell tray rotated by a servo would be a good start. If powder hoppers are available that have an adjustable charge button on the bottom, then this could be operated via a servo to measure and supply the powder. A movable funnel with a top that is long and narrow could be placed between the hopper and the tray, and moved foward/back over each shell for the loading. The operation would be for the funnel to be moved across the tray, stopping over each shell. The hopper would drop a charge in the funnel to fill the shell. When the row is filled, the funnel would move back to the start position, the table would rotate to the next row of shells, and the process started again. A small vibrator motor could be attached to the funnel to make sure all the powder slid down to the shell.
So in researching the project I am thinking about I’ve found a magnetic encoder kit you can install into a rc servo that gives 0.08 degree rotation accuracy. It even compensates for gear backlash because the output shaft is the part with the encoder on it.
If I could build something like this it would be more adjustable to changes and other operations I might want to automate. I don’t know that I want to build something that is limited to just this one task.
Does anybody know of any other kits to modify a rc servo with some type of encoder? What really makes me want to build something with an encoder is that you could program the robot by positioning it by hand and storing the positions or even move the robot and playback the movements exactly as it was moved because the encoders would capture the movement.
I thought the Gecko Servo Drive would be useful with a 60rpm brushed gearmotor for the first joint of a robot arm. I’m sure I want to go way overboard and start a robot arm but I do not want it to be a toy. I want it to be as real as a lab type robot arm, no plastic servos. I agree that steppers would be cheaper as has been said above but I want the robot to know where it is at all times, it also needs to be able to detect a collision. I could do steppers with encoders but I was thinking I might as well go all the way and use brushed dc gearmotors.
Alan,
Have you used a G320? Do you know where I can find more information on the step and direction signals? My plan is to attach a high resolution encoder directly to the joint so that gearbox backlash is accounted for. I will use an optical quadrature encoder with index to define home position. I’ve actually looked all over their website for specs on how to tell the drive what do to.
Jim,
Are the RoboClaw motor controllers on Lynxmotion sort of like the Gecko drive but cheaper? Can it read an encoder and try to maintain position with a DC motor or is the encoder input designed for speed feedback only?
I don’t know if this helps me get where I’m trying to go with this project but I have a Propeller Microcontroller Development kit with proto board stashed in the closet. I have done anything but blink LEDs with it yet.
Well the step and direction, I’d assume it it a 5vdc TTL input, maybe if the direction is high you will turn clockwise, if low it will turn counterclockwise. If the pulse or step input goes through a high to low transition it will seek the next mark on the optical encoder. So I assume if I want to turn the motor one revolution and I have a 5000PPR encoder I could program a microcontroller to pulse 5000 pulses. If I wanted accelleration and decereration I would change the pulse width at the beginning and the end. I also assume that if the robot picked up weight and was able to turn the gearbox that the encoder would pulse and it would fight the weight to maintain position and fault out if it could not maintain position such as when a collision would occur.
I couldn’t find any documentation to support my assumption. Am I right in the asumption above?
On a CNC do you use the encoder on the motor or the linear DRO? Can’t you tune the PID loop to eliminate hunting or does it always run back and forth one pulse on the encoder just so it knows where it is? How do you compensate for backlash on a CNC? I’m sure the tolerences on a CNC are tighter than anything I am attempting.
Yes, basically the steps match the encoder for the G320 and similar step/dir driven DC servo drivers.
The direction line does indeed specify the direction. However, one does NOT change the pulse width to change accel/deceleration. The step pulses stay basically squarewaves. The pulse RATE changes.
The following error is monitored for faults.
Unless you have an exceedingly expensive mill (or lathe) with glass linear encoders, zero-backlash ball-screws, you put the encoder on the motor tail shaft. Backlash comp is done in software. Ball-screws help.
Normally one reads the Controller program screen to see the axis locations, not a DRO. No, you don’t dither (move back and forth) to improve resolution.
A good place it ask these questions is the CAD_CAM_EDM_DRO list on Yahoo, or my CNC list posted below. These are good questions, but a bit out of scope for your needs, I think.
Again, consider the wood-router type machines, and how they’re built. You’re not cutting metal, just positioning a relatively small mass with out cutting forces.
