Homing and Limits

Whenever a servo system starts up from either power-on or a system reset, all axes of motion must home themselves so that they know for sure where they are. Incremental encoders can only tell the controller that they have moved one "tick" on the encoder wheel, forward or reverse. It is up to the controller to count the ticks and add to (forward) or subtract from (reverse) the current position count. Without knowing where actual mechanical components of the system are at startup, the computer connected to the controllers can't know where anything is; it needs a place to start counting from. (For more on encoders, see servomotors.) The designer has to add an input device, usually some form of switch, the system will use to find "home" on each axis. Here, we will call these "homing switches" as opposed to "limit switches" for the sake of clarity. They are functionally different in the case of a servo system, although the term limit switch can be used for both.

In the simplest sense, a limit switch is used to limit the travel of a mechanism. Limits can be hard or soft. A soft limit sends a signal to the control system that it's time for the mechanism to stop or reverse direction. The control system should then receive this piece of information and act upon it. Should the controller fail, due to hardware or programming errors, and the mechanism not respond correctly, something will usually be damaged. If the motion of the offending axis continues in its malevolent direction, it should reach a second switch, the hard limit switch. This switch is not connected through software, rather, it as directly as possible cuts power to that axis, and perhaps all motor power to the rest of the machine as well. The hard limit switch should be designed to deal with full speed impact of the mechanism, and not fail from the blow. Placing the body of a plastic microswitch directly in the path of a traveling axis, for instance can result in the switch body being crushed. Use of levers, rollers, spring fingers, etc., and giving some space for "overtravel" of the switch before a mechanical stop is reached goes a long way toward reliability. Above all, if there would be danger of personal injury or equipment damage from an axis missing a limit do not depend on software alone to catch the error.

In the specific case of the machine being discussed here, two of the axes do not require any limit switches. The base motor and camera rotator are free to rotate and have no mechanical endpoints to their travel. There is nothing for either of them to collide with. They do, however, need homing switches. The camera arm does have definite ends to its travel, so in addition to its homing switch it has two hard limit switches, one at the top, and one at the bottom. Fortunately, they have only been used once, and that was during early testing.

For homing, all three axes use magnetic reed switches. They are small glass tubes with bendable contacts inside them. When the field of a nearby magnet flows down the length of the switch, the bendable contacts are attracted to each other and touch, closing the switch. Optical (photointerruptor), hall effect magnetic, or perhaps some sort of mechanical switch would have been other options. Reed switches were chosen because they are easy to mount, and with a strong magnet, are easy to keep out of the way of the mechanism. Accuracy was not a problem since during the homing exercise, the magnet always approaches the switch from the same direction at the same speed.

Reed switches are glass, so after connecting wires to them, they were bonded into heat shrink tubing for protection. The switch for the camera rotator was glued into a section of round brass tubing for mounting.

The reed switch for homing the base is slipped into a channel in a piece of 80-20. A wide notch was milled in the 80-20 to clear one of the mounting bolts for the bottom bearing.

The reed switch for homing the camera arm is also slipped into a channel in a piece of 80-20. This made it easy to adjust the home position of the axis. The magnet is a good distance from the switch when it activates.

A word or two about homing algorithms is in order. When the machine starts up, it has No Idea where it is. It has to find the homing switch without running out of travel. When travel is finite as is the case with the camera arm, the homing switch will, by design, be placed near one end of the axis's travel, not somewhere in the middle. The first thing the homing algorithm wants to know is if it's already sitting on top of the homing switch. If it is, it has no way of knowing by how much, so its first action will be to move off the switch plus some small amount of distance for wiggle room. Now no matter where the axis is in space, it's free to move slowly toward the switch, ready to stop instantly and zero out its position counter. This controller is now home. Homing can be rather more complex than this, depending on the nature of the mechanism but the basic idea will be the same.