A solenoid is a coil that pulls (or
pushes) a metal rod called
a plunger when current flows through it. There are many kinds of
solenoids: pull and push types, with and without springs (to push back
the plunger when current no longer flows), with and without latches,
etc. The main advantage of solenoids over motors is that they can
effect a linear motion using a compact mechanism with no gears. On the
negative side, solenoids have a very short stroke (the length of
movement of the plunger).
In this project I tried to utilize a small solenoid that I removed from an old Iomega 100MB Zip drive. The solenoid itself, which you can see in the picture below, is smaller than 2x3x3 Lego units. It is a pull-type solenoid with no spring to push the plunger back out. The plunger ends with a wide disk that is designed to pull something with it. The resistance of the coil is only 3.6Ohm, so if you connect it to a 9V supply, it will draw 2.5A (after a ramp-up period).
The next few paragraphs explain how to integrate the solenoid into a Technic structure, how to use it to fire the Technic cannon (a spring-loaded Technic part that fires a little missile with a foam head when you pull back a trigger axle), and how to drive the solenoid from the NXT, either using a motor port (really easy) or a sensor port (more challenging).
A video shows the sensor-port activated solenoid in action: the mechanism charges for 15 seconds and then fires. You can of course wait for some external event before firing, like detecting a sound or a nearby object.
Warning: connecting your NXT to any home-made gizmo (like the one described here) can damage it. Beware.
The first step was to integrate the solenoid into a Technic contruction. I ended up encasing it in a Technic box with a 2x3x3 cavity with a slot in front for the plunger. To make the fit tight, I used cut-to-size pieces of thin craft foam (about 1mm).
Fortunately, the little disk at the end of the plunger fits very nicely into the slots in the axle joiners, so I used them as a Technic extension of the plunger. I initially used a rubber band to pull back the plunger (but see below for a better idea).
The useful travel of the plunger is about 1/2 Lego unit. If you pull it by more that this, the solenoid might not be able to pull it back. This is not much. Here is the whole thing together.
An alternative construction might be to encase the solenoid in two 2x3 bricks stacked one on top of the other and hollowed out. This would be much more compact than the Technic box that I constructed. You would need to close off the bottom with a 2x3 plate, but the studs of the plate might not leave enough vertical room for the solenoid (but I didn't try). A potential disadvantage of this costruction is that the front wall of the bricks might further limit the movement of the plunger, which is short anyway.
It took me a while to figure out what this construction is good for. The stroke of the plunger is short, and to keep the plunger pulled requires a lot of current. These considerations rule out many potential applications, like activating a pneumatic valve (this is a common use for solenoids, but this particular solenoid is a poor fit fot the Lego pneumatic valves).
Then I remembered that there is one Technic mechanism that is activated by a small momentary movement: the Technic competition cannon. As it turns out, the solenoid is perfect for triggering the Technic cannon.
The one difficulty is mounting the cannon so that the solenoid can trigger it. I had to offset the solenoid mechanim and the cannon by half a stud in two dimentions, which is what the cams and the "Technic Axle Joiners Perpendicular Double Split" are used for. The plunger assembly pulls a "Technic Pole Reverser Handle" that is mounted on the cannon's triger axle; the handle is not pushed all the way on the trigger, but there is a gap of about 1/4 of a unit to keep the plunger horizontal.
I connected the solenoid to one of the NXT's motor ports (more specifically to port A) in series with 2 resistors of 1Ohm nominal resistance each. The resistors were a little higher than specified, about 2.5Ohm total. I am not sure that they are necessary, but I felt that this would be safer than connecting a coil with only 3.6Ohm resistance directly to the motor port. With the resistors the total resistance is more than 6Ohm, which limits the port's current to less than 1.5A.
I activated the solenoid by turning on Port A at 100% power for 10ms. This activates the solenoid reliably and fires the cannon. To understand what exactly is going on, I measured the port voltages and the current through the solenoid using an oscilloscope. The graphs below show the results.
As expected, the port voltage goes high almost immediately. The current ramps up because of the inductance of the solenoid's coil. This causes the port's voltage to drop from about 7V (The NXT used rechargeable batteries) to less than 5V. Then the current continues to drop. I am not sure why, perhaps because of the movement of the plunger. At some point the current climbs again, perhaps the result of the plunger reaching the end of its stroke. When the NXT powers off the port, the current ramps down gradually (again due to the inductance), which causes a short negative-voltage spike until the protection diode in the NXT starts conducting. The protection diode stops conducting when the votage across it drops to below the diode's threshold, so the low negative voltage in the port lingers for a while longer.
The next driving method uses a sensor port rather than a motor port. This leaves the motor ports available, well, for motors. The circuit is an I2C circuit so it can be used on the same port with other I2C sensors.
The key here is to slowly charge a large capacitor using the 4.3V supply, and then to discharge it all at once through the solenoid to activate it. To work, the capacitor need to have (1) high enough capacitance to store enough charge to activate the solenoid, and (2) low enough internal resistance (called ESR) to discharge quickly enough to generate a large current through the solenoid.
The graphs above show roughtly how large the capacitor needs to be. The activation above uses about 500mA for 10ms. This is equivalent to 0.005 coulomb of charge. At 5V, a capacitor would need to be 1mF (1000uF) to store that much charge. In reality, the capacitor would need to be larger, because as the capacitor discharges, the voltage across it drops, reducing the current through the solenoid. So the capacitor would need to be several times larger.
After a few experiments I was able to get it to work. The circuit slowly charges seven 1000uF capacitors. A PCF8574 (actually a PCF8574A, which has a different I2C address) controls a darlington transistor configuration. Most of the time, the transistors should be off, which essentially disconnects the solenoid from the circuit. When the 8574 turns its outputs on, this turns on the darlington pair and discharges the capacitors through the solenoid, thereby activating it.
The capacitors are charged through a 1k resistor, which gives a 7-seconds RC constant. In the program, I let it charge for 15 seconds before firing, to let the capacitors charge more fully. You can use a larger value resistor, say 2.2k, to reduce the maximum charging current, but this will also slow down the charging phase.
Turning on the transistors requires a bit of care. The 8574 can only supply about 100uA of current, which is not enough to turn the darlington pair on completely. The effect that you see on the scope is (1) the pin of the 8574 does not reach anywhere near Vcc, and (2) the voltage of the capacitors drops down linearly, which means that the switching transistor is only allowing constant current to flow, which in turn means that it is not saturated. I solved this by adding a 1K pullup resistor to the darlington base, which provides enough base current to energize the solenoid. But this is an awkward solution since the pullup resistor is now drawing about 5mA when the capacitors are charging. Not terrible but still a waste.
An I2C chip like the MCP23008 that can provide 25mA (in any direction, to Vcc or to ground) would be a better choice here (but I don't have one right now).
I used a discrete darlington pair (with a high-voltage power transistor that I removed from an old PC power supply, but you can use any NPN transistor that can pass 2A or so). To activate several solenoids, a chip like the ULN2003 or the ULN2803 would be a better choice: these chips combine 7 or 8 darlington pairs that can pass 500mA each in a small package. (Thanks to Michael Gasperi who suggested the use of the ULN2003 for a stepper controller for the NXT.)
© 2007, Sivan Toledo