MAGNETIC LEVITATION DEVICE
These Zetex power transistors are tiny plastic-case types, and I found that they run fairly hot. The overheating problem would be even worse if we used a 12v solenoid with its smaller resisitance, or if we tried to levitate a larger bar magnet. If I were to build this device again, I would use TO-220 style power transistors, such as the TIP-120 series. Metal heat-sinks are probably not required for these TIP types. For example, try these:
I was able to float a 2" long by .25" diameter alnico bar magnet with as little as five coil assemblies (two under each end of the magnet and one extra assembly positioned on axis with the magnet, to stop end-to-end motion.) VERY unstable though; the levitation lasted just long enough to snap a photo before the magnet wiggled out of the coils. The 12-coil, two-rail device was better, but still unstable. Placing bulk copper or aluminum near the floating magnet (a copper rod or thick plate, or even stacks of pennies) can give E/M damping and partially prevent the wiggling motion.
HOW IT WORKSFirst the basic circuit operation: each magnetic field sensor controls the polarity of one electromagnet coil, and also the polarity of the coil affects the magnetic field sensor. This forms a feedback loop which controls the average current in the coil. When the end of the coil's iron core becomes "north", the sensor detects this and makes the current in the coil reverse direction. Then, when the end of the core tries to become "south", the sensor reverses the current again. In other words, the sensor trys to keep the magnetic field turned off! It's a feedback system which dynamically creates *zero* magnetic field.
What good is this? A device that creates zero magnetic field? Very
useful, actually. Superconductive plates also create zero field inside
themselves and thereby repel both poles of a permanent magnet. That's how
superconductive levitation works: a substance with zero-field is a natural
repeller of magnet poles. Therefore the above coil-assemblies in my
maglev device will repel a bar magnet pole regardless of whether it is an
N or an S pole.
OK, so normally the feedback is working, and the coil's polarity is
rapidly changing back and forth, with an overall average of zero field.
When a the N pole of a bar magnet approaches the Hall sensor, the circuit
will still try to keep the field inside the Hall switch at zero. To do
this, it sends a current through the coil, so the current is more in one
direction than the other, which makes the coil's magnetic pole, on the
average, become "N" rather than zero, and it repels the magnet. (Alike
poles repel, so the North pole on the end of the coil repels the North
pole on the magnet.) The two oppositely-pointing magnetic fields cancel
out to zero right between the magnet and the coil, right where the Hall
sensor sits. If the N pole of the bar magnet gets even closer, the Hall
sensor will tell the electromagnet to repel the bar magnet even more
strongly. If instead we hold the "S" pole of the bar magnet near the
sensor, the average current in the electromagnet will reverse, and the bar
magnet will still be repelled. It's a negative feedback system based upon
a naturally occuring oscillator and Pulse-Width-Modulation. Cool, eh?
Before undertaking a huge maglev railroad project, build a single
"repulsor" coil to get a feel for operation, and to demonstrate the
interesting repulsion effect. If you build large numbers of coil circuits
and they turn out not to work, you'll have to debug ALL of them. Better
to get the bugs out of a single one first.
DETAILS OF OPERATIONWhen each Hall switch is turned on, it pulls down on the voltage at the base of the first NPN transistor. This turns that transistor off. As a result, the 10K resistor can pull up on the base terminals of the two power transistors. This turns the NPN PWR transistor on, and it turns the PNP PWR transistor off. This connects +24 volts to the electromagnet coil, and a current appears in the coil. A magnetic field appears, which turns the Hall switch off, and this causes all of the transistors to change state. The current in the electromagnet coil reverses, which turns off the Hall switch, and the whole cycle repeats. As a result, the device acts like a buzzer, the Hall switch turns on and off very rapidly, and the magnetic field wiggles north and south a little, but the average field is zero. When a magnet pole approaches, the Hall sensor starts keeping the coil switched to north longer than to south (or vice versa) in order to cancel out the field, keeping the field inside the sensor averaging at just about zero.
A pair of rows of these devices acts as a maglev railroad track. A bar
magnet will fly along above them if it is gently thrown between the rows.
Other ideas: build one coil into a hollow silver plastic rod, run it from
batteries in the rod, then and show your physics teacher that one end of
your "metal rod" repels BOTH ends of any bar magnet. He/she will freak,
because only a room-temperature superconductor can do this. Room-temp
superconductors don't exist. Yet.
It's simple, but unstable
One serious problem with the maglev device: instability. Magnet wobbling builds up until the magnet is thrown clear. This occurs because tiny movements in the suspended bar magnet trigger the compensating magnetic field after a small time delay. This will trigger slightly larger movements, which triggers even larger movements, and after a few seconds the magnet will be wiggling so violently that it will be thrown out of the device. It's like the opposite of friction. The device has slight "negative stability." If you can get the magnet to stop moving initially, the oscillations will build up very slowly or not at all.
A simple cure for the wobbling: place a thick non-ferrous bar or plate
right below the bar magnet. A hunk of aluminum works fine, and copper
works even better. The electrical resistance of the bar will offer some
mechanical damping to changing fields, this will dampen the magnet motion,
and the quivering will die away. Another solution: Replace each circuit
with an ANALOG Hall sensor and a DC power amp (needs big heat sinks,)
route the analog hall sensor signal through the amp and to the coil, and
use op-amp chips to add a bit of differentiated Hall-sensor signal into
the main signal going to the DC power amp. This will create some
programmable damping via analog computation, and will cancel out the
effects of the loop-delay which causes the oscillation in the first place.
And if you succeed in all this, you will be an expert in linear control
theory and proportional motion control systems!
Hall effect sensors for Maglev CradleChoose a sensor which has a digital switching output (not a proportional analog voltage output.) The sensor should respond equally to fields of either polarity, so choose a "bidirectional" type that has equal and opposite B(on) and B(off)ratings. I suspect that high-sensitivity detectors might work better than insensitive ones, since this should reduce the "dither" current in the levitation coil and make the system run a bit cooler. Anything below 10mT (100 gauss) is high-sensitivity. I've not tried swapping insensitive for sensitive detectors. Maybe a more sensitive detector would give less time delay, more negative damping, and therefor better stability. That's an experiment for someone to try.
HAL102-ND Hall effect sensor IC
Price: about $1.00 Supplier: digikey.com Output type: sinking (O.C) V(supply) = +.3V to +13.5V I(supply) = 20mA Io(max) = 250mA Bon = +6.5mT, +65gauss Boff = -6.5mT, 65gauss Hyst = 4.5mT, 45gauss Max freq = 50KHz _ / | || Pin 1 output | | || Pin 3 GND | | || Pin 2 Pwr (+12Vdc) \_| Bottom ViewFor projects involving large arrays of pcb-mounted sensors, see Panasonic DN6849S-ND from Digi-key. That sensor has a 4-lead surface mount package and costs about $1.50