1. Brief description
Levitron is a device that keeps an object in balance with the forces of gravity using a magnetic field. It has long been known that it is impossible to levitate an object using static magnetic fields. In school physics, this was called a state of unstable equilibrium, as far as I remember. However, with a little desire, knowledge, effort, money and time, it is possible to levitate an object dynamically by using electronics as feedback.It turned out this:
2.Functional diagram
Electromagnetic sensors located at the ends of the coil produce a voltage proportional to the level of magnetic induction. In the absence of an external magnetic field, these voltages will be the same regardless of the magnitude of the coil current.
If there is a permanent magnet near the lower sensor, the control unit will generate a signal proportional to the field of the magnet, amplify it to the desired level and transmit it to the PWM to control the current through the coil. Thus, feedback occurs and the coil will generate such a magnetic field that will keep the magnet in balance with the forces of gravity.
Something abstruse everything turned out, I'll try it differently:
- There is no magnet - the induction at the ends of the coil is the same - the signal from the sensors is the same - the control unit gives the minimum signal - the coil works at full power;
- They brought the magnet close - the induction is very different - the signals from the sensors are very different - the control unit gives out the maximum signal - the coil turns off completely - no one holds the magnet and it starts to fall;
- Beckons falls - moves away from the coil - the difference in signals from the sensors decreases - the control unit reduces the output signal - the current through the coil increases - the induction of the coil increases - the magnet begins to attract;
- Beckons is attracted - approaches the coil - the difference in signals from the sensors increases - the control unit increases the output signal - the current through the coil decreases - the induction of the coil decreases - the magnet begins to fall;
- A miracle - the magnet does not fall and is not attracted - or rather, it falls and is attracted several thousand times per second - that is, a dynamic balance arises - the magnet simply hangs in the air.
3.Design
The main element of the design is an electro-magnetic coil (solenoid), which holds a permanent magnet with its field.78 meters of copper enameled wire with a diameter of 0.6 mm are tightly wound on a D36x48 plastic frame, about 600 turns. According to calculations, with a resistance of 4.8 ohms and a power supply of 12V, the current will be 2.5A, the power will be 30W. This is necessary for the selection of an external power supply. (In fact, it turned out to be 6.0 Ohm, they hardly cut more wire, rather saved on the diameter.)
A steel core from a door hinge with a diameter of 20 mm is inserted inside the coil. Sensors are fixed at its ends with hot glue, which must be oriented in the same direction.
The coil with sensors is mounted on an aluminum strip bracket, which, in turn, is attached to the housing, inside which is the control board.
On the case there is an LED, a switch and a power socket.
The external power supply (GA-1040U) is taken with a power reserve and provides current up to 3.2A at 12V.
An N35H magnet D15x5 with a glued Coca-Cola can is used as a levitating object. I must say right away that a full jar is not good, so we make holes at the ends with a thin drill, drain a valuable drink (you can drink if you are not afraid of chips) and glue a magnet to the top ring.
4.Schematic diagram
The signals from the sensors U1 and U2 are fed to the operational amplifier OP1 / 4, connected according to the differential circuit. The upper sensor U1 is connected to the inverting input, the lower U2 is connected to the non-inverting one, that is, the signals are subtracted, and at the output OP1 / 4 we get a voltage proportional only to the level of magnetic induction created by the permanent magnet near the lower sensor U2.
The combination of elements C1, R6 and R7 is the highlight of this scheme and allows you to achieve the effect of complete stability, the magnet will hang in its tracks. How it works? The DC component of the signal passes through the divider R6R7 and is attenuated by 11 times. The variable component passes through the C1R7 filter without attenuation. Where does the variable component come from? The constant part depends on the position of the magnet near the lower sensor, the variable part arises due to the oscillations of the magnet around the equilibrium point, i.e. from a change in position in time, i.e. from speed. We are interested in the fact that the magnet is stationary, i.e. its speed was equal to 0. Thus, in the control signal we have two components - the constant is responsible for the position, and the variable is responsible for the stability of this position.
Further, the prepared signal is amplified by OP1/3. With the help of a variable resistor P2, the necessary gain is set during the tuning phase to achieve equilibrium, depending on the specific parameters of the magnet and coil.
A simple comparator is assembled on OP1 / 1, which turns off the PWM and, accordingly, the coil when there is no magnet nearby. A very convenient thing, you do not need to remove the power supply from the outlet if the magnet is removed. The trigger level is set by the variable resistor P1.
Next, the control signal is applied to the pulse-width modulator U3. The output voltage range is 12V, the frequency of the output pulses is set by the values of C2, R10 and P3, and the duty cycle depends on the level of the input signal at the DTC input.
The PWM controls the switching of the power transistor T1, which in turn controls the current through the coil.
The LED1 LED can not be installed, but the SD1 diode is necessary to drain excess current and avoid overvoltage at the moments when the coil is turned off due to the phenomenon of self-induction.
NL1 is our homemade coil, which is dedicated to a separate section.
As a result, in equilibrium mode, the picture will be something like this: U1_OUT=2.9V, U2_OUT=3.6V, OP1/4_OUT=0.7V, U3_IN=1.8V, T1_OPEN=25%, NL1_CURR=0.5A.
For clarity, I apply graphs of the transfer characteristic, frequency response and phase response, and oscillograms at the output of the PWM and coil.
5. Choice of components
The device is assembled from inexpensive and affordable components. WIK06N copper wire turned out to be the most expensive, for 78 meters WIK06N paid 1200 rubles, everything else, taken together, was much cheaper. There is generally a wide field for experiments, you can do without a core, you can take a thinner wire. The main thing to remember is that the induction along the axis of the coil depends on the number of turns, the current through them and the geometry of the coil.As magnetic field sensors U1 and U2, SS496A analog Hall sensors with a linear characteristic up to 840 gauss are used, this is the very thing for our case. When using analogs with a different sensitivity, you will need to adjust the gain by OP1 / 3, as well as check for the level of maximum induction at the ends of your coil (in our case, with a core, it reaches 500 gauss), so that the sensors do not saturate at peak load.
OP1 is LM324N quad op amp. When the coil is off, it gives out 20mV instead of zero at output 14, but this is quite acceptable. The main thing is not to forget to choose from a bunch of 100K resistors the closest in actual value for installation as R1, R2, R3, R4.
Ratings C1, R6 and R7 were selected by trial and error as the best option for stabilizing magnets of different calibers (N35H magnets D27x8, D15x5 and D12x3 were tested). The R6 / R7 ratio can be left as is, and the value of C1 can be increased to 2-5 microfarads, in case of problems.
When using very small magnets, you may not get enough gain, in which case cut the value of R8 to 500 ohms.
D1 and D2 are ordinary 1N4001 rectifier diodes, any will do here.
The common TL494CN chip is used as a pulse-width modulator U3. The operating frequency is set by the elements C2, R10 and P3 (according to the 20 kHz scheme). The optimal range is 20-30 kHz, at a lower frequency coil whistle appears. Instead of R10 and P3, you can simply put a 5.6K resistor.
T1 is an IRFZ44N field effect transistor, any other from the same series will do. When choosing other transistors, it may be necessary to install a radiator, be guided by the minimum values of the channel resistance and gate charge.
SD1 is a VS-25CTQ045 Schottky diode, here I grabbed it with a large margin, a regular high-speed diode will do, but it will probably get very hot.
LED1 yellow LED L-63YT, here, as they say, the taste and color, you can set them up a little more so that everything glows with multi-colored lights.
U4 is a 5V voltage regulator L78L05ACZ to power the sensors and the op-amp. When using an external power supply with an additional 5V output, you can do without it, but it is better to leave the capacitors.
6.Conclusion
Everything worked out as intended. The device works stably around the clock, consumes only 6W. Neither the diode, nor the coil, nor the transistor are heated. I am attaching a couple more photos and the final video:
7. Disclaimer
I am not an electronics engineer or a writer, I just decided to share my experience. Maybe something will seem too obvious to you, and something too complicated, but I forgot to mention something at all. Feel free to make constructive suggestions both for the text and for improving the diagram so that people can easily repeat this if they so desire.Magnetic levitation always looks impressive and bewitching. Such a device today can not only be bought, but also made by yourself. And in order to create such a magnetic levitation device, it is not necessary to spend a lot of money and time on it.
This material will present a diagram and instructions for assembling a magnetic levitator from inexpensive components. The assembly itself will take no more than two hours.
The idea of this device called Levitron is very simple. The electromagnetic force lifts a piece of magnetic material into the air, and in order to create a floating effect, the object rises and falls in a very small range of heights, but at a very high frequency.
To assemble a Levitron, you only need seven components, including a coil. The scheme of the magnetic levitation device is presented below.
So, as we see from the diagram, in addition to the coil, we need a field effect transistor, for example, an IRFZ44N or another similar MOSFET, a HER207 diode or something like 1n4007, 1KΩ and 330Ω resistors, an A3144 Hall sensor, and an optional indicator LED. The coil can be made independently, this will require 20 meters of wire with a diameter of 0.3-0.4 mm. To power the circuit, you can take a 5 V charger.
To make a coil, you need to take the base with the dimensions shown in the following figure. For our coil, it will be enough to wind 550 turns. Having finished winding, it is desirable to insulate the coil with some kind of electrical tape.
Now solder almost all the components except the Hall sensor and the coil on a small board. Place the Hall sensor in the hole of the coil.
Fix the coil so that it is above the surface at some distance. After that, power can be supplied to this magnetic levitation device. Take a small piece of neodymium magnet and bring it to the bottom of the coil. If everything is done correctly, then the electromagnetic force will pick it up and keep it in the air.
If this device does not work properly for you, then check the sensor. Its sensitive part, that is, the flat side with the inscriptions, should be parallel to the ground. Also, for levitation, the shape of the tablet, which is inherent in most sold neodymium magnets, is not the most successful. So that the center of gravity does not “walk”, you need to transfer it to the bottom of the magnet, attaching something not too heavy, but not too light either. For example, you can add a piece of cardboard or thick paper, as in the first image.
.
In this article, Konstantin, the How-todo workshop, will show us how to make a levitron.
So, levitron. The principle of operation of this pribluda is simple, like a self-tapping screw. We use an electromagnet to lift a piece of some magnetic material into the air. To create the effect of soaring, the electromagnet is turned on and off with a high frequency.
That is, as it were, we lift and throw a magnetic sample.
The scheme of such a device is surprisingly simple, and it is not difficult to repeat it. Here is the schematic.
We need materials and components.
LED of any color, it is not required.
Transistor IRFZ44N, almost any field worker similar in parameters will do.
Diode, here the author uses HER207, some 1N4007 will work just as well.
Resistors for 1 kOhm and 330 Ohm (the latter is optional).
Hall sensor, I have this A3144, it can also be replaced with a similar one.
Copper winding enameled wire with a diameter of 0.3 0.4 mm, 20 meters. The author has a wire of 0.36 mm.
A neodymium tablet-type magnet, 5 by 1 mm in size, is also not very important, within reason.
An unnecessary five-volt charger from the phone is suitable as a power source.
Glue, paper, soldering iron... standard soldering kit.
Let's move on to assembly. First you need to make a cardboard coil for the body of the future electromagnet.
The coil parameters are as follows:
6 mm diameter of the inner sleeve, the width of the winding layer is approximately 23 mm and the diameter of the cheeks, with a margin, is about 25 mm.
As you can see, Konstantin built a case for the reel out of cardboard and trimmed a notebook sheet, well lubricating them with superglue.
We fix the beginning of the wire in the frame, be patient and start winding about 550 turns.
The winding direction does not matter. You can even wind it in bulk, but this is not our method.
We wind 12 layers, turn to turn, isolating each layer with electrical tape.
After spending an hour and a half, we fix the end of the wire and set aside the coil.
We proceed to soldering, everything is according to the scheme, without any differences.
We extend the outputs of the Hall Sensor with wires and insulate with heat shrink, because it must be placed inside the coil.
Actually, everything, it remains only to set up, for this we install the Hall sensor inside the coil and fix it with improvised means.
We hang the coil, we supply power.
By bringing the magnet we feel that it is attracted or repelled, depending on the polarity.
At some distance, the magnet tries to hang, but does not hang for a long time.
We study the documentation for the sensor, where it is specially shown in the pictures on which side it has a sensitive zone.
We take it out and bend it so that the flat side with the inscriptions ends up parallel to the ground.
We push it back, this time everything is much better.
But still does not soar.
The problem lies in the shape of the magnet, namely the flat shape of the "tablet".
Not the best you can think of for levitation. It is enough just to shift the center of gravity down. We organize it with a piece of thick paper.
By the way, before gluing the counterweight, do not forget to first look at which side the magnet is attracted to the coil.
Own now everything more or less works, it remains only to center and fix the sensor.
Here it is told and shown how to make a cool levitron with your own hands!
I was forced to assemble this craft at the university :)
I did it in tandem with a classmate, whose task was to make a freak case, and from me - an electronic stuffing.
How cool everything turned out - judge for yourself, write comments, it will be interesting to read, discuss.
I don’t remember exactly how we came up with the idea of making a Levitron, the theme of the craft was free. The design seems to be simple, but it attracts the eye.
In general, the Levitron itself is a device that supports an object in an environment that does not come into contact with any surface in any way, except through air. It will also work in a vacuum.
In this case, the electronics makes the magnet float, and the magnet can already be glued to, for example, a can of a delicious inexpensive drink :)
If you search carefully on the Internet, you can see many different options for an electromagnetic levitron, for example:
They can be divided into suspended and repulsive. If in the first case it is necessary to simply compensate for the force of gravity, then in the second it is also a displacement in the horizontal plane, since according to Earnshaw's theorem "any equilibrium configuration of point charges is unstable if nothing acts on them except for the Coulomb forces of attraction and repulsion." - quote from wiki.
It follows from this that the suspended levitron is easier to manufacture and configure, if at all necessary. I didn’t want to bother much, so they made a hanging levitron for the university, which is discussed here, and the repulsive one has already made a loved one for myself :) It will be written about in another article. A little later I will delete this text and give a link to it here. It works great, but it also has its drawbacks.
In turn, all suspended levitrons can also be conditionally divided into digital and analog according to the method of holding an object at the same distance. And according to the type of sensors, they can be divided into optical, electromagnetic, sound, and, probably, everything.
That is, we receive an analog signal about the distance of the magnet to the Levitron, and we correct the force of influence on the magnet already in a digital way. Hi-tech, however.
The idea itself was borrowed from the geektimes website, and the printed circuit board was already made personally for our set of parts. Also in the original project, three-pin SS49 sensors were used, but the deadlines were very tight, they were, to put it mildly, unreasonably expensive for us ($4 per piece versus $6 for 10 pieces in China - link for example), so we used four-pin Hall sensors. I had to change the scheme and make structural additions to the device. Also, for greater showiness, a block of LEDs was added, which light up smoothly when the magnet is brought up, that is, when the Levitron starts to work and smoothly turn off when the magnet is removed. All this will be reflected in the diagram.
Actually, the Levitron circuit on four-pin sensors:
And the Levitron circuit on three-pin sensors and a simpler backlight:
The principle of operation is quite simple. The coil, which is an electromagnet, attracts a magnet when energized - the object is attracted. A sensor attached between the magnet and the coil detects an increase in magnetic flux, which means the magnet is approaching. Electronics monitors this and disconnects the coil from the voltage source. The magnet starts to fall under the force of gravity. The sensor detects a decrease in the magnetic flux, which is immediately detected by the electronics and voltage is applied to the electromagnet, the magnet is attracted - and this happens very often - about 100 thousand times per second. There is a dynamic balance. The human eye fails to notice this. The oscillator frequency is set by a resistor and a capacitor at pins 5 and 6 of the TL494 chip.
The second sensor on the other side of the electromagnet is needed in order to compensate for the magnetic field created by the coil itself. That is, if this second sensor did not exist, when the electromagnet was turned on, the system would not be able to distinguish the intensity of the magnetic field of the neodymium magnet from the magnetic field created by the electromagnet itself.
So, we have a system of two sensors, the signal from which is fed to the operational amplifier in a differential connection. This means that only the voltage difference received from the sensors appears at the output of the operational amplifier.
For example. One of the sensors has an output voltage of 2.5 V, and the other one has a voltage of 2.6 V. The output will be 0.1 V. This differential signal is located at pin 14 of the LM324 chip according to the diagram.
Further, this signal is fed to the next two operational amplifiers - OP1.1, OP 1.3, the output signals of which go through the diode valve to the 4th output of the TL494 microcircuit. A diode valve on diodes D1, D2 passes only one of the voltages - the one that will be higher at face value. Conclusion No. 4 of the PWM controller drives as follows - the higher the voltage at this pin, the lower the duty cycle of the pulses. Resistor R9 is designed so that in a situation where the voltage at the inputs of the diode valve is less than 0.6 V - pin No. 4 is unambiguously pulled to the ground - while the PWM will produce the maximum duty cycle.
Let's return to the operational amplifiers OP1.1, OP 1.3. The first serves to turn off the PWM controller while the magnet is at a sufficiently large distance from the sensor so that the coil does not work at maximum idle.
Using OP 1.3, we set the gain of the differential signal - in fact, it sets the feedback depth (OS). The stronger the feedback, the stronger the system will react to the approach of the magnet. If the OS depth is not sufficient, the magnet can be brought close, and the device will not start to reduce the power pumped into the electromagnet. And if the depth of the OS is too large, then the duty cycle will begin to fall before the force of attraction of the magnet can hold it at this distance.
It is not necessary to set a variable resistor P3 - it serves to adjust the frequency of the generator.
OP1.2 is a 2.5 V voltage generator required for four-pin sensors. It is not needed for SS49 3-pin sensors.
I forgot to mention the elements C1, R6 and R7. Their trick is that the constant signal here is cut down by 10 times due to resistors, and the variable signal quietly passes further due to the capacitor, thereby achieving the emphasis of the circuit on sharp changes in the distance of the magnet to the sensor.
Diode SD1 is designed to dampen reverse surges at the moment the voltage is turned off on the electromagnet.
The node on T2 allows you to smoothly turn on and off the LED line when pulses appear on the electromagnet.
Let's move on to the design.
One of the key points in the Levitron is an electromagnet. We made a frame based on some kind of construction bolt, on which round sides were cut out of plywood.
The magnetic flux here depends on several key factors:
- the presence of a core;
- coil geometry;
- coil current
To put it simply, the larger the coil and the more current flows in it, the stronger it attracts magnetic materials.
PEL wire 0.8 mm was used as a winding. They wound by eye until the dimensions of the coil seemed impressive. It turned out the following:
It may not be possible to find the necessary wire in our area, but it is quite easy to find it in online stores - 0.4 mm wire for winding the coil.
In the meantime, the coil was wound and the board was prepared and etched. It was made using LUT technology, the board drawing was made in the Sprint LayOut program. You can download the Levitron board from the link.
The board was etched in ammonium persulfate residues, an empty can of which was successfully used later in this project :)
I want to note that the placement of parts, as well as the wiring of the tracks, imply very accurate soldering, since it is easy to make connections where they should not be. If there are no such skills, it is quite permissible to do this with large-sized components on a breadboard, like this, and make connections using wires on the reverse side.
As a result, the payment turned out like this:
The board very ergonomically fit into the dimensions of the coil and was attached directly to it with the help of powerful hot melt adhesive, thereby turning into a single monoblock - plugged in the power, set it up and the system worked.
But this was all before the electromagnet was ready. The board was made a little earlier, and in order to somehow test the performance of the device, a smaller coil was temporarily connected. The first result pleased.
The sensors, as already mentioned above, are used from the position tracking systems of BLDC engines, four-pin. Since it was not possible to find documentation on them, I had to empirically find out which conclusions are responsible for what. The form factor looks like this:
Meanwhile, a large electromagnet arrived in time. This one gave me a lot of hope :)
The first tests with a large electromagnet showed a rather large working distance. There is one caveat here - the sensor, which is located on the side of the neodymium magnet, should be a little further from the coil for reliable operation of the electronics.
The last photo is more like a space satellite. By the way, this is how this Levitron could be designed. And for those who intend to repeat the design - everything is ahead :)
As a levitating object, it was decided to use a can of soft drink. We sculpt a magnet to the bank on double-sided tape, check.
Works great, in general, the device can be considered ready. All that's left is the exterior. A support beam was made from bars and sticks, the body of our monoblock was made from the same empty plastic can from ammonium persulfate. Only two wires for power come out of the monoblock, as intended.
By this time, the circuit for smooth switching on the LED line was already soldered by surface mounting, the line itself was successfully mounted on the ubiquitous hot melt adhesive.
A block borrowed from some printer, converted from 42 V to 12 V, acts as a power supply.
I will also show the appearance of the power supply :)
Next, a stand was made of plywood, in which a power supply and a connector for connecting 220 V were placed. A cloth fabric was glued on top for beauty, the whole structure was painted yellow-black. The jar was changed, as during the experiments it was a little wrinkled.
From all this, in addition to the effect of levitation, it turned out to be a very wonderful night light.
I will add the video a little later, but for now, to top it all off, I want to say that my design was easily repeated by a 13-year-old student of my radio club.
So far, the appearance has not been brought to the finished version, but the electronic filling works as expected. Photo of his design:
A short video about what the made levitron is like:
www.youtube.com/watch?feature=player_embedded&v=vypjmqq9...
If someone is not afraid to do the same interesting thing, then here is a detailed instruction for you:
A bit of theory
Let's start, perhaps, with the mechanical scheme of the platform levitron, which has developed in my understanding. The magnet that hovers above the platform, I will here for brevity call the word "chip".
Levitron platform sketch(top) is shown in Fig. one.
On fig. 2 - power diagram of a vertical section along the central axis of the platform (as I imagine it) at rest and without current in the coils. All is well, except that the state of rest in such a system is unstable. The chip tends to move away from the vertical axis of the system and slam onto one of the magnets with force. When "feeling" the space above the magnets with a chip, a force "hump" is felt above the center of the platform with the top lying on the central axis.
mg - chip weight,
F1 and F2 - the forces of interaction of the chip with the magnets of the platform,
Fmag - the total impact that balances the weight of the chip,
DH - Hall sensors.
On fig. 3. shows the interaction of the chip with the coils (again, in my opinion), and the rest of the forces are omitted.
Figure 3 shows that the purpose of coil control is to create a horizontal force Fss, always directed towards the equilibrium axis when a displacement occurs. X. To do this, it is enough to turn on the coils so that the same current in them creates a magnetic field in the opposite direction. There was nothing left: measure the offset of the chip from the axis (the value X) and determine the direction of this displacement using Hall sensors, and then pass currents in the coils of a suitable strength.
A simple repetition of electronic circuits is not in our traditions, especially since:
- two TDA2030A are not available, but there is TDA1552Q;
- there are no SS496 Hall sensors (available for about $2 each), but there are sensors similar to HW101, 3 pcs for free in each CD or DVD drive drive;
- too lazy to mess with bipolar power.
Datasheets:
SS496 - http://sccatalog.honeywell.com/pdbdownload/images/ss496.seri...HW101- http://www.alldatasheet.com/datasheet-pdf/pdf/143838/ETC1/HW101A.html
The circuit consists of two identical amplifying channels with differential inputs and bridged outputs. On fig. 4 shows the complete diagram of only one amplification channel. The chips used were LM358 (http://www.ti.com/lit/ds/symlink/lm158-n.pdf) and TDA1552Q (http://www.nxp.com/documents/data_sheet/TDA1552Q_CNV.pdf).
A pair of Hall sensors is connected to the input of each channel so as to supply a differential signal to the amplifier. The outputs of the sensors are switched on in opposite directions. This means that when a pair of sensors is in a magnetic field with the same intensity, zero differential voltage is supplied from it to the input of the amplifier.
Balancing resistors R10 are multi-turn, old, Soviet ones.
In an attempt to squeeze a sufficiently high gain out of the amplifier, I got a banal self-excitation, presumably due to a mess on the circuit board. Instead of “cleaning”, frequency-dependent RC chains R15C2 are introduced into the circuit; they are not required. If you still had to install them, then the resistance R15 must be selected as the largest, at which self-excitation goes out.
The power supply of the entire device is an adapter (pulse) for 12V 1.2A, reconfigured to 15V. Power consumption in the normal state (with the fan turned off) turned out to be quite modest: 210-220 mA.
Design
A 3.5” floppy drive shroud was chosen as the case, which roughly corresponds to the dimensions of the prototypes. To level the platform
legs are made of M3 screws.
A figured hole is cut out in the upper part of the body, clearly visible in Fig. 5. Subsequently, it is closed with a decorative mirror plate made of chrome-plated brass, fixed with screws from hard drives.
1 - installation locations for magnets (bottom) and balance indicators (optional)
2 - "pole pieces" of coils
3 - Hall sensors
4 - backlight LEDs (optional)
The Hall sensors are located in the holes of the fiberglass base of the platform and are soldered on the unbent legs of the connectors (I don’t know the type). The connectors looked like in Fig.6.
The sensors are soldered from the motors of the CD or DVD drive. There they are located under the edge of the rotor and are clearly visible in Fig. 7. For one channel, you need to take a pair of sensors from one engine - so they will be the most identical. Soldered sensors - in Fig.8.
Plastic spools for sewing machines were bought for the spools, but there was little space for winding on them. Then the cheeks were cut off from the spools and glued onto segments of a thin-walled brass tube with an outer diameter of 6 mm and a length of 14 mm. The tube used to be a segment of a telescopic rod antenna. On four such frames with a 0.3 mm wire, the windings are wound “almost in layers” (without fanaticism!) Until they are filled. The resistance is aligned to 13 ohms.
Magnets - rectangular 20x10x5 mm and disk magnets with a diameter of 25 and 30 mm 4 mm thick (Fig. 9) - I still had to buy ... Rectangular magnets are installed under the base of the platform, and chips are made from disk magnets.
View of the device from below and behind (upside down) - in fig. 10 and 11 (one legend for both figures). The mess, of course, is picturesque ...
The U2 TDA1552Q chip (3) is located on the heat sink (9), which used to work on the video card. The radiator itself is fixed with screws on the bent parts of the top case cover. A power socket (1), control sockets (2) and a thermal control unit (5) are also fixed to the radiator (9).
A piece of fiberglass, which used to be a keyboard, serves as the base of the platform. Coils (7) are fixed on the base with M4 screws and nuts. Magnets (6) are fixed on it with the help of clamps and self-tapping screws.
The control jacks (2) are made from a computer power connector and are fixed on the back of the unit near the balancing resistors (10) so that they are easily accessible without disassembly. The jacks are connected, of course, to the outputs of both channels of the amplifier.
The circuit of the preamplifier and its power stabilizer, including balancing resistors (10), is mounted on a breadboard and, as a result of adjustment, turned into a picturesque pigsty, which had to be refrained from macro photography.
1 - fastening the power socket
2 - control sockets
3 - TDA1552Q
4 - power switch
5 - thermal control unit
6 - magnets under the clamps
7 - coils
8 - magnetic shunts
9 - heat sink
10 - balancing resistors
Adjustment
Setting zeros at the outputs of both channels at each debugging start is mandatory. It is possible without fanaticism: + -20 mV is quite acceptable accuracy. There may be some interference between the channels, so with a significant initial deviation (more than 1-1.5 volts at the channel output), it is better to set zeros twice. It is worth remembering that with an iron case, the balance of a disassembled and assembled device is two big differences.
Checking channel phasing
The chip must be taken in hand and placed above the center of the platform of the included Levitron at a height of approximately 10-12mm. Channels are checked one by one and separately. When the chip is shifted by hand along the line connecting the sensors opposite from the center, the hand should feel a noticeable resistance created by the magnetic field of the coils. If no resistance is felt, and the hand with the chip "blows" away from the axis, you need to swap the wires from the output of the channel under test.
Adjusting the position of the floating chip
On videos about homemade platform levitrons, you can often see that the chip is hovering in an inclined position, even if it is made on the basis of disk magnets, that is, it is quite well symmetrical. Not without distortion in the described design. Perhaps the metal case is to blame ...
First thought: move the magnets down from the side where the chip is unnecessarily “supported”.
The second thought: move the magnets further from the center from the side where the chip is unnecessarily “supported”.
The third thought: if the magnets are displaced, then the magnetic axis of the system of permanent magnets of the platform will be skewed relative to the magnetic axis of the coil system, due to which the behavior of the chip will become unpredictable (especially with different weights).
The fourth thought - to make the magnets stronger on the side where the chip is tilted - was discarded as unrealizable, because there was nowhere to get a wide range of magnets to fit.
The fifth idea: to make the magnets weaker on the side where the chip is unnecessarily “supported” – turned out to be successful. Moreover, it is quite simple to implement. A magnet, as a source of a magnetic field, can be shunted, that is, a part of the magnetic flux is short-circuited, so that the magnetic field in the surrounding space becomes a little weaker. As magnetic shunts, small ferrite rings (10x6x3, 8x4x2, etc.) were used, plucked free of charge from dead economy lamps (8 in Fig. 10). These rings just need to be magnetized to a too strong magnet (or two or three) on the side that is farther from the center of the platform. It turned out that by choosing the number and size of shunts for each "too strong" magnet, it is possible to quite accurately level the position of a floating symmetrical chip. Don't forget to perform electrical balancing after every change in the magnetic system!
Options
Options include: amplifier unbalance indicators, thermal control unit, lighting, and adjustable platform legs.
The amplifier unbalance indicators are two pairs of LEDs located at the same radii as the sensors, in the thickness of the fiberglass base of the platform (1 in Fig. 5). LEDs, very small and flat, used to work in some kind of modem, but they will also work from an old mobile phone (in SMD version). The LEDs are recessed in the holes, since the chip, breaking off from the center, flops onto the nearest magnet and is quite capable of destroying the LED.
The indicator scheme for one channel is shown in fig. 12. LEDs must be with an operating voltage of 1.1-1.2 V, i.e. simple red, orange, yellow. At higher LED voltages (2.9-3.3 V for super-bright ones), the number of diodes in the D3-D6 chain should be recalculated to minimize the "dead zone" - the minimum voltage at the channel output, at which none of the LEDs glows.
I arranged the indicators so that the one towards which the chip is shifted from the center shone. Indicators help to easily hang a chip over the Levitron, as well as to level the platform. In the normal state, they are all redeemed.
The diagram of the thermal control unit is in fig. 13. Its purpose is to prevent the final amplifier from overheating. At the output of the thermal unit, a fan 50x50 mm 12V 0.13A from the computer is turned on.
In the thermal node circuit, it is easy to recognize a slightly modified Schmitt trigger. Instead of the first transistor, a TL431 chip was used. The type of transistor Q1 is indicated conditionally - I stuck the first NPN that came across that could withstand the operating current of the fan. A thermistor found on an old motherboard in a processor socket was used as a temperature sensor. The temperature sensor is glued to the heatsink of the final amplifier. By selecting the resistor R1, you can adjust the thermal unit for operation at a temperature of 50-60C. Resistor R5, together with the collector current Q1, determines the amount of hysteresis in the circuit relative to the voltage at the control input U1.
In the diagram in fig. 13 resistor R7 is introduced to reduce the voltage on the fan and, accordingly, the noise from it.
On fig. 14 shows how the fan is embedded in the bottom cover of the case.
Another way to use a thermal node is to connect a final amplifier chip to the MUTE control pin (Fig. 15). The value of the R5 rating indicated on the diagram assumes that MUTE (pin 11 of the U2 chip in Fig. 4) is connected to the power supply through a 1kΩ resistor (NOT directly, as in the datasheet!). In this case, a fan is not needed. True, when the MUTE signal is applied to the amplifier, the chip falls, and after the MUTE signal is removed, it itself (for some reason?) does not take off.
Illumination - 4 bright LEDs with a diameter of 3 mm, located obliquely to the center in the holes of the platform base and decorative plate in those places where the chip does not fall. They are connected in series and through a 150 Ohm resistor to the 15V device's general power supply circuit.
Conclusion
load capacity
In order to “finish off” the topic, the “cargo characteristics” of the Levitron with chips of 25 and 30 mm in diameter were removed. I here called the cargo characteristics the dependence of the height of the hovering of the chip above the platform (from the decorative plate) on the total weight of the chip.
For a chip with a 25 mm magnet and a total weight of 19 g, the maximum height was 16 mm, and the minimum was 8 mm with a weight of 38 g. Between these points, the characteristic is almost linear. For a chip with a 30 mm magnet, the load characteristic turned out to be between the points of 16 mm at 24g and 8 mm at 48g.
From a height below 8 mm from the platform, the chip falls, being attracted to the iron cores of the coils.
DO NOT do like me!
First, do not save on sensors. "Naked" Hall sensors, taken out in pairs for each channel of two engines (that is, almost the same!) - still show their ugly large temperature coefficient of resistance. Even with the same power circuits and back-to-back switching of the sensor outputs, you can get a noticeable zero shift at the channel output when the temperature changes. Integrated sensors SS496 (SS495) have not only a built-in amplifier, but also thermal stabilization. The internal amplifier of the sensors will make it possible to significantly increase the overall gain of the channels, and the circuit for their power supply is simpler.
Secondly, one should, if possible, refrain from placing the Levitron in an iron case.
Thirdly, bipolar power is still preferable, because gain control and zero adjustment are easier.
Thank you for your attention!