"Son, i have two wall clocks, one is nice but it makes funny noise, and the other is ugly but works good. I want you to put the good mechanism to the nice clock and throw away the other".
That is what my mother told me a few hours ago, and that is what i did, but i didn't throw away any the other parts... I used it to feed my curiosity once more. That is how this experiment began.
The original purpose of this experiment...
The good, the bad and the ugly
The original purpose of this experiment was to see how the clock is working. Recently, i had a discussion with a friend of mine. He said that there must be a step motor inside with 60 steps per rotation. Rather impossible for such a low cost mechanism. It was just about time to find it out.
The change (for my mother) was done in a hurry, as more important things were yet to come. Just for the records, the ugly clock had it's mechanism (the good one) fixed with clips, while the beautiful clock had the noisy mechanism fixed with a nut and a drop of hotmelt glue. I removed the good mechanism and used only hotmelt to fix it on the beautiful clock housing. End with this.
And now for the entrails. The mechanism was opened by carefully releasing 3 clips. Everything used the upper housing as the second base, thus when i removed the upper housing, everything was free to be removed. And i removed everything. The "motor" mechanism was then revealed to me.
The housing opened by carefully releasing 3 clips
Everything was free to be removed
This is the "motor" mechanism
I like to move it move it!
So i had the motor in my hand. A coil is placed on one side of a U metal. The edges of this U metal surrounds a round magnet. The first gear is fixed to this magnet:
So now, i only had to find out what kind of signal was powering this coil. Well, actually i was pretty sure by this time that the coil would be powered with alternating pulses, so that the North and South poles would change, causing the magnet to rotate. Standard AC single pole motor. But for the sake of the experiment (and for calming my curiosity) i just had to do it. Here are the results:
Positive and negative rectangular pulses, 16 pulses per second. I was rather surprised by the frequency of the signal. I mean, 2 Hz is what i expected, not 8 Hz! The seconds pointer should not rotate like the normal clocks does, 2 clicks per second. But due to my rush, i did not see the clocks in operation before the surgery. So, either i would have to "imagine" the rotation of the seconds, or i had to re-assemble the entrails and see it in action. You guessed correct, i re-assembled it. Moreover, i soldered 2 wires for the power supply. The following video shows exactly that this clock had a smoother rotation than the normal clocks:
By the time i saw the pointer moving, i remembered that i had seen before similar clocks rotating that way. So, finally, how this Chinese clock works? A permanent magnet rotates inside an alternating magnetic field. The frequency of this alternating field determines the speed of the rotation. The rest is done with mechanical gears. If i was forced to name this motor, i would not call it step-motor. The principle of operation is the same as a simple AC single phase motor, only that it runs with a very low frequency.
What i could not understand, is how this motor rotates always to the right direction. A single phase AC motor is coin-toss whether it will rotate clock wise or counter clock wise. There are some methods to determine the rotation, but this would require a secondary coil, something that does not exist here. So??? How???
How the wall clock motor works anyway???
NOT a toroidal magnet, although it looks like one
It took me a while to understand what happens. First of all, the magnet. Although it looks like a toroid magnet, it is not. And it could not be a toroid magnet, because such a magnet has no polarization, no North or South pole, only toroidal magnetic lines. But to make sure that my guess is right, i did a very simple test. Holding the magnet, i tried with a metal the magnetic field in some places. I found out that there are 2 places, one across the other, that the magnetic power is stronger. This is the proof that this is not a toroidal magnet. I marked with blue pen these points.
But this does not solve the mystery of the correct rotation. Taking a closer look, right below the U metal of the coil, there is another almost rectangular piece of metal. The magnet goes through the center of this metal. A closer look at this piece, reveals an not-normal corner. This corner is bigger than the other 3. And frankly, this is not accidental. Look at the following video:
I have completely removed the coil. Only this piece of metal is there. I turn the magnet a little bit, and then i let it free. You will notice that no matter where i leave it, one of the 2 points where the poles are (the small blue lines on the gear of the magnet) will always go there where this big corner is! Brilliant! So, what happens is now simple to understand. The magnetic field of the coil has an angle of about 35-40° from this corner. When a pulse arrives, the magnet poles are polarized accordingly. This pulse lasts for a while and then no current goes through the coil at all. When this happens, no magnetic field is acting on the magnet. Thus, the magnet will rotate so that the closest pole (North or South) will be lined up with the bigger corner, because this corner acts more to the magnetic field of the magnet. This will rotate the magnet 35-45 degrees MORE from it's last position. This rotation is the key! When the magnetic field from the coil changes direction, the magnet is already rotated a little bit MORE to the correct direction, thus it is impossible to rotate the other way! Simple, easy and absolutely ingenious!
Do you see the small rectangular piece of metal?
Take a closer look. It has one corner bigger than the others.
No matter where you release the magnet, it's closest pole (marked with blue line) will always line up with the bigger corner
What else can we do with that?
The previous experiment just gave me an idea. I suppose that the crystal that provides the pulses to the clock circuit is the classic 32768 Hz crystal. How can i tell? Well, this crystal is widely used for clock applications because it can be divided by a power of 2 and provide an integer number of frequency, and it is cheap. But, as before, for the sake of the experiment, i had to see it on the oscilloscope.
The probe connected parallel to the crystal
And the result on the monitor...
The fun circuit
The mechanism of the clock, a little bit "changed" to fit my needs
Hmmmmm, 34.2 KHz. I suppose i was wrong. It looks like that not everyone uses the good old 32768 crystal. Anyway, the circuit that this crystal is connected to will do the divisions. The output is driven to the coil. I remove the coil and i put a circuit that i made, a simple transistor amplifier. But i did not drive both the positive and negative pulses. Instead, i had a diode to cut off the negative edges. I drove the output of the transistor to another transistor that performed a simple inverter, and the output of the inverter powered an LED. Voila!
The camera i used is an old Canon and not a video recorder. This video shows the LED to flash slower than 8 Hz, but this is only due to the camera's sampling rate. It flashes much faster. Anyway, this is a nice and easy way to get a precision oscillator out of a... wall clock!
Here is the circuit. LSP1 and LSP2 are the inputs where the clock circuit is connected to the coil. The coil needs to be removed. If not, funny things may happen, or the transistor will be destroyed due to the high voltage of the coil when the current shuts off. The Q2 and R2 can be totally omitted without any problem at all. I just put it there, because the ON-time of the LED was more than the OFF time, and the flash was not visible to the camera. Use only 1 transistor to get the pulses. Use these pulses for any kind of circuit that requires such a frequency. Have fun.
The crystal oscillator is likely running at your original frequency of 32768Hz. The scope probe loads the crystal circuit altering the frequency. The I.C. oscillator that the crystal is connected to has an amplifier input and output. It's best to probe the amplifier output since it has a much lower impedance. The frequency will still be altered, but not as much.
@saideep Yes. Do not confuse the transistor inverter with the electric motor inverter. A transistor inverter (amplifier) is when you provide 1 and you get 0 (and opposite). It simply inverts the signal.
very nice and informative post
and i have a question "what does the sentence mean the output of on transistor given to other transistor performs as simple inverter".do u mean that entire circuit acts as inverter. If yes please tell me how?
I like to experiment with mechanical pendulum clocks. I placed a small permenent magnet at the bottom of a pendulum so it swings through a stationary coil. The coil voltage pulse as the magnet passes is sensed by a PIC ADC input that is quickly reconfigured as an output so it pulses the pendulum by repelling the permenent magnet at the same time during each half swing, then it's reconfigured again as an ADC input in preparation for the return swing. I pulse a different PIC output pin with alternating pulse polarities to drive a clock motor similar to this at 0.5 Hz. (1 tick per second...the pendulum is almost exactly a meter long). I interfaced two pushbuttons to the PIC as well. When one is pushed and held, it inhibits pulsing the clock motor until it's released. When the other is pushed it speeds up the clock to 10 Hz (20x normal). These allow easily synchronizing it to WWV.
The pendulum rod by the way is an old 4 ft. flourescent tube with the upper end carefully opened and the phosphor washed out with a long bottle brush (outdoors with plenty of water...the phosphor is slightly toxic and there are a few micrograms of mercury present too). I adjusted it's period by adding or removing dry sand from it to set it precisely to 0.5 Hz (two PIC pulses per full swing so the clock motor gets driven 180 degrees per second). I used the glass tube because it has a very low thermal coefficient of expansion so the pendulum's timing is less effected by temperature changes. It makes a rather precise, old timey pendulum clock with high tech electronics.
Presently building another pendulum with a 10 second period. Actually it's more like a slow teeter totter seesaw with the pivot a couple of millimeters above the center of the horizontal pendulum rod that's balanced with weights on both ends. It's PIC will drive a wall clock that runs at 10Hz for one second every 10 seconds. The second hand will therefore stop for 9 seconds, then advance 10 seconds in only a second. Should be an interesting conversation piece.
Also used the PIC to measure time of each half swing of the pendulum in yet another experiment. The PIC detects excursions outside normal half swing times and remembers the time it occurs. This became a somewhat crude seismic sensor. There is a quarry about two miles from my home and the pendulum detects the anomalous half swing times every working day at about 10:15 to 10:30 AM when the first explosives are used at the quarry. The pendulum is more sensitive to the explosions when swinging toward and away from the quarry rather than at right angles to it. Typically further explosions occur at 2 hour intervals for the rest of the day. These explosions are too weak to feel, even when I expect them. I had some fun speaking with the foreman at the quarry about my "seismic sensor". He found it intriguing. I guess several of these pendulums swinging in different directions could determine the direction of an earthquake. Several groups like these could triangulate the epicenter.
The 32768 Hz circuit I removed from the clock that pulses once per second is driving a PIC that drives a 24 hr. 6 digit LED clock.
I need an output of 50hz-100hz driving that clock movement stepper motor. The wave form should look simmilar to your posted osccilloscope image. What kind of circuit can I use making that type of pulse pattern ( 1.5v -1.5v)?
Jim, you did not use the first transistor??? I am surprised! What voltage do you get from it? Can it really sink a PIC's input? I did this a lot time ago, so i do not really remember what voltage i did get, but to add a transistor means that i had problems driving anything with it. The second transistor is only to drive the LED.
Yes firoz that is exactly the reason. The internal circuit of the clock has more divisions so the final is 1 pulse per 2 sec (0.5 Hz), which is rather strange. I suppose that the "seconds" pointer of your clock would make one move per 2 seconds is that correct? do you remember? Is it possible to reassemble and check? It may also be that your clock does not have a seconds pointer at all.
Another more accurate method is to take the 15625khz line frequency supplied by your rubidium locked local TV station, divide it a few times, add a PLL with Crystal and you have a Lab Frequency standard as well as a very accurate clock.
Excellent detailing and presentation of the step by step exploration of the circuit's working. Really appreciate that.
I was looking for this detail and this really helps.
You have made me realize how losing our curiosity, we also lose the chance to explore and learn from such ingenious designs in seemingly the most simple, most common articles we daily use and see around us.