I am in the process of designing a metal fatigue demo setup. For this purpose I need a heavy pendulum, swinging at quite large angles (say 30 degrees to either side). In order to keep the pendulum swinging permanently some excitation circuit will be needed to compensate for bearing/aerodynamic drag.
Being a mechanical engineer I have only faint ideas how to do this electronically. My simple idea is to use a coil with a soft iron segment attached to the pendulum. To see what I mean watch at:
The excitation could be accomplished by adjusting the current strength (or current pulse length) through the coil, controlled by the length of the light pulse on the light sensitive element. My question is: does this 'simple-mind' excitation mechanism make sense? If so, how could I accomplish it electronically? If not, what would be a sensible alternative?
Google Foucault or Foucault's Pendulum. you need a little delay circuit between the sensor and the actuator, and the system transfer function must meet the +360' phase shift criterion (Barkhausen) for oscillation.
OK, light sensor is a photodiode or phototransistor at the target end, so its enabled once the arm hits, making it self starting.
That trips a opamp comparator so you have a adjustable threshold for the switching. The comparator output triggers a timer in monostable mode, which means its output changes state for a time N, giving you maximum time on the magnet to ensure full travel, then the monostable drops to its normal state and the excitation cuts off, the hammer falls and the cycle repeats. A second monostable may be added to provide a delay, to ensure full energy is transferred if you have a really fast, powerful magnet.
Or just a simple 555 in astable mode (oscillator) and ignore the need for a sensor and delay.
In the upper part of the pendulum shaft, use a core of soft iron. Wind a coil around the core. Make the bobbin big enough so that the core will not hit the bobbin as core moves back and forth. Use a capacitive detector to detect when the core approaches the bobbing. Excite the coil with DC. The resultant magnetic field causes the core to move toward the center of the coil. Use the capacitive effect to also shut off the coil after the core moves away from the bobbin. Adjust the turn-off point so that just enough impetus is given to the core to overcome all drag.
If you make it so that the sensor detects the pedulum passing a point in one direction only, the solution will be easier.
You want to apply a force that always speeds the pendulum up. Doing this mechanically and optically is not super easy. Adding just a little electronics and a second optical sensor, will help a lot.
The first optical sensor will be just to the left of the bottom of the swing and the second will be just to the right. Passing the first, arms the circuit and passing the second triggers it. The circuit I am thinking of right now has a couple of LM555s in it but other designs will work too.
The first LM555 is triggered by the light at the first opto being blocked.
The second is triggered only if the first has its Q high and the light at the second opto is blocked.
The second's Q goes back to the THR of the first to clear it.
The second is a one shot that sets how long the pulse on the coil is.
Yes, MooseFET, that seems like a good idea! If the magnet/coil is placed just below the pendulum it would not force the pendulum out of the plane of swing. It will however need a measure to prevent the swing from going astray.
Since you're not an electronics guy, I'd suggest using all commercial parts: an optical interrupter, a solenoid, and a solid-state relay, possibly a time-delay relay to allow adjustment of the added impulse.
Do you expect this to be a high-Q system? If so, a small solenoid and a short kick will keep it going.
Perhaps a 'sensible alternative' would be to not attempt to re-invent the wheel?
What you have described is the Hipp clock movement designed in the mid
19th century. Stripping out refernces to how the clock movement itself was driven, it worked like this:
'The pendulum was not impulsed on every swing or even at a set interval of time. It was only impulsed when its arc of swing had decayed below a certain level. The pendulum carried a small vane, pivoted at the top, which was completely free to swing. It was placed so that it dragged across a triangular polished block with a vee-groove in the top of it. When the arc of swing of the pendulum was large enough, the vane crossed the groove and swung free on the other side. If the arc was too small then the vane never left the far side of the groove and, when the pendulum swung back it pushed the block strongly downwards. The block carried a contact which completed the circuit to the electromagnet which impulsed the pendulum. The pendulum was only impulsed as it required it.'
Another description and an excellent animation can be found here:
I have seen two different ways to transfer energy to the pendulum: one is to do it as in the reference you linked to with a solenoid and core to one side and the other is to have an electromagnet magnet mounted vertically beneath the pendulum. (And possibly displaced to one side?) The pendulum would then have a soft iron plate mounted on the bottom, leaving the pendulum bob adjustable for time keeping but in you case it could be the pendulum bob itself.
I did also come across a link to an article about an electronically driven Hipp clock but I haven't studied it as the original concept is so simple (and you still have to sense what the pendulum is doing!)
My first thought was a big electromagnet under the middle of the pendulum's arc; but then I thought of pendulum clocks, with an escapement. This doesn't necessarily need the wheel - it could be replaced with a solenoid near the suspension point, and timed with an LED/PHD or so.
You may wish to consider an alternative to a pendulum for inducing fatigue. Consider a setup where the specimen is horizontal with one end fixed to a motor shaft and the other end with a weight attached. The weight causes a bending moment producing tension in the top side of the specimen and compression in the underside. As the motor rotates, the compression and tension rotates around the specimen. These are called 'Wohler' fatigue testing machines. Here's a better description:
I wonder if a permanent magnet will give you more 'bang' for each amp in the coil than soft iron? Then make sure things are phased such that energy is only added to the system.... How does your light sensor know which way the bob is swinging?
There are some simple pendulum drives that use a magnet attached to the pendulum and two coils. A sense coil detects when the magnet approaches and triggers a pulse through the drive coil which gives the pendulum a little extra kick each time it pass.
I have a table top clock with a quartz movement and a phony pendulum. The pendulum doesn't establish a rate. It just rocks back and forth to give the appearance of a mechanical movement. It has been running for something like 5 years on one AA battery (both the movement and the pendulum).
The setup in your link will probably work as well.
One problem: Most of these 'clock pendulums' are designed to have very low losses. They use knife edge, or very expensive low friction mounts. As a result, the low efficiency drives don't need to add much energy for each swing. As you appear to be flexing some material as a test of fatigue, the losses will be higher and the energy transfer will have to be higher as well. None of these pendulum designs appear to be well suited to this kind of transfer.
I'd think about a simple motor drive and crank to bend the sample back and forth.
Paul Hovnanian mailto:Paul@Hovnanian.com
The interesting question is whether you are suspending your pendulum from a wire or a leaf spring.
If you are hanging your pendulum from a wire, a soft iron bob that you can attract with an electromagnet is about the most complicated device you can use; a coil sitting directly under the suspension point is simple, but can't start the prendulum swinging. A coil off to one side can start it up and keep it going at any amplitude you care to build up. You will need some way to work out where the bob is to time the current pulse into your solenoid; a light source at the suspension point and a photo-detector under the track of the bob - probably directly under the suspension point - would do the trick. You might want to modulate the light source and demodulate the sensor output to eliminate any direct effect of ambient light.
If you suspend your pendulum with a leaf spring, you can do better; you can put a permanent magnet into the bob with its north-south axis aligned with the direction of swing. You can drive this with a solenoid - again aligned along the direction of swing, - set directly under the rest positon of the pendulum. This coil can then both push and pull the pendulum. The permanent magnet lets you use sensitive linear Hall effect magnetic field sensor to keep track of the postion of the pendulum bob. Two - as before - would make life easier.
A serious system would use a linear stepping motor system, You'd need a double leaf (parallelogram) spring suspension and your soft iron bob would have a series of grooves cut into its bottom face, with the groove spacing roughly comparable with the clearance between the bob and the linear motor beneath it. For further details search on the Sawyer motor.
If the pendulum swings at 90 degrees to the intended direction, we need to not add energy to that swing.
Instead of one coil place two coils along the line at right angles to the intended swing. One slightly to each side of the intended path. If the pendulum swings towards either, it will be repelled and lose energy from that direction of motion because it is moving towards something that is repelling it. When it moves in the intended direction it will be moving away from the repelling magnet.
OK Here 's a little more background information about this thing. I work at the Loads and Fatigue department of our company. We tell our customers how good we are in predicting crack growth in metals. To have an appealing demo of this I want to make a "crack calendar": a simple to understand device to place in our hallway (the heavy pendulum) that uses the forces at the suspension point to crack a thin aluminum plate specimen. This specimen is instrumented with 'break wires'. These wires will be placed such that the crack arrives at (and breaks) a new wire each month, lighting up a corresponding LED. Technically the easiest thing would be to devise a fatigue machine rather than a pendulum. But for customers that would be "another incomprehensible fatigue test". A pendulum would be simple to understand, plus it has a 'Clock-like' appearance.
About the suspension: the safest way would be to use a leaf spring. But the disadvantage of this is that the point of application is not very well defined (in the crack growth calculation). An alternative would be to use a knife edge, with the disadvantage that it is unable to resist out-of-swing-plane forces.
About the drive mechanism: the simplest solution seems to me to have a permanent magnet at the low end of the pendulum and a repelling electromagnet below it at the center of the swing. No need to sense the direction of swing, just its arriving at the center position. Probably the time needed to build up a magnetic field is enough of a delay to let the weight pass the center position.