Monday 21 June 2010

Timing Revisited


Following on from the discussion “Timing Is Critical”, the above circuit provides control of the Bedini Motor that was hitherto not available. Maybe the time is coming when this should no longer be considered to be a Bedini Motor and we can wave goodbye to the SSG.

The circuit utilizes a CMOS 4538 Dual Retriggerable Monostable Multivibrator as the main controlling element. The transistor is non-critical and any NPN transistor can be used. The reed switch could be any form of sensor; Hall-Effect, optical or a trigger coil; the only proviso is that whatever the sensor is, it must provide the 4538 with the correct logic levels. (We used a Meder Electronics Reed Switch from Mouser - 876-KSK-1A35-1520)

The first monostable is used to control the speed of the motor. It is used in a retriggerable configuration. The time constant (RC) is chosen to set the RPM of the motor. So, if an upper limit of 2,500 RPM is required then this equates to a period of 60*1000/2500mS or 24mS. If C is chosen to be 100nF then the corresponding resistance is 24mS/100nF = 240K. (This assumes only one rotor magnet; our design uses two magnets, so the resistance is halved - 120K.)

The second monostable is triggered by the leading-edge of the first monostable´s output and provides the control of the power pulse duration. The approximate duration should be about one tenth that of the speed monostable; this is reflected in the capacitor size of 10nF. There is no `hard and fast´ rule for the optimal duration - it needs to be of sufficient duration to accelerate the rotor. (As will be shown later, it is non-critical and is largely dependent on the type of rotor and bearing quality.)

When the rotor starts spinning, the first monostable is triggered which in turn triggers the second monostable and therefore fires a power pulse to drive the rotor. This sequence continues until the rotor is spinning at a rate that is just greater than the corresponding period of the first monostable. At that time the monostable has not completed its timeout period and as the monostable is retriggerable, a new timeout interval is restarted - therefore, for that trigger there was no consequent power pulse. Only when the rotor has slowed down fractionally will another power pulse be delivered. So there is now a balance, speed is maintained constant and there is a considerable saving in power consumption.

So, even though the power pulse duration might not be optimum, the action of a more powerful pulse will simply push the rotor a bit faster and the time interval to the next pulse will have been extended.

A third monostable could be introduced between the existing two to control the interval between the trigger pulse and the power pulse. In practice, this is not necessary as the trigger sensor normally can be moved physically in relationship to the power coil - but the option exists !

The first monostable can be reconfigured to be non-retriggerable. This means that a power pulse will occur for every trigger pulse. Now what happens is that the rotor will rotate at a rate that is always a little greater than the monostable period - sometimes considerably greater ! This mode consumes more power but is better suited to those who want to continue to `recharge´ their batteries !

From this most basic of circuits, there are now opportunities to add greater functionality and to consider different design options. For example, in our levitating motor, the pulse synchronizes with the rotor sending it into a wobbly resonance. One solution is using two monostables, to set an upper and lower speed. When the rotor exceeds the upper speed, it then freewheels until the lower speed is reached; the sequence repeats and voila - no more wobble and another slight saving in power consumption !

As specialist digital systems design engineers, we consider that monostables belong to the analog domain, not part of a digital system. So, another approach is to use a crystal oscillator and digital counters. Now all sorts of options become available - counting the RPM is child´s play, as is replicating the functionality of the monostables. Logic can be introduced to respond to power availability. It becomes easy to determine the rotor state - is it speeding up, slowing down or stopped ? With the appropriate rotor design, self-starting can be achieved too.

Operating parameters can be dynamically changed to maintain efficient and smooth running ... but that is moving into the realms of microprocessors and software. Isn´t that overkill for a motor ? No, it´s fun and that´s what it´s all about ! It beats the `socks off´ trying to charge car batteries and potentiometer tweaking ! Nor is it expensive.

For those of you making Bedini Motors from Imhotep axial fans - have you ever stopped to look at the technology and functionality of the chips that drive these motors ? Some are utterly amazing and you throw them away and replace it with a neon, a 2N3055 and a potentiometer then call it progress - shame on you ! If all that you want is the back emf from the motor then add a diode and extract it and leave the technology intact !



This recording is from our levitating motor using the above circuit. It accelerates from start to 2,500 RPM and there are two brief periods when friction was applied to the rotor. The clicks are the power pulses - as the speed stabilizes, the number of pulses decreases. The power consumption peaks during startup at 2.8mA @ 3.0V, dropping to less than 1mA when the rotor reaches its target speed.

(The hum in the background is the induced voltage produced by the free spinning rotor.)


Our levitating motor uses two magnets. There is a tendancy for the circuit to synchronize with one magnet, so the power pulse applied to that single magnet makes the rotor wobble and resonance is established - this undesirable effect is clearly audible.

Visitors