Complexity vs Efficiency

A natural question to ask is why to go with such a relatively complicated circuit as mine and not with something much simpler. Why not simply permanently connecting a rectified voltage from the dynamo in parallel with rechargeable batteries and lights, as in the setups described in the pages by Andrew Henry, or Sheldon Brown, or in the circuits by Steve Kurt? Below I try to explain why all the hassle.

The voltage from a dynamo is alternating, changing sinusoidally with time. For a parallel connection with a battery, the current from a dynamo must be rectified, given one direction, by passing the current through a bridge out of diodes. When Schottky diodes are used for the rectification, there is a drop of voltage across two diodes of about 0.6V and regular diodes give an even larger drop. In the low-voltage bicycle circuit, the drop represents the loss of power of about 20%. The circuits mentioned above suffer from this loss.

At the normal bike speed, after rectification, the voltage from a dynamo changes in magnitude from 0 to about 8.2V. In the parallel connection, it is only when the voltage from the dynamo exceeds the voltage across the battery, of ~6V, that the dynamo powers the lights. This only happens during about 50% of time of the cycle of the dynamo. During this 50% of time the dynamo powers the lights and charges a bit the battery. During the other 50% of time, the lights are powered by the battery! In effect, with a simple circuit, you will be actually draining the battery while riding with the dynamo on. In addition, you will be somewhat overvolting the lights; Steve's circuit limits the overvolting but has the problem with battery drainage. The batteries must last longer than twice the time without the dynamo as they get a bit charged during half of the cycle. You can experience the reduction in the power taken from the dynamo by comparing the effort needed in cranking a wheel when the dynamo is connected in parallel with the battery to the lights to that when the dynamo alone is connected to the lights. The reduction is very significant. What does not come from the dynamo must come from the batteries. In addition, the charging and discharging of the batteries, at typical currents, involves at least 20% loss in efficiency. This additionally reduces the performance of the above circuits.

My circuit also operates in the manner described above, and suffers the deficiencies, but only when the dynamo voltage is deemed to low to drive the lights on its own. Then the power switches to the battery and the dynamo then provides assistance if it can, i.e. when its voltage around the peak exceeds that of the battery. Should there be a better way still? Yes, one could envision chopping the low voltage of the dynamo, using an inductor, and producing voltage pulses exceeding the battery voltage for extra fractions of the dynamo cycle. Maybe just a transformer would do. When the dynamo voltage is adequate, i.e. most often under normal operation, my circuit connects lights directly to the dynamo. If there is an excess power available, it is used to charge batteries. If the batteries are full, then the standby circuitry takes less than 1% of the power.

When lights are off, the dynamo charges the batteries in the simplest circuit in an unregulated fashion. That means that even when batteries are full they continue to be charged. This proceeds with a large current and heats up the batteries gradually destroying them. Steve Kurt's budget circuit prevents that from happening by dumping the current into a transistor instead and heating the transistor. While the batteries are spared, the consumed power is comparable with the lights on or off. The operating dynamo acts as a partly engaged electromagnetic brake with these circuits, no matter whether any useful work is done powering the lights or charging the batteries. With my circuit, if the lights are off and the battery is full, the electric power consumption is less than 1% of that with the lights on. Still another Steve's circuit has also that feature but suffers from the drawbacks of the parallel dynamo and battery.

Although my circuit has different satisfactory features, there are reasons to look for a better circuit. One reason is to get a more continuous delivery of power to the lights. Because of the decision making on the voltage, the lights dim a bit when switching from the dynamo to the battery, though not the other way around. A related issue is that of finding a better utilization of the dynamo at low speeds.