Marko Mäkelä’s electronics projects: Dynamo-Driven LED Tail Light

[One LED driven by an LP2951][Four LEDs as a rectifier bridge feeding an LM134]Background

I equipped my bicycle with a SON dynamo hub and a B+M D’Lumotec Oval Senso Plus front light. Because no suitable model of LED tail light was available at the time of the purchase, I decided to replace the incandescent bulb in the original tail light with a LED. It turned out to be a good decision, because the B+M 4DLitePlus does not work when connected to the front light, apparently due to insufficient operating voltage.

Obviously, a LED cannot be directly connected to a dynamo. The dynamo produces alternating current. When the load is small, the voltage can easily exceed the maximum reverse voltage of a LED. So, a rectifier bridge will be needed. Furthermore, very few LEDs can survive a current of 0.6 amperes, which the SON28 is able to deliver. The Luxeon III Emitter is an exception, because it tolerates up to one ampere. It makes an excellent front light. A tail light is better driven with a smaller current, such as 20 mA.

When the dynamo is driving the B+M D’Lumotec Oval Senso Plus front light, it will simultaneously deliver about 3.5 volts to the tail light. The voltage drop of a red LED is at least about 1.4 volts. The cheap solution would be to connect two LEDs in series and to complete the circuit with a resistor. However, the voltage will rise at higher speeds, causing the LEDs to receive more current. Furthermore, the maximum reverse voltage of the LEDs can be exceeded with this half-wave rectifier circuit. Last but not least, should the wiring or the front light fail, the dynamo would deliver at least 20 volts to a small load, which would likely destroy the LEDs in the tail light. To prevent situations like this, a current limiter with a wide input voltage range is needed.

The schematic diagrams on this page were drawn with CadSoft EAGLE 4.16.

Constant Current Source with a Transistor

In its simplest form, a constant current source consists of a single NPN transistor that is given an appropriate bias voltage. The load is connected between the positive supply and the collector. The current is controlled by a resistor that is connected between the emitter and ground. Unfortunately, this circuit works best with a very narrow input voltage range. The current tends to be proportional to the supply voltage.

The LM134/LM234/LM334 Adjustable Current Source

[Parts of the LM134 light][Schematic Diagram of a 3..7-volt AC quad LED Tail Light]The LM134 is programmable from 1 µA to 10 mA at a voltage drop of 1 V to 40 V. If you find the maximum current insufficient, you can connect multiple LM134 in parallel.

The minimum input voltage is about 3 volts AC. If the rectifier bridge is built of four red LEDs with a dropout voltage of 1.4 V, the set current will be reached when the voltage across the circuit exceeds 2*1.4 V + 1 V for the LM134. If the minimum input voltage is higher than 4 volts AC, you may wish to connect some LEDs in series with the LM134 to reduce the amount of power wasted in the LM134.

I built the circuit with four bright 5 mm red LEDs that are specified as 18 Cd with 20 degrees viewing angle at 20 mA. The LEDs are too bright to look at already when driven at 10 mA. The dropout voltage is 1.7 V, but the light seems steady. Because of the higher dropout voltage, the minimum input is about 3.5 V alternating current.

Note that the maximum input voltage is only about 7 volts AC, because the maximum reverse voltage of each LED in the rectifier bridge is about 5 volts.

The LM134/LM234/LM334 Adjustable Current Source with a Single LED

[Parts of the LM134 light][Schematic Diagram of a 4..20-volt AC LED Tail Light]Using LEDs for voltage rectification has three drawbacks:

The LEDs are only illuminated when there is some forward current and voltage across them. During the reverse cycle, the LED will be completely dark.
limited breakdown voltage
The maximum reverse voltage of a LED is about 5 volts. Normal diodes can withstand up to hundreds of volts in the reverse direction.
Multiple LEDs require more space and complicate the design of good optics.

After upgrading my bicycle in the Spring of 2007, I rebuilt the tail light inside a rectangular Cateye RR-395BT tail reflector that is a perfect match for the Tubus Logo carrier. The power cord is a coaxial audio cable that runs through the frame to the rear fork and through the hollow tube of the carrier. I had to drill two holes for the cable: near the top of the down tube and near the left top of the carrier. Because of the limited space that is available inside the reflector, I had to implement the rectifier bridge with discrete diodes. The physical circuit is L-shaped and enclosed in heat-shrink tubing. There is no capacitor, because the LM134/LM234/LM334 has no problem with pulsating current. The allowed input range is about 3 to 20 volts AC.

This tail light was in daily use for a year. However, I replaced it with a B+M Toplight Flat Plus, which features better optics and a capacitor-backed parking light function.

The LP2951 Adjustable Micropower Voltage Regulator

[Parts of the LP2951 light][Schematic Diagram of a 3.5..30-volt DC LED Tail Light]The LP2951 is a programmable low-dropout voltage regulator for up to 30 V and 100 mA. It can be configured as a constant current source with a dropout voltage of 2 V. With one ultra-bright red LED (1.4 V voltage drop), the input voltage range is about 3.4 V to 30 V plus the dropout voltage of the rectifier bridge.

Contrary to my hypothesis, the capacitor at the output of the rectifier bridge (C1) cannot be omitted. Apparently the LP2951 was damaged after I removed the capacitor. Also, the LM117/217/317 configured as a current source would be unable to regulate the current if the capacitor were omitted.

A better efficiency for small currents can be obtained by replacing the LP2951 circuit (between the LED and ground) with one or more LM134 connected in parallel.

Switching Regulators

The circuits outlined in this document are essentially linear regulators. Current sources based on switching power supplies would yield a better efficiency at the cost of added complexity. When driving one LED, the maximum efficiency of the LP2951 circuit is only about 40 %. The more voltage has to be dropped, the worse the efficiency. However, typically the total power consumption is small, less than 100 milliwatts, or a few per cent of the power consumed by the front light.

A voltage-doubler circuit combined with a step-down switching regulator, such as the LT3474 (4..36 V, 1 A) could be worth experimenting. A parking light could be implemented by equipping the input of the switcher with a large capacitor.