A boost converter circuit changes a varying lower voltage, (a battery pack), to a regulated higher voltage, (which goes to the LEDs). This is accomplished in two (simplified) steps. The first step is to pump a inductor with current from the power source. The second step is to try to "turn off" the current in the inductor. Due to the nature of a inductor, it wants to try to keep the current going even if it was turned off, (like a flywheel). So to compensate for this, the inductor increases its voltage in an attempt to keep the current going. This increase in voltage is then transferred to the output of the circuit. By varying the ratio of inductor on time to inductor off time the output voltage can be changed, (this is called the duty cycle) The Maxim integrated circuit controls the duty cycle and ensures that the output voltage stays constant (within +- 1% ideally).

A short circuit description: the components in the box on the circuit diagram marked "keep wires in box as short as possible" are the actual boost voltage conversion circuit. The rest of the circuit is the controller for the boost voltage conversion circuit. Note that there are two feedback mechanisms in the circuit. The first one is the usual voltage feedback, which controls the output voltage. The second one is an averaged output current feedback. The current is sensed in R1. The reason for the averaged output current feedback is to ensure that the LEDs are not burnt out due to too much current. The circuit has no protection for the LEDs other than this, and there is a chance that due to component aging and temperature variations in the circuit that the LEDs might be over stressed with out it.

The power inductor value that is suggested in the Max 1771 data sheet (22uH) is too small in value for a 9 LED load. This makes the circuit operate in discontinues conduction mode. The consequence of this is that the circuit has higher than necessary inductor currents, which lead to greater resistive power losses in the Mosfet and inductor. I found that 150 to 100 uH seemed to give the best performance. There is a section in the Mohan text, listed in the reference section that gives formulas to figure this all out. The key to applying the formulas is to assume the worst case scenario. Note: if more LEDs are added to the circuit the inductor value can be reduced.

The efficiency of this boost regulator varies somewhat with varying input voltages, but in general it is about 94% efficient for fresh batteries, and about 70% efficient with dead batteries (for 4 AA alkaline batteries). With 3 AA alkaline batteries the efficiency was 94% with new batteries, and with 2 new batteries the efficiency was 68%. By the way, the circuit will operate with 2 to 6 AA batteries for the power. With more batteries the light will last longer.

Looking at the numbers one would ask what was gained with the complex boost regulator. The biggest advantage is that the circuit will extract the last bit of life out of every battery. Where as with a normal incandescent lamp the batteries must be thrown out when the individual cell voltage is about 1.2 volts, the boost regulator will function with individual cell voltages as low as .5 volts. This allows one to use the batteries longer with no noticeable degradation in light output. Testing the actual circuit with 9 LEDs, and 4 AA Duracell batteries, (which were a year after their use by date) the lamp was lit at 100% brightness for 23 hours; the last 2 hours the light dimmed somewhat since the input voltage was so low (.88 volts). This is since loses in the circuit are inversely proportional to input voltage. As a side note, the circuit will not turn on with total battery voltages below 2.5 volts, but once the circuit is turned on it will operate with battery voltages below 1 volt.

The key to building this circuit is sort and direct wiring of the components and proper component selection (just use parts from parts list). The circuit board listed in the parts list is fairly small. If the circuit is built on this same pref-board, the wiring lengths should work out fine. Connect the wires as listed on the schematic diagram, paying special attention to the connection lengths of the wires between the Mosfet, capacitors, the diode, and the inductor. I attached the circuit board to the lamp housing with RTV silicon. This works, but makes future modifications a little tough. The ambitious might figure out a better way. After connecting the circuit, there is one adjustment to be made. Using a voltmeter, measure the voltage across the bank of LEDs. It should be 10.5 volts. If it is not, turn the screw on R4 until it is. Actually it would be better to measure the output current when making this adjustment (75 mA), but that is a little tougher.

To house the circuit an el-cheapo Rayovac headlamp from Wal-Mart was used. The regular incandescent lamp was ripped out and a circuit board with the LED lamp was inserted in its place. Leave the original on-off switch in the headlamp, use it to turn the new LED circuit on and off. It is important that when soldering the components into the circuit board that the corners of the board are left free. This is since you must nibble the corners of the circuit board off to make the light fit in the lamp housing.

The wire that connects the battery pack to the lamp on the Rayovac headlight is too short to put on a helmet, so disconnect the original wires at the battery pack end, and remove the plug at the light end. Take a piece of lamp cord and solder this in, in place of the original wire. In addition, epoxy in the wire at both ends where the wire passes through the housing, this will provide strain relief, and make sure no water gets in. Some epoxy’s don’t stick well to certain plastics so make sure to test the strain relief operation before using the light. A good idea is to make a knot in each end of the wire so a epoxy failure won’t allow the wires to rip out.

Since the LEDs vary some in their individual voltage characteristics, it is best to match each series LED string. Connect each LED in series with a 1000 ohm resistor, and connect this to a 9 volt power supply. With a voltmeter measure the voltage drop across the LED. Some LEDs will have a slightly higher voltage drop, some lower. What you want to do is equalize each of the series LED strings so that the total voltage drop is about the same for each string. Eg, if one LED has a lower voltage drop, and another has a higher voltage drop and another LED’s voltage drop is in the middle you would put all these LEDs in one series string so that the voltage drop is equalized with the 2 other strings.

Some improvement that would make this circuit better would be more LEDs. The circuit can drive 30 LEDs with no problems, but it would be awfully expensive to build It would be $112 just for LEDs. 20 LEDs might be a better compromise. The circuit now is set up to operate with 3 parallel strings of 3 LEDs in series. The circuit might be a little more efficient if instead there were 5 parallel strings of 2 LEDs in series. However, this change would necessitate that the output voltage be change from 10.5 to 7 volts. To do this the resistor network consisting of R2, R3, and R4 would need to be re calculated. Also, the current sensing resistor, R1, would need to be changed. Better mechanical lay out and interfacing to the light are some other areas to improve on.

Note: I have been using this light for over a year with no problems. One time while doing a sump the internals of the light got wet, but the circuit still worked. ROBUST ! In the next version I would most likly pot the circuit in silicon so if the housing leaks the circuit don't get wet.

Hosfelt Electronics 800 524 6464

Digi-Key Electronics 800 344 4539

Mohan, Undeland, and Robbins, "Power Electronics", Wiley Inc, pp 172-177

Nevin W. Davis, "AN INVESTIGATION INTO THE FEASIBILITY OF DESIGNING A "REAL" CAVE LAMP USING WHITE LEDS" contact author, (a grotto member), for availability

Agilent Technologies, HLMP-CW15, (Data sheet for white LEDs), http://www.semiconductor.agilent.com/led_lamps/hlmpcw15.html

Maxim Integrated Products, max1771 data sheet, http://dbserv.maxim-ic.com/quick_view2.cfm?pdf_num=1030

Rayovac Corp, "Application Notes and Product Data Sheet", http://www.rayovac.com/

Thanks to Nevin Davis for his Technical Assistance

As a post-script the price of the LEDs went down to $2.99.

Useful Links

CREG

Radio Location Methods

NSS Homepage

Go Penn State!

PA Cave Conservancy

Power Electronic Soc.

.

Copyright Notice: Commercial use is prohibited without express written authorization from Nathan Hirneisen. Blah Blah Blah.....(like in Peanuts)

  Page Created and Copyright (c) Nathan Hirneisen, Aug 2001