circuit's design

White LED Drive Circuit

How do you get 3.5 volts to drive a white LED when you only have a 1.5v battery?
(With special thanks to Dave Eselius for keeping this project alive until it was finished.)

New on this site since original publication of this page in 2002:

•  Simplest LED Flasher Circuit (in the world? You tell me.) (this page, Click here).
•  Rusty Nail Night Light demonstration added (this page, Click here).
•  Alternative_Types_of_Cores  (this page, Click here).
•  Toroid and Ferrite Bead Winding Notes (Click here).
•  New resistor-capacitor circuits to drive LEDs from 1.5 and 3 volts (Click here).
•  White LED Stroboscope (Click here).
•  Solar Powered Garden Light with PC layout (this page, Click here).
•  A Note About Transistor Selection (this page, Click here).
•  Common sources of ferrite beads (this page, Click here).
• A note of warning about LEDs and Eye safety (this page, Click here).
• Some tips on getting it running (this page, click here).
• Contributors to this page (this page, click here).

-  You may see copies of this page on other sites, but you will only see updates and improvements at  www.projects.cappels.org, the only site authorized to publish this material.

Be Careful About Peak Current A note of caution: These LEDs are comparatively expensive, so I suggest putting a small resistor (1 to 10 Ohms) in series with the cathode of the LED and measuring the peak current as inferred from the IR drop using either a scope or a peak detection probe (as described elsewhere in these pages) while testing this circuit to be sure you don't exceed the Manufacturer's recommendations. In lieu of specific guidance from the manufacturer, for greater reliability try to keep the peak current to half of the manufacturer's recommended maximum.

Overview
A minimum number of parts yields a compact switching converter that can provide sufficient voltage to drive white LEDs. The resulting lamp is much more efficient, in terms of lumen hours per pound of battery, than incandescent bulbs, and because the color of the light is determined by fluorescence of phosphors within the LED assembly, the color of the lamp does not change perceptibly as the battery runs down. As a result, very long battery life is achievable. This circuit is suitable for powering flashlights, emergency lighting, and other applications in which it is desirable to power white LEDs from one or two primary cell batteries using a low cost circuit.
The Circuit
The circuit could not be simpler than this. The transistor, 1K resistor and the tapped inductor form a blocking oscillator. When the power button is pressed, the transistor is biased on through the 1 k resistor. Voltage that appears from the tap on the inductor to the collector causes the voltage on the 1K resistor to be even higher than the battery voltage, thereby providing positive feedback. Also because there is voltage across the inductance between the tap and the collector of the transistor, collector current increases with time (this is in addition to a starting value that relates to the current supplied to the base, but this part of the collector current is rather small. Because of the positive feedback the transistor stays saturated until something happens to change its base current.
At some point the IR drop across the inductor from the tap to the collector approaches the battery voltage (actually battery voltage - VCEsat). As this happens voltage is no longer induced in the winding from the tap to the 1k resistor and the base voltage starts to drop, and this forces the base voltage to go negative, thereby accelerating the switching off of the transistor. Now, the transistor is off, but the inductor continues to source current and the collector voltage rises.
Quickly, the collector voltage gets high enough for the LED to conduct current, and it does for a little while, until the inductor runs out of current, then the collector voltage starts to ring toward ground base voltage swings positive again, turning the transistor on again for another cycle.

The Inductor
If you aren't designing this as part of a commercial product, you have a lot of latitude in the design of the inductor. The size of the core, its permeability, and its saturation characteristic (Physical dimensions, u and Bs) determine how many amper-turns it can sustain before it saturates. If the core saturates before the IR drop from tap to collector approaches the battery voltage, the circuit will switch quickly anyway because saturating the core makes the coil look like a resistor and coupling between the collector half and the base half (the side with the 1k resistor) drops to very little, so the effect is the same as the IR drop approaching battery voltage. The wire size determines how many amps the circuit will dray (well, ok, milliamps) before the IR drop gets large enough for the circuit to switch. The inductance constant of the core (physical dimensions and u mostly) determine how man microseconds it takes the collector current to rise to the point the circuit switches off, and it also determines how long current will be delivered to the LED when the transistor is off. Nearly every inductor parameter affects the performance of this circuit.
I have made this with ferrite beads a few millimeters in diameter and toroid cores up to a few centimeters across (take a look at the Rust Nail Inductor further down on this page). Here is the general relationship between core size and characteristics:
Large core: Easy to wind, lower frequency operation, higher power.
Small core: Harder to wind, higher frequency operation, lower power.
How to get started: Get a transformer core, preferably ferrite, and wind 20 turns on it. Tap by pulling a short loop of wire off to the side, then continue winding another 20 turns. Increase turns to lower frequency, decrease turns to increase frequency. I've used as little as 10 turns, center tapped (5+5) and operated this at 200 kHz. Experiment. Scroll down to the bottom of the page - the tiny ferrite core used in the #222 bulb base run at about 200 kHz.

A Circuit Enhancement
The neat thing about this circuit is that it has the bare minimum number of parts to do the job. The LED runs from pulsating DC and since its forward voltage is higher than the battery voltage, it does not interfere with the switching of the transistor. The drawback is that the peak-to-average current of the LED is pretty high, it could be 3:1 or 5:1, depending on the circuit (mostly on the inductor and battery voltage). If you want the LED to be brighter for a given peak current, you can add the diode and capacitor shown in the schematic below.

Something that one online critic has pointed out is that where space if is available, it is a good idea to add a bypass capacitor between the negative terminal of the battery and the center tap of the inductor. Some batteries have a high impedance, and the decoupling capacitor might serve to increase the output of the circuit. A value of 10 microfarads should be sufficient unless you are using a very high inductance inductor, in which case a larger one might be better. 
Where Can You Put the Power Supply?

 
Since this circuit has so few parts, I was able to make a couple of them so small that I could fit the entire circuit, including the inductor, 1 k resistor, 2N4401 (TO-92 package by the way), rectifier, chip capacitor, and Nichea NSPW315BS along with a little glue to hold it in, into the base of a number 222 incandescent penlight bulb. 
Using the LED replacement for the number 222 lamp, allows operation in a compact penlight. It gives enough light to walk outside on a moonless night. Based on about 35 ma currant drain at 1.5V, I estimated operating time, I estimated that this flashlight will operate for at least 30 hours of continuous use. That's a long time if your holding down the button. Data sheets for several of Duracell's alkaline cells can be found here.
Even though the battery voltage drops, the color of the light is always a crisp bluish-white. If its taken care of, this light "bulb" should last longer than me. I've had this one for 18 months as the latest update to this page and use this flashlight every night. I only had to change the battery twice. If the battery contacts didn't start to get intermittent from the corrosion, I wouldn't have know it was time to change the battery since the light was working fine. 


LED replacement for the 222 lamp is on the left. The entire power supply is contained within its base.



The hard part was taking apart the #222 lamp so I could reuse the base. 

Rusty Nail Night Light

These blocking oscillator type power supplies work best with ferrite cores, and sometimes they can be hard to locate. Some readers have expressed anxiety over making inductors, and that is understandable since to many, inductors have an aura of mystery about them.

Just to prove that these inductors aren't magic, or even that critical for that matter, I wound one on a rusty nail that I noticed laying beside the road one day while waiting for a tow truck. It is a 2-1/2 inch (6.5 cm) long flooring nail, which serves as the inductor's core.

The wire is a twisted pair of #24 solid copper wire that I pulled from a length of CAT-5 (ethernet) cable, which is similar to the wire used to connect telephones inside buildings. I wound 60 turns of the twisted pair in about three layers around the flooring nail, then I connected the start end of one conductor to the finish end of the other conductor and that made it into a 120 turn center tapped inductor.

I connected it to a 2N2222, a 1K resistor, a 1.5 volt penlight cell, and a white LED. Nothing happened. Then, I put a .0027 uf capacitor across the 1 K resistor (it happened to be on the work bench) and the LED came on. Sometimes you need .001 uf or so. The LED glows nicely and the circuit draws 20 milliamps from the AA cell. The waveform on the oscilloscope looks terrible, but the point is that the circuit oscillated with even this rusty nail, and it boosted the output of the 1.5 volt AA cell to over 3 volts peak to drive the LED.

Those who are familiar with some aspects of coil core selection would quickly point out that the eddy currents would be huge since iron has a low resistance compared to ferrite, or air for that matter, and that there would also likely be other types of large losses.  The point here is not that you should run out and buy some flooring nails to make LED lamps, but that this circuit was not "designed", but was thrown together and worked quite readily. If a rusty nail and some telephone wire is enough to light up a white LED, then the inductor is not so critical. So, relax, go buy a ferrite core and get started on your project.



(Above) Rust Nail Night Light. This LED power supply was thrown together in a few minutes using scrap parts.
Well, ok the LED itself was not really scrap.




Common Sources for Ferrite Cores

Wolfgang Driehaus from Germany wrote to point out that ferrite cores are used in compact fluorescent lamps, and he further stated that he has had success in making the cores work in this LED power supply circuit. It was only a day after receiving his email that I looked up at the ceiling and saw some lamps that needed replacement. Here is what I found.






Some compact florescent lamps from Sylvania had failed in my home. After buying new Philips lamps to replace them, I ventured into the garage to take one of the Sylvania lamps apart. The first problem was getting to the electronics in the base of the lamps. In later correspondence, Wolfgang showed me that the base of the lamp can be pried open and the circuit board removed without having to break any glass.  Be careful not to break the glass tubes in the lamp, as they contain mercury, which is toxic.




Inside the base of the lamp, as Wolfgang Driehaus, I found three inductors with ferrite cores, as well as a pair of high voltage transistors, a high voltage capacitor, and some other potentially useful components. The inductors were wound on three types of cores, which are a bobbin core (left, covered in heat shrink tubing), a toroid core (center) and an E-E core, (right).

I wanted to confirm the usefulness of the cores for myself, so I removed the existing windings from the bobbin and toroid cores. I cracked the E-E core in several places during the process of separating it from the coilform, so I did not have a chance to try it in the power supply circuit.

On the bobbin core, I wound 50 turns of #32 magnet wire, pulled out a center tap, and then wound another 50 turns. I connected this to a 2N4401, a 330 Ohm base resistor, and a white LED, according to the circuit at the top of this page. When I connected a power supply set to 1.5 volts, the LED lit up brightly. Ok, that is solid confirmation that the bobbin core from this particular Sylvania light works in this application.

On the toroid core, I wound 10 turns of #26 wire wrapping wire, pulled out a center tap, and wound another 10 turns. Connecting it in the same circuit (2N4401, 330 Ohms, white LED) with a 1.5 volt power supply, I saw that the LED lit up, but not as bright as with the bobbin core, but then again, I had only put 20 turns on the toroid.

So now we have a very common source of toroids. Compact florescent lamps are available in places, and as Wolfgang pointed out, they eventually wear out and need replacement.

Another reader pointed out that another source of ferrite cores is the shielding beads used on computer peripheral cables.  Those plastic-encased lumps on monitor, keyboard, and some USB cables are actually ferrite cores. If you are about to toss an old keyboard into the recycling bin, why not cut off that ferrite bead first? Christian Daniel of Gernany wrote, noting that the shielding beads are not idea for this kind of use, so you might want to try this last.



Alternative Types of Cores

If you can't find a ferrite core, or even an old rusty nail, all is not lost. You can still make a pretty good white LED power supply by using a non-magnetic core. It sounds like an oxymoron, but a nonmagetic core has little effect on the magnetic flux from the windings, and therefore does not interact strongly with the circuit -it is there mainly to provide mechanical support. Two experimenters have provided reports of their experiences with nonmagnetic cores, each with its unique attributes.

The wood core inductor

Bill Levan in the United States came up with a wood core inductor. His circuit powers a white LED from a 1.2 volt 700 milliamper-hour cell. Mr. Levan reports that he used the rusty nail night light circuit, but found that he did not need the capacitor across the resistor.

 

The coin is a United States quarter dollar, 2.54 cm  (1 inch) in diameter, for size comparison.
Take a close look at the LED - it really is on.

If the wood is dry, the material itself will not have a significant effect on circuit operation. Wood that has a lot of water in it might lower the efficiency of the circuit a little bit, but most likely, you would not be able to tell.

Mr. Levan's wood core is 5.08 cm x 12.7 mm x3.18 mm  (2 inches x 1/2 inch x 1/8 inch). The wire is 30 gauge solid conductor insulated wire wrapping wire, Radio Shack #278-501.

Wind 100 turns, pull out the center tap, then wind 100 turns more. In total, there are 200 turns.

You can contact Mr. Levan with questions about his wood core inductor and the circuit at the email address below.
  (The email address is an image.)

Air Core Inductor

Antonis Chaniotis in Greece converted his children's incandescent night light to use two LEDs in parallel, and increased battery life from one night to about 30 hours of light over three nights.




The LEDs, while emitting green light, are electrically similar to the white LEDs in the other circuits because, similar to most white LEDs, the die emits ultraviolet light, which excites green phosphors. Of course in the white LEDs, the phosphors emit white light. The large capacitor in the picture is 100 uf 25 volts, connected from the emitter of the transistor to the tap on the inductor, and it acts as a bypass capacitor to insure that the circuit sees a low impedance from the battery and switch. It may increase efficiency, especially as the batteries runs down and the batteries' internal resistance increases.

The base resistor is 10k, and the batteries is made of two 1.2 volt rechargeable cells in series.

Mr. Chaniotis' analyzed the circuit with SPICE and confirmed his findings by experiment. Interestingly, his analysis showed that for his circuit the optimum location for the tap is not in the center.

The coil is a total of 35 turns 80 mm (3.2 inches) in diameter with no core.  Start by winding 14 turns, the start is the collector winding. Pull out the tap for the battery connection, and then wind an additional 21 turns for the base winding.

Once, I made a similar coil for a different kind of application. I used a plastic food storage container from the kitchen as a coil form, then after winding, carefully slipped the coil off the container and held the wires together with tape. From the photograph, Mr. Chaniotis wrapped some wire around the bundle to hold it together.

You can contact Mr. Chaniotis with questions about his air core inductor and the circuit at the email address below.
(The email address is an image.)



Solar Powered Garden Light


This simple flyback/blocking oscillator LED power supply was adapted to and built into solar powered garden light by a talented experimenter in the United States known as "mrpiggss". The method of switching the power supply off during daylight hours, thus allowing the rechargeable cell to recharge, was taken from a circuit design by Nick Baroni, of Willetton, Washington, and published on the siliconchip.com.au website.  Here is the design from mrpiggss' workbench.



The 1.5 volt cell used in this picture was replaced with a 1.25 volt nickel-cadmium cell.




Actual values used in mrpiggss' circuit.

Mr. Baroni's original circuit used a BC547, but mrpiggss found that a BC547C (the version he bought) would work as Q1 but not Q2 (see "A Note On Transistor Selection, below). In his version, mrpiggss used 2N4401 transistors for both Q1 and Q2, to keep the bill of material as simple as possible.  He also noted that if R1 was 15k, the sunlight gating function would be more sensitive, thus keeping the LED power supply off until it was darker than when R1 is 22k.

The core for L1 came from ebay with no part number or supplier but it is the size of a penny and about 3mm thick. The wire came in a 3 pack of magnet wire from Radio Shack and and are pure copper.  Being green in color and 30 gauge, the wire is easy to handle and wind. The cores have 40 turns, 20+20 (wind 20 turns, pull out the center tap, then wind 20 turns more). and removing the coating is easier than with really thin wire. The insulation on this wire can just be burnt off with a lighter, and it comes off just like magic. No scraping or sanding, although he use a small emery board to give it some tooth for soldering. 

Another very nice method of removing the insulation from magnet wire came from Christian Daniel of Germany: To quote his email:
     "Scratching away the insulation is difficult with thin wires, it's too easy to cut or weaken them. I prefer fine corund sanding paper. Or - for very thin wires - I use an Asperin(R) pill and press the wire with the tinned hot soldering tip on it : opacht ! Put your eyes and nose away ! The hot organic acid destroys the insulation and it can be tinne nicely."
 
The solar panel is a standard one-battery panel. it has no info on it but puts out 1.5 volts in full sun. No idea about the current rating.

D1 can be nearly any silicon or germanium diode, provided it is rated to handle the solar panel's output current into a short circuit. For most of the panels used in garden lights, this means basically, any diode you can buy. A low power Schottky or germanium diode would have a lower forward voltage drop than a small signal silicon PN diode. The 1N4001 series, as used by mrpiggss is a good choice as well because its large junction area results in a relatively low forward voltage drop.




This is the "circuit side" of the printed circuit board.
The components are mounted on the opposite side. It should be noted that
the basing for the transistors corresponds to the 2N4401, and NOT the BC547.

mrpiggss' email address is "thetraindork (at) gmail.com" -please note that this email address needs to be retyped with "@" in place of (at). This is meant to stump spam robots, not people.

 A Note About Transistor Selection

The transistor used in this circuit can be any one of a wide variety of transistors. I recommend trying the 2N4401, 2N3904, and 2N2222, listed in no particular order. Dariusz Flaga in Poland, noted that the BC338 is popular in Europe and that its specifications, including voltage and current ranting, and saturation characateristics, suggest that it is a good fit for this application. Mr. Piggs in the U.S.A. (see the Solar Powered Garden Light, above), successfully used the BC547 on several prototypes.

You can even use PNP transistors, but if you do, remember to reverse the battery and LED connections. I have received email from a couple of project builders who used very high gain transistors and had trouble with their circuits. Transistors with high DC current gain (hfe) tend to switch slowly and may operate inefficiently or not at all.  Stay away from  the very high-gain audio transistors in particular.  The small capacitor across the base resistor, as shown in the Rusty Nail Night Light can speed up the transistor and make oscillation possible when you have low a inductance inductor, or if the transistors are a little slow. Speedup capacitors beyond  .0033 uf (33000 pf) are probably excessive and may actually slow down the circuit by causing the base to be over-driven.  If your circuit doesn't work, consider using a faster transistor.



Some tips on getting it running

Bob Parrott sent in the following helpful hints:

I had great fun playing around with your blocking oscillator circuit and thought your readers might like to try this variation if they have difficulty getting the thing to start oscillating.

Initially, I tried a BC107 transistor (because it was to hand) but it would only oscillate if I lowered the base resistor to 220 ohms and when I measured the current taken from the battery it was about 90mA - even before the LED was connected.

A BC108 oscillated easily as did a H945 removed from a dead switcher - also a good source of coils/ferrites.

But I was intrigued with getting the BC107 to run and added a small capacitor (22nF) across the base resistor to 'kick-start' the oscillations. It worked so well - with various transistors and coils that I was further intrigued to see how much I could increase the resistor value - hence the 20k trimpot. (The 22nF also got over the problem of the oscillator failing when I tried to add an ammeter in the battery circuit).

I found it would continue to oscillate right up to 20k ohms and this also had the effect of reducing the supply current (osc only) from 90mA to 800microamps - very important if using batteries.

Bob Parrott's email address is:  beep1952@hotmail.co.uk



Danger of Eye Damage From Visible Light Emitting Diodes





Contributors to This Page

What started out as a simple web page to share a simple circuit that allows white LEDs to be driven by a single 1.5 volt cell has grown bit-by-bit over the years to make it even more helpful to experimenters.  The contributions of the individuals listed below, in the order of the appearance of their contributions on this page, is noted with appreciation.

Wolfgang Driehaus             - Ferrite cores from compact florescent lamps.

mrpiggss                               - Garden Light.

Bob Parrott                          - Tips on Getting It Running.

Geoff Davies                         - LED Safety

Bill Levan                              - Wooden Core

Antonis Chaniotis                 - Air Core

Christian Daniels                  - EMI Cores, removing insulation from wires.


And many others who provided suggestions and feedback to improve the page, but who were not directly quoted.