A low power, long life LED flashlight circuit

Posted by Matthew Fri, 03 Mar 2006 04:17:00 GMT

Scouring the Internet for information on LEDs, I accidentally stumbled upon a PDF detailing a flashlight made from PVC. For the torch, he biases ultra-bright LEDs with ballast resistors as described in my LED lighting guide. This is a simple solution, but a with a slightly more complicated circuit we can extend battery life by over ten times!

Download PDF instructions for the inefficient version

PVC flashlight LEDs

Electrical losses in the flashlight

In this flashlight, each LED/resistor combination consumes 4.5volts at 30mA or about 135mWatts. The ballast resistor alone consumes 1.1volts at 30mA or about 33mWatts. Therefore, 25% of the power being drained from the batteries is lost in the resistor of each LED chain!!

Battery life

The ‘C’ size batteries in the PVC flashlight have a capacity of around 4 500mAh. If you build your flashlight with seven LEDs, there is a constant current drain of 7 * 30mA or 210mA. Dividing this into our battery capacity, it becomes obvious this flashlight will only stay lit for about a day. This is a far cry from the 50-100 hours claimed by commercial flashlights running on smaller ‘AA’ batteries.

Cutting your losses

As mentioned above, the ballast resistors are wasting 25% of our battery power. Electronic Design proposed a simple circuit to resolve this in a recent article. The front end of their circuit draws less than a milliamp of extra current.

LED Boost Controller

The circuit is best described in two parts: one, the boost circuit function of Q1 and Q2, and two, the control circuit of Q3 and JFET1. Assume Q1 is off. With the battery voltage slightly above Q2’s VVB, a positive Q2 base current [iB = (battery voltage VBE)/RJET1] would flow. Q2 turns on, which switches inductor L1 to ground.

The end result is a 23volt pulse (as shown in Figure 2) across the series of ultra-bright LEDs. At 278kHz, the human eye cannot distinguish the difference between these pulses and a constantly lit LED. This saves even more battery power.

As the battery voltage decreases, the pulses become further apart. The brightness remains indistinguishable until the voltage falls near 2volts. (The circuit does not function well below 2volts) I doubt the PVC flashlight has this efficiency near the end of its life.

The extended battery life

According to Electronic Design, this circuit consumes an current equivalent to about 17mA. Powered by our ‘C’ cell in the PVC, this circuit could run for 265 hours! This is ten times the original PVC design.

Comments

Leave a response

Alan said about 13 hours later:

Cheap, rugged, long lasting. Good idea!
Clu said about 13 hours later:

This was posted previously on the Make blog entry for the 1,024,000 mcp portable light: “I know that some might think that pulse with modulation (PWM) might be better solution “but” when using with white LEDs, PWM dramatically reduces the life of the phosphor. PWM does have its advantages in portable devices but if you’d like to have the most efficient LED array, use resistors and proper power source and avoid PWM if possible. ”

Is that (PWM) esentially what this cicuit is doing?
AIVAS said about 13 hours later:

Yes this circuit is a PWM.
Clu said about 16 hours later:

Any credence to “PWM dramatically reduces the life of the phosphor.”
rakess said about 17 hours later:

Any ideas on how to make one of these kinetic powered … would be a realy fun project to try
onlyocelot said about 18 hours later:

278Khz, eh O_o

Thereminists beware, 270Khz is a very popular fixed-oscillator frequency, so having this flashlight around and on will make an 8Khz scream.
Matthew said about 23 hours later:

In regards to PWM, for dimming LEDs it is reccomended over just lowering the current through the LEDs.

The fact that PWM decreases the lifespan of the LEDs is believable. However, in this situation the end justifies the means. The simplicity of the circuit and elongated battery life are worth the small risks.
bertrik said 2 days later:

First of all the circuit is not a PWM, looks more like a step-up converter. For modern LEDs, PWM is less efficient than simply a lower current. The extended battery life is a bit misleading I think, because the LEDs are probably just getting much less current than in the original design.
Matthew said 2 days later:

betrik: You are right, the circuit is technically not PWM, but it is implementing a very similar idea.

The circuit is definitely more efficient than a ballast resistor. However, it does gain most of its ‘efficient’ from the pulsing. Honestly, if you implemented this circuit, you would barely notice any difference in brightness. The human eye reacts somewhat logarithmically to light. (The brighter the light the less sensitive we are.) It is also difficult to distinguish the difference as the period between pulses decreases.

I have seen PWM recommended in LED documentation on several occasions. It may or may not be more efficient depending on your implementation. As far as performance and specs, no two LEDs are guaranteed to have similar responses outside their rated current.

In this particular application, the circuit required to drive the LEDs at a constant current without extraneous losses is significantly more complex.
TooZ said 4 days later:

But did I read their article right that they are using 20mA White LED’s? Most of the ones I have seen are ~30mA. I wonder how much that will kill the theoretical efficiency.
harrygibus said about 1 month later:

Is the voltage on this circuit 6v? Am I right counting 4 cells on the diagram?

Thanks, Jay
harrygibus said about 1 month later:

Thanks anyway, I read the original article and saw the voltage listed on their diagram.

Does anyone know how you would go about dimming the LEDs in conjunction with this circuit design? I’ll looking for a simple solution (like a potentiometer somewhere in the circuit). I’m new to this so I am also doubting the answer is that simple?

Thanks, Jay
TooZ said about 1 month later:

I am not sure where you would want to put a pot to dim the lights, but keep in mind that you are going to be burning power by having the pot there and you won’t get nearly as much efficiency by having it. I was warrant a guess that dimming this wouldn’t be a quick and easy fix due to the way it is designed. Don’t take my word as a final answer though, I would love to hear what someone else has to say on the topic.
craig said 3 months later:

i believe puting a pot in the circuit would allow you to adjust the frequency does “dimming” the leds by reducing the on time without raising the load wattage of the circit significantly. can anybody confirm this?

Mike said 7 months later:

Has anyone actually built this circuit. I have modelled it in LT Spice and cannot get it to oscillate.

As the original text suggests I have used a 1N4148 in place of the schottky as I do not have a spice model for the SD101. I have tried various schottkys in this place and cannot get the posted waveform at the collector of Q2.

Can someone help me correct my Spice model before I go out and buy the components for this circuit?

C2 N002 N001 0.1µ
R2 N001 N008 820k
R4 N002 N006 1meg
Q3 N006 N008 N001 0 2N3904
Q1 N010 N007 0 0 2N3904
Q2 N003 N010 0 0 2N3904
L1 N001 N003 100µ
R1 N003 N007 22k
C1 N003 N007 220p
R3 N008 N006 1.2meg
JFET1 N001 N006 N010 2N5461
V1 N001 0 6
D§LED1 N003 N004 QTLP690C
D§LED2 N004 N005 QTLP690C
D§LED3 N005 N009 QTLP690C
D§LED4 N009 N011 QTLP690C
D§LED5 N011 N013 QTLP690C
D§LED6 N013 N012 QTLP690C
D§LED7 N012 0 QTLP690C
D§SD1 N003 N002 1N4148
.model D D
.lib C:\Program Files\LTC\SwCADIII\lib\cmp\standard.dio
.model NPN NPN
.model PNP PNP
.lib C:\Program Files\LTC\SwCADIII\lib\cmp\standard.bjt
.model NJF NJF
.model PJF PJF
.lib C:\Program Files\LTC\SwCADIII\lib\cmp\standard.jft
.tran 0.01
.backanno
.end

Par said 8 months later:

Could this type of circuit be built with only 1 or 2 LEDs and an on/off switch? I need to build a small circuit board with only 1-2 white LEDs using 9v or less and it needs to last at least 14 days constant or 28 days, running it 8-9 hrs a day. Is it possible? Can ya help me out? I am not extremely knowledgable in this field. Thanks, Phil
Acme Fixer said over 2 years later:

The two transistor circuit, Q1 and Q2, is basically a blocking oscillator like the Joule Thief. Remember that this circuit works by regulating the output to a constant voltage, which isn’t really what the LEDs want. The higher the output voltage, the higher the rectified voltage across C2, which then restricts the current thru the JFET, and hence to Q2 and the V boost circuit.

I’m not sure this circuit requires a JFET, other than to limit the current, but the same thing can be done with a resistor and a transistor. You can remove the JFET and replace it with a 1k resistor, and the circuit should work okay, since that’s the same circuit seen in other articles. But the first thing to do above all else is to replace Q2 with a 2N4401, since a 2N3904 can’t handle the high peak current over 100mA that is required. Also it doesn’t seem necessary to use a Schottky diodes, a 1N4148 should work okay, since there is just a very small current thru the diode.

Just tack solder it together and see what happens when the parts are changed. Have fun.
Flashlight z said over 3 years later:

Great DIY guide. Only thing lacking is the apparent battery life. How would the cost compare to the battery drain versus just buying commercial?
zinaung said over 3 years later:

I am zinaung .My education is 2nd year B.A(geog). My hobby is construction of simply circuits ,so daily reach to circits your website is most attractive as learning pages I think.Can your some circuits sent me? If you will link me,my G-mail is “bozinaungkalate.sgmm”.
ny dentists said over 3 years later:

Dentists in New York, NY, See Reviews and Book Online Instantly. All appointment times are guaranteed by our dentists and doctors.

This comment has been flagged for moderator approval. It won't appear on this blog until the author approves it.