I took 3 days of my holiday to deal with the problem that there is no easy step up converter to power a white LED on 1 AA cell.
Either you can build the SatCure circuit, which makes a lot of trouble winding the coil, or you can use a ready to go IC which can only be bought from the manufacturer and is SMD (hard to solder). Getting the coils for these circuits is hard also, because RadioShack and similiar shops dont have them.
Now I made a circuit which works only with resistors, capacitors, and easy-toget-and-cheap Transistors (BC547C and BC557C, both at about 10 cents apiece). The disadvantage of this circuit is that is relatively large (16 transistors!) and can power only one LED at 20mA, pulling 170mA out of the battery (not very good efficiency).
How it works:
1. An instable multivibrator generates a rectangular signal of about 10 kHz.
2. This signal drives a H-bridge of 4 Transistors to lower impedance.
3. This signal is sent through a voltage multiplying cascade of diodes and capacitors.
4. To minimize voltage drop across the diodes in the cascade, they are replaced by transistors (so the 0.6V voltage drop across the diodes are reduced to about 0.1 to 0.2 V).
I(LED)=20mA at Uin=1.500V
I(LED)=3mA at Uin=1.000V
I(batt)=170mA at Uin=1.500V
I(batt)=39mA at Uin=1.000V
The input voltage should not exceed 1.6V, because this will make the output voltage drop and the circuit will draw large currents from the battery.
If you have any questions, suggestions or improvements, please tell me.
20 mA is not the ultimate current you can run through a LED, but it is what if specified by the manufacturer. At this current the LED has no bluish tint. An increase up to 40 mA doesnt double the brightness, it is just an extra bit.
Besides, withe these cheap everywhere-to-get transistors, I wasnt able to get more current...
But I am working on an improved version perhaps with the little extra current.
The problem when stepping up 1.5V without an inductor is that you have no room for loosing 0.6V at a diode like in usual step up circuits using capacitors. With a coil, you can use the voltage spike generated when the current through the coil is suddenly turned off, it is many times the input voltage. When you turn the coil on, turn it off, use the spike to charge a cap through a diode, and turn it on again, it is really simple to build a step up converter.
When you use a cap, you dont have this voltage spike. You have to charge a cap, then change polarity somehow and then use this charged cap plus the input voltage to charge another cap at the doubled voltage. The problem is, usual circuits work with diodes do control the charging of the caps. But when you use only 1.5V the voltage drop is enourmous.
You charge one cap via a diode.
This make 1.5-0.6=0.9V on the cap.
Now you charge this cap plus the input voltage through another diode into a second cap: 0.9+1.5-0.6= 1.8 V !! The voltage gain is only 0.3V in a 2 stage voltage doubler!
Not to mention, what happens, when the battery drops to 1.4V or lower...
Ok, in the above described circuit I replaced the diodes by transistors which are current controlled by a low current flowing from the base to the emitter. This make the collector-emitter to conduct without the usual drop of 0.6V (in reality there are about 0.1 to 0.2V drop).
So the lower 2 parts of the circuit do the following:
They charge an AC (generated by the upper part) aternating on the 100yF capacitors, always adding the supply voltage.
The upper branch generates 3 times the input voltage over ground (three caps), the lower branch generates 3 times the input voltage below input voltage.
The outputs of these two branches (where the LED is connected) have (3+1.5V)+(3x1.5V)-1.5V = 7.5V when no load is present. When you connect a load, the voltage drops to about 3.5V at 20mA.
The problem is: To get 20mA flowing through the output (at virtually 4.5V in one branch) the first transisor has to switch 3x the current = 60mA. Due to the fact that the transistor is only switched on half the time, it has to switch 120mA to get 60 mA median current. Those BC types used are rated at 100mA continuous, 200mA peak current. So they are just a little bit below the limit at the moment.
I am working at a circuit using FETs to switch the current (These are kind of transistors, but they are not current but voltage controlled and do need virually no power to be turned on and off). The problem is I need about 3 volts to switch these, and I only have 1.5V, so I would have to generate an auxiliary voltage to control the FETs, but probably I will get a solution...
I hope, you understand the circuit a little bit more now (on first view it looks complex, but it isnt). If you have any questions, please tell me.
Phantomas2002: I just wanted to say that I really enjoyed your description of your circuit. Thanks for taking the time to share it and explain it. I'm going to be taking electronics classes in college soon so I'm trying to learn all I can beforehand. It's a really great idea to have an alternative to using a hard to find coil. In most applications that would use one single battery, space is a serious limitation. I worry that it would probably get difficult to place so many components in a very small space.
There are IC which do work as voltage doublers. A good example is the MAX660 from Maxim (www.maxim-ic.com). But they are hard to get (only from the manufacturer, not from radio shack) and only available in SMD.
But if you want to build a circuit with these, just call Maxim, and order two ICs. Two of these IC in serial would make a voltage step up by factor 4. Use a ow dropout voltage regulater behind them and you have the perfect 1 cell power source for a white LED.
Would you be interested in a circuit working with an easy-to-wind inductor (15 windings of 0.5mm copper wire on a simple ferrite core) and two or three resistors? As far as I saw the postings here, the Satcure circuit has a lot of windings on the ferrite core.
Phantomas2002: How efficient are IC step-ups like the one you mentioned? Are they much more efficient than a coil, much less efficient, or about the same? I think a circuit involving tiny IC's would be easy to put in a small space, but it would depend on how efficient they are as to whether it would be worth it to use them rather than some other design such as a coil.
At good circuit design you can get an efficiency of 90+ percent.
By the way, I think a really sophisticated flashlight should be made of metal, not plastic. My experience shows that all the Maglites are good to convert. The D size models have a lot of space in the head. You just have to get the switch out with a imbus key in the little hole in the switch. Then saw off the part which holds the lamp and you have plenty of space.
And besides, these MagLites look really good...
Several companies offer inductorless voltage multiplier ICs. You can select devices which multiply the input voltage by 1.5 or by 2. They work with several very small ceramic capacitors (low loss) at high frequency. Most of the modern devices (like nationals 335x series or the maxim 660) use carefully crafted internal MOSFET transistors to move electricity between the capactors with little loss. They easily reach 95% efficiency is used correctly.
Having said that - remember that the power to the LED will be 95% of the power from the battery. The current from the battery can then be calculated. So for example:
In a real circuit, you'd have to use a doubler, followed by a 1.5 or more multiplier to get the 3.4-3.6V needed for a white LED. In this case you'd waste some power in a series resistor. For example:
Doesn't that bug you? bugs me! [img]images/icons/mad.gif[/img] This brings the over all efficiency to (83-17-8)/83=about 70%. If you drop the series resistor - you won't win back much because the IC's will current limit and their efficiency goes down.
Annoying as inductors are - I'm working on "one cell - one bright LED" and "two cell dimmable - 8-12 LED" inductor based circuits aiming for 90% or more over all efficiency. My goal for the two cell is to be as bright as a trek 7 in a two AA compact light for 5 hours continuous burn, or 15hours at 50% dimmed, or 35 hours at 25% (90 hours at 10%, etc). It would have a "turbo" switch to overdrive the LEDs for 150% brightness at 2+ hours continuous burn. If I add a small $3 microcontroller I could add fancy auto off or strobe modes. I realize three cell designs are so much simpler, but I want a compact unit - and I want to be able to tailor the light output to the task - so I can trade light for battery life or vice-versa. One all purpose light that runs on cheap batteries - rather than needing many specialized lights. As long as I'm here - anyone have opinions about these tradeoffs?
<LI>Fancy auto-off & strobe modes or none?
<LI>continuous dimming versus set points of 150%, 100%, 75%, 50%, 25%, 10%.
<LI>Plastic versus aluminum for flashlight body (I'm going to convert an off the shelf unit).
<LI>Waterproof? Able to float?
<LI>I like two AA sized lights. Would a two "C" unit offer better optical characteristics (larger reflector?)
<LI>Push switch versus twist on?
What do you think?
This project may take a while as we're about to move....
interesting thread, all the more reason to believe my eternalight classic is my best all around light. question is, are those 4 leds burning at highest setting bright enough? they are for me and the many brightness levels/battery drain levels make it even more interesting as far as a "task" light. the pulse-width modulation is a small trade-off for me. if anyone can improve on the switching technology to eliminate the visible "pulsing" i'd be tempted to buy and compare [img]images/icons/smile.gif[/img]
php_44, I like the sounds of that. It sounds similar to the Eternal, but in a smaller package, maybe more conventional flashlight size? Put me on the list for one of them [img]images/icons/smile.gif[/img]
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by franken: ..if anyone can improve on the switching technology to eliminate the visible "pulsing" i'd be tempted to buy and compare [img]images/icons/smile.gif[/img]<HR></BLOCKQUOTE>
Try putting an electrolytic cap (+100uF?) in parallel with the LEDs. It might smooth out the pulses and it shouldn't hurt the eternalight to try. Just make sure you got the polarity right and the circuit is off when the cap is added/removed.
Of course, this will probally void the warranty, so don't try it if you wan't to keep that.