DMF boost circuit.

September 4, 2009

One of the ideas I've been playing around with for a few years is how to run a white or blue LED off of one or two AA batteries. The problem is that a single cell battery only provides 1.2 - 1.4 volts, but white and blue LEDs require 3.3 - 3.7 volts. What I needed was someway to boost the voltage. Now there are lots of ways to boost a voltage and I've tried a few of them. I'd like to come back and discuss this more and provide some links. Currently my favorite (and newest) solution is utilizing an ON semiconductor NCP1402 IC and their basic application circuit. The NCP1402 is a 200mA Pulse Frequency Modulation Step-up micropower switching regulator. It comes in a couple different models, depending what you want for output: 1.9V, 2.7V, 3.0V, 3.3V, 4.0V and 5.0V. The IC will startup with an input voltage of .8V and operate down to .3V, so you can easily power this with a single cell battery like an AA and get out 4V. The circuit is really just a modified boost circuit where the NCP1402 does all the hard work. There are lots of other ICs out there for making boost circuits, but they rarely accomodate such low input voltages. The one issue I did have with the NCP1402 is that it only came in a surface mount SOT23-5 package, which I consider to a VERY small package and difficult for a home hobbyist to work with. I decided to give it a try though and was pleasantly surprised with the results.

There are commercial pre-built circuits that do the same thing that my circuit will do and they were also an inspiration to create my own. Bodhilabs.com sells circuits they call Vpacks with 3.0 and 5.0V outputs. You can see a picture of their pcb on the website and it looks a bit different than the circuit I built. But it's an easy way to get a pretty cheap prebuilt boost circuit like I describe on this page.

Let's get to the details!

Parts List

  • IC1 - NCP1402 (Free samples from ON website, or $.80 from Digikey)
  • L1, Inductor, 47uH, .72A (Digikey part: 513-1073-1-ND, $1.00)
  • D1, Schottkey diode, 20V, .5A (MBR0520LT1, free samples from ON, or $.55 from Digikey)
  • C1, Input capacitor, low profile tantalum, 10uF, 16V (Digikey: 718-1320-1-ND, $.31)
  • C2, Output capacitor, low profile tantalum, 68uF, 10V (Digikey: 718-1485-1-ND, $.44)

    So you can see that overall the parts for the circuit add up to under $4.00. The real cost is in the time it takes to create the printed circuit boards. Because all of the parts we're using are surface mount and it's a simple circuit, we were able to layout the circuit board using just the top layer (no need to drill and holes for thru-hole parts). I use the freeware version of Eagle CAD for drawing schematics and laying out circuit boards. In this case I had to create a part for the NCP1402 because I couldn't find it in any libraries. Here's the schematic and the PCB layout.

    If you'd like to use or modify the circuit I've already created, it's available as an Eagle project in the the file dmf_booster.zip. This file also contains the Eagle library 1dmf.lbr that contains some necessary parts for the schematic and pcb. I believe you need at least version 4.5-ish of Eagle to use these project files.

    My method of etching a custom printed circuit board is described in my webpage on How to etch PCBs at home. The end result was this circuit board.

    As you can see, the first time I etched the circuit I really wasn't paying attention. I had forgotten to print and etch the mirrored image of the PCB, so everything was backwards. This almost still worked, except that the NCP1402 chip pinout wouldn't work backwards. I was pleasantly surprised how well the PCB came out, considering that my etching method doesn't always give the most precise lines and I'm working with surface mount parts.

    Now it was simply a matter of populating the board. This involved soldering all the tiny SMD parts. It actually wasn't as bad as I was expecting. I pre-soldered all the pads on the PCB and used some tweezers to help hold the parts in place while I quickly tacked them to the board. Here's the end product.

    I did a quick test by soldering a 1xAA battery holder to the inputs of the board and then measuring the output voltage. I had used the 4V NCP1402 circuit and the measured output voltage was 4.02V. Excellent! So now another quick test; I attached a white LED and small current limiting resistor to the output and it worked great. The output was drawing ~.020A and the input was supplying ~.052A. A quick shot of the circuit driving the white LED from a single AA battery.

    The circuit is quick small in area, but could be made even smaller by modifying the layout slightly and trimming more off the edges.

    September 24, 2009 - I shrunk the PCB down to a little more than the size of a dime and am driving a Cree XLamp XR-E white LED in a standard 2 D-cell flashlight. The Cree in nominally rated at 3.7V at 1A (it's really bright), but the forward voltage drops as the current drops: at 750mA it's 3.5V, at 350mA it's 3.3V, etc. On page 7 of the XR-E spec sheet is a graph of the forward current vs forward voltage. Since the NCP1402 can't supply more than 200mA, I extrapolated from the graph that the LED voltage would be closer to 3V. That means I can use the 3.3V version of the NCP1402 and conserve power.


    October 2009 - I shrunk the PCB down even further and have succeeded in modifying a regular flashlight bulb with a high-power LED and booster circuit in a 2-D cell maglight. Essentially I broke the glass off a flashlight bulb and added a super smushed booster circuit PCB on top of it and then a .5" diameter high-power LED. (I'm writing this in 2012 and I've forgotten exactly what model LED it was, but I got it from www.dx.com.) I think the pictures say it all. After installing the modded bulb in the maglite, it works great!








    I can be reached at