Three inductor-free circuits that allow a white LED or UV LED to be
driven
from one (1.5 volt) or two flashlight cells (3 volts total) are
described.
The circuits on this page are:
The two cell (3 volt) circuit four transistors.
The simple single cell (1.5 volt) circuit requires six transistors, but
it isn't
as bright as the two cell circuit.
The higher output single cell (1.5 volt) circuit requires 13
transistors, but it
is as bright as the two cell circuit.
A Simpler and more efficient circuit
that uses an inductor is described on
THIS PAGE.
(Above) A battery operated LED light, powered by a single carbon zinc
AA cell is shown above. It should be easy to improve upon the
industrial design of the black plastic case with a little work. With
an alkaline cell, battery life would be 20 to 30 hours. There is a
small ON/OFF slide switch on the top of the case.
right 2005 Richard Dwight
Cappels www.projects.cappels.org
(Above) Higher Output Single Cell (1.5 volt) Pulse Boost LED circuit in
an enclosure. In a world without
a enclosure tooling budget, one uses what one can find. This is quite a
bit brighter than the simpler single
cell circuit. This one has a small slide sitch which is mounted near
the battery, to turn it on and off.
The
problem with driving White and UV LED's
The
operating voltage of white LEDs coupled with my emotional
need for
low cost for a given function, motivates me to make LED drivers
suitable for battery operated lanterns that can run from a small number
of cells. White and Ultraviolet LEDs available at the time of this
writing, require
higher voltages than colored LED's. In the case of the ones I bought
most recently, from 2.8 to 4 volts, depending upon operating current
and the individual device.
A 006P 9 volt transistor radio battery can drive a 4 volt LED quite
easily if the correct resistor is placed in series. The 9 volt
transistor radio battery is
among the most expensive of the common primary batteries, and has a
rather low energy capacity, so it does not seem to be a practical
choice for my uses.
Primary batteries, such as alkaline and carbon zinc flashlight cells
are priced by the cell rather than the available energy. For example, a
flashlight D cell costs about twice what penlight AAA cell of
the
same brand and similar construction does, but the D cell is capable of
delivering much more than twice the energy, so the cost of light is
much lower
with
the D cell.
Unfortunately, three or more of these approximately 1.5 volt cells are
needed to directly light a white LED. This solution is good for some
applications, but is not desirable in all applications, such as
rechargeable lights, where one cell would be adequate and would
significantly lower cost. Another place multiple cells is a problem is
in compact lamps where the space is a problem.
To me, the AA cell seems to be a good tradeoff among size,
weight, energy capacity and cost. This was the reason behind my
first LED power supplies based on blocking oscillators. These work very
well and are pretty efficient -even the one that's built into the lamp
base and runs at a couple hundred kilohertz.
Many people seem to have trouble finding the parts to make the
inductor for the blocking oscillator. Ferrite cores are exotic to many,
and they seem to have become more
difficult to obtain in the last few years. Thus, I decided that this
project would not use any inductors, and would instead use readily
obtained components. The circuits described here can be from
simple discreet components, which means that the components will be
more
easily obtainable to some and it gives the builder an opportunity to
customize the design to her or his own needs.
The three white LED power supply circuits described here are based
on a simple pulse boost circuit. The single cell version was made from
the two cell version by adding an additional boost stage, and the
Higher Output Single Cell circuit was made by adding an additional
pulse
boost stage and some peak detectors to the LED is operated from direct
current. Though
it cannot be seen from the pictures below, the two cell version
produced a brighter light than the simple single cell circuit and the
Higher Output Single Cell circuit is at least as bright as the two cell
circuit.
Copyright
2005 Richard Dwight Cappels www.projects.cappels.org
(Above) A Two Cell (3 volt) Pulse Boot LED driver circuit uses four
transistors.
It was powered from a bench top power supply.
Copyright 2005 Richard Dwight
Cappels www.projects.cappels.org
(Above) Single Cell (1.5 volt) Pulse Boost LED driver circuit uses six
transistors.
The single cell board was made by adding two PNP transistors, three
resistors and
a capacitor to the Two Cell circuit.
(Above) Higher Output Single Cell (1.5 volt) Pulse Boost LED driver
circuit uses thirteen transistors
and is as bright as the 3 volt drive circuit. Some of the additional
transistors serve to increase the
efficiency of the voltage multiplier stages over that of the simpler
circuits. A word of caution about
soldering to button cells. I don't recommend it as they tend to
explode under the heat of the soldering
iron. It is far better to use a holder. When I soldered this one, I
used a hot, well-tinned iron, lots of
active flux, and let the cell cool between steps in the soldering
process. I
wore safety goggles over my eyeglasses.
After the picture was taken, I connected the circuit to an AA cell
battery and put it in the frosty plastic
enclosure shown near the top of this page.
The
Circuits
I'll start with the two cell version since it is the most fundamental
version of the circuit. See the schematic below.
The three volt cicuit uses
few parts but requires two cells.
The circuit is composed of two sections. In the schematic above: A
multivibrator on the left and a pulse boost stage on the right.
The mulitivibrator is the simplest discrete transistor oscillator I
could come up with that will run reliably below 1 volt and provides
square wave outputs of opposite phase. Since this circuit has been more
than adequately explained in many publications and web pages, I'll just
say that when Q1 turns on, the negative going collector pulse
turns off Q2 until the base capacitor (C1) charges up to where Q2 turns
on, which turns Q1 off until C1 charges to the point that Q1 comes on
again. Thus, by varying the base resistors (R2 and R3) and the base
capacitors (C1 and C2) the frequency is varied. The frequency is also
directly proportional to the battery voltage minus the base-emitter
voltage, but the operating frequency is not very important. It just has
to be high enough so that the boost capacitor, C3 is a low impedance at
that frequency.
The boost circuit, consisting of Q4, Q4, and their associated
components, charges up boost capacitor C3 and then switches C3 to
be in series with the LED and its dropping resistor, R6, and the
battery voltage. Ideally, it would put a pulse of two times the battery
voltage across the LED and its resistor, but we live in a real world
and the voltage comes out less that that.
In operation, the base of Q4 is switched by a square wave that is an
inverted version of the square that switches Q3, so when Q4 is on, Q3
in off, and when Q3 is on, Q4 is off.
When Q4 is on, the negative end of C3 is held at ground and since Q3 is
off at that time, C3 charges through R6. On the next half cycle from
the oscillator, Q4 turns off and Q3 turns on, switching the positive
end of C3 from about +3 volts to ground. This makes the negative end of
C1, and therefore the cathode of the LED go to - 3 volts. Since the
anode of the LED is connected to + 3 volts through R7, the LED and its
dropping resistor have 6 volts across it.
In real life, the voltage isn't that high because of circuit losses,
but its high enough to drive the LED.
I bought the white LED used in this circuit from
www.besthongkong.com.
The LED is their part #BUWLC333W20BA13. In the two cell (3 volt)
circuit I built, the LED
was driven to a little less than 30 milliamps peak when the circuit was
supplied by a
3.0 volt power supply. Remember that this was at a 50% duty cycle, to
the brightness would appear to be that of the LED being driven at half
that current (which , because of the eyes' logarithmic response, looks
almost the same!). The input current to the circuit was about 50
milliamps DC and the oscillator ran at about 50 kHz. The circuit
continued to light the LED as I cranked the power supply voltage down,
until it reached about 2.0 volts. In the single cell (1.5 volt)
version, the LED was illuminated when the power supply voltage was
cranked down to about 1.08 volts.
There are a couple of points concerning the operation of Q4 that
warrant some discussion.
Q4 could be replaced with a resistor, or a diode with the cathode
grounded, but the output voltage would be lower. Of course, one
possible advantage of a diode is that it does not need base drive, so
an oscillator with a single output could be used.
When the collector of Q4 is driven negative, current is drawn through
R8. This results in a few milliwatts being lost. If the battery voltage
were to be a lot higher, say more than 7 or 8 volts for a 2N4401, the
base-emitter junction would be reverse biased into avalanche. At that
point, a lot more current would be drawn and that could affect
reliability as well as efficiency, so this circuit cannot be extended
to much higher voltages without modification.
The component values are not critical. I used 2N4401's and 2N2907's
because they were handy, but other small signal transistor such as the
2N2222, 2N4124 and 2N3904 can be used for the NPN. Similar
substitutions can be made for the PNP -2N3906, 2N4126, etc.
I imagine that the passive components can vary in value by 2:1 or more
without much affect on performance. I you don't have these
particular values, then experiment with what you have available.
Just
keep the value of R1 = R3, R2 = R4, C1 = C2, and C5 = R6 = R9 =
R10 to keep the square wave near 50% duty cycle. Oscillator frequency
is not critical, but for frequencies below about 30 kHz, you should
think about making C3 and C4 larger. C3 and C4 can be a larger
value without significant
effect on performance.
None of the circuits shown on this page have decoupling capacitors
across the power source. These circuits rely on the low output
impedance of the battery cells to supply peak currents to the circuit.
In some cases there might be an improvement in performance if a
capacitor is placed from the + power supply after the switch to the
negative battery terminal.
Copyright
2005 Richard Dwight Cappels
This single cell version is not
very bright, but then again, it
doesn't use many parts, either.
The single cell circuit, shown above, is the same as the two cell
circuit, except that R7 was changed to 22 Ohms, a second pulse boost
circuit was added in the form of Q5 and Q6 along with their associated
components. With the LED I used, the current was mostly a function of
the voltage across the LED, and the 22 Ohm resistor had little effect
at that current. The 22 Ohm resistor could be omitted, but in earlier
training I was taught that when you have a direct path through
semiconductors from the power supply to ground, the chances of smoke
increase a lot, so the 22 Ohm resistor stays in my version.
The second pulse boost circuit operates the same way as
that made form Q3 and Q4, but instead of using NPN transistors to make
a negative going pulse
on the cathode of the LED with respect to ground, this one uses PNP
transistors to make a
positive going pulse on R7 with respect to the battery's positive
terminal voltage. This means that ideally, R7 will be switched to a
positive voltage that is two times the battery voltage with respect to
ground and at the same time, the cathode of the LED is switched to
negative 1 times the battery voltage with respect to ground. That would
be 4.5 volts peak voltage, but in my real world version it was only 4
volts peak.
In my circuit, the LED drew 14.4 milliamps peak, at about a 40% duty
cycle when the input voltage was 1.525 volts. The input current ot the
circuit was 34.2 milliamps. The oscillator ran at 70.4 kHz, and the LED
was extinguished when the input voltage dropped below about 1.08 volts.
This single cell circuit uses a
lot of parts, but the result is a fairly bright light.
A Higher Output Single Cell (1.5 volt) inductor-free version was also
designed. See the schematic above. It needs 13 transistors to
obtain significantly brighter and much more
efficient than the simpler, six transistor single cell circuit, but at
a similar efficiency.
Since building the first single cell inductor-free version, I though
about the power wasted in R6 and R9, and realized that in order to make
the LED brighter, the circuit would have to be able to supply even more
voltage, and running the LED from DC instead of pulsed current would
also make it brighter if the DC level is higher than the averaged
pulsed current, so I chose circuit modifications that would accomplish
these functions and that resulted in the schematic above. Yes, its a
lot of transistors, but transistors are inexpensive and a lot easier
for some to obtain than ferrite cores and other types of inductors. If
built on a surface mount PCB, even this 13 transistor circuit could be
quite small.
For the sake of power efficiency, resistor R6 from the earlier version
is replaced
by an active switch, Q3 in the higher output version, and R9 from the
earlier version is replaced by Q8 in the higher output version. With
transistors instead of resistors, only the current needed to charge C3
and C4 passes through the transistors. When Q4 is conducting, Q3 is
off, saving power over the older circuit in which Q4 would have shunted
current through its collector resistor (R6 in the old circuit) to
ground. Similarly, Q8 only charges C24 and switches off when Q7 turns
on.
The input current to this power supply is 49 milliamps x 1.415 volts =
69 milliwatts, while the output power is 3.49 volts X 8.2 millimaps =
28 milliwatts, so the efficiency could be stated as 40%. This is
difficult to compare with the old circuit, which provides pulsed
current to the LED because I don't have enough data to accurately
calculate the older version's efficiency (the older version and I are
on
different continents at the moment), but based an estimated voltage
drop across the LED, the effiecncies are roughly the same. This
indicates an increase in efficiency in the first two multiplier stages
(Q3 through Q8) since the efficiency is about the same, in spite of the
addition of Q9 through Q13 in series with the load.
To get a little more voltage, an additional boost stage was added. This
is made of Q9, Q10, and Q11 along with their associated parts. Ignoring
losses, the maximum available output voltage of this circuit is the
cell
voltage + voltage across C3 + voltage across C4 + voltage across C5,
which would equal 4 times the cell voltage, or about 6 volts DC. Since
the voltage across the capacitors droops a little bit and there are
saturation losses in the transistors, the maximum unloaded output
voltage of the one I built was 4.95 volts when the input voltage was
1.415 volts, or 3.5 times the cell voltage. When the LED loads the
supply with 8.2 milliamps DC, the voltage across C6 drops to 3.49 VDC.
The last two transistors in the circuit, Q12 and Q13 are low voltage
drop rectifiers. Q13 peak detects the pulses from C4, which go below
ground, and Q12 peak detects the pulses resulting from C3 and C5 being
put in series with the cell, which is above ground. The peak voltage is
stored in C6. The reason the transistors have such low voltage drops is
that when, using Q12 as an example, the voltage on the emitter of Q12
goes more than a base-emitter junction voltage above the cell voltage,
base current is drawn through Q12, causing Q12 to saturate. The
saturation voltage of Q12 is much lower than the forward drop of a
rectifier would be. In this case, the forward voltage is only about 40
to 50 millivolts. Larger area transistors would lower this drop.
Construction
of the simple single cell (1.5 volt) version
Click the image above for a larger one.
(Above) A view of the single cell (1.5 volt) version of the circuit
sitting inside the plastic box. The left-most two transistors in the
top row are the multivibrator transistors. The right-most transistors
in the top row are the NPN pulse boost transistors. The transistors in
the lower-right of
the board are the PNP boost circuit added to make the two cell (3 volt)
version
into the one cell (1.5 volt) version. The yellow "blobs" are the 0.33
uf tantalum capacitors. The board and AA cell will be held in place by
a
piece of foam rubber that is compressed between the components and the
lid of the plastic box. In this picture, the LED is on. And yes,
that is Thai script on the Eveready AA cell.
I built mine on punched phenolic board and just connected the leads of
the components and soldered them together. I ran insulated
jumpers where wires had to cross over. The circuit is pretty
tolerant of layout, as you can see from the picture below.
Click the image above for a larger one.
(Above) The circuit along the top is the ground bus, and along the
bottom is the +1.5 volt bus. The LED is to the right. I do not
recommend this be taken as good construction practice
-it is barely suitable for the quickest breadboard, though with luck,
it will continue to work for many years without problems. Note from the
residue on the board that this was soldered with a liberal amount of
neutral PH flux. With some care while placing the components on the
board, I managed to get by with only three jumpers. The LED cathode
lead was insulated with a piece of paper when I closed the circuit up
in its case.
Its easy to miswire the circuit, particularly if you crowd the
components together and are hand wiring it. I built the multivibrator
first and saw that it was running, then added the NPN pulse boost
circuit and saw that it was running, but using a 3 volt power supply
and connecting the LED, then I built on the PNP pulse boost circuit and
reconnected the LED and powered it up with a 1.5 volt AA cell.
Seeing the LED shine brightly when I touched the wires to the AA cell
was a thrill.
A hole was drilled and hand filed in the black plastic enclousre for
the LED and another hole was
drilled and hand
filed for a slide switch. Both were press fitted into their holes -no
glue here. The battery is an AA carbon-zinc cell and it was soldered to
wires going to the switch and the circuit board, and another wire was
soldered from
the circuit board to the switch, then the cover plate was screwed into
place. Nothing fancy, just a small book light for now.
Construction
of the Higher Output Single cell (1.5 volt) version
I hand wired these components on a piece p;re-etched of pad-pre hole
circuit board. This resulted in a much more solid assemly than the
other one, where component leads were just pushed through holes and
connected together becuase in this one, the leads were soldered to pads
on the board. Holes for the LED and on/off switch were cut into the
frosted plastic box with an X-Acto knife after starting the holes with
a soldering iron. The holes were made so that the LED and switch could
be pressed into the hole and held in place without glue or screws.
Unfortunately, there was no room for a battery holder, so the battery
and swtich are soldered to each other and to the circuit board. This
should be no problem though, because the battery won't have to be
changed for a good long time.
Danger of Eye
Damage From Visible Light Emitting Diodes
HOME
(More Projects including LED circuits)
Contents ©2005 Richard Cappels All Rights Reserved.
http://www.projects.cappels.org/
First posted in July, 2005. Revised
August, 2005, December 2006.