26 February 2010

Drive 12 LED$ with one I/O line

Few Design Ideas expand the I/O of a pin-limited microcontroller (references 1 through 4). The circuit in this Design Idea uses an alternative method (Figure 1). It limits external additional parts to one IC, and it drives as many as 12 LEDs in dot-bar or bar-graph mode. You can use the same technique in a dot-bar design
(Figure 2). If you need a seven-segment LED display, you can use the circuit in Figure 3, which shows how to rearrange the circuit according to a classic multiplexed, four-digit common-cathode display. The prototype display uses Kingbright’s (www.kingbright-led.com) SC52-11EWA high-efficiency LEDs, which emit 2000 to 5600 mcd at a forward current of 10 mA. The driver is a 12-stage NXP (www.nxp.com) 74HCT4040 binary counter or a 74HC4040 version for a lower power supply.Listing 1, which you can download at the online version of this Design Idea at www.edn.com/100204dib, contains an assembly-language routine.It generates a precise quantity, Q, of high-frequency pulses, which deliver the number, N, that the outputsof the 74HCT4040 require. The relations are Q52N21 in dot-bar mode and Q52N21 in bar-graph mode.    Listdesignideas

 which is also available at www.edn.com/100204dib, is a full example of using this routine with Microchip’s (www.microchip.com) PIC10F202, a member of the PIC10F series, which is the company’s mostpin-limited microcontrollerfamily.Although the PIC’s internal unique clock frequency is 4 MHz, you’ll noticelittle flicker effect. You can reducethe flicker by using a midrange pin-limited PIC microcontroller, such as the PIC12F629, which has an internal.
clock frequency of 20 MHz. Listing 3, also available at www.edn.com/100204dib, uses a look-up table to convert the desired number into seven-segment code to replace the 12 LEDs with a four-digit display.EDN References Anonymous, “Microcontroller provides low-cost analog-to-digital conversion, drives seven-segment displays,” EDN, May 10, 2007, pg 80, www.edn.com/article/CA6437954. Raynus, Abel, “Squeeze extra outputs from a pin-limited microcontroller,” EDN, Aug 4, 2005, pg 96, www.edn.com/article/CA629311. Jayapal, R, PhD, “Microcontroller’s single I/O-port line drives a bar-graph display,” EDN, July 6, 2006, p 90, www.edn.com/article/CA6347254. Lekic, Nedjeljko, and Zoran Mijanovic,“Three microcontroller ports drive 12 LEDs,” EDN, Dec 15, 2006, pg 67, www.edn.com/article/CA6399101.

25 February 2010

Wind Power & Solar Power (Info)


Wind is the movement of air from an area of high pressure to an area of low pressure. In fact, wind exists because the sun unevenly heats the surface of the Earth. As hot air rises, cooler air moves in to fill the void. As long as the sun shines, the wind will blow. And as long as the wind blows, people will harness it to power their lives.
Ancient mariners used sails to capture the wind and explore the world. Farmers once used windmills to grind their grains and pump water. Today, more and more people are using wind turbines to wring electricity from the breeze. Over the past decade, wind turbine use has increased at more than 25 percent a year. Still, it only provides a small fraction of the world's energy.
Most wind energy comes from turbines that can be as tall as a 20-story building and have three 200-foot-long (60-meter-long) blades. These contraptions look like giant airplane propellers on a stick. The wind spins the blades, which turn a shaft connected to a generator that produces electricity. Other turbines work the same way, but the turbine is on a vertical axis and the blades look like a giant egg beater.
The biggest wind turbines generate enough electricity to supply about 600 U.S. homes. Wind farms have tens and sometimes hundreds of these turbines lined up together in particularly windy spots, like along a ridge. Smaller turbines erected in a backyard can produce enough electricity for a single home or small business.
Wind is a clean source of renewable energy that produces no air or water pollution. And since the wind is free, operational costs are nearly zero once a turbine is erected. Mass production and technology advances are making turbines cheaper, and many governments offer tax incentives to spur wind-energy development.
Some people think wind turbines are ugly and complain about the noise the machines make. The slowly rotating blades can also kill birds and bats, but not nearly as many as cars, power lines, and high-rise buildings do. The wind is also variable: If it's not blowing, there's no electricity generated.
Nevertheless, the wind energy industry is booming. Globally, generation more than quadrupled between 2000 and 2006. At the end of last year, global capacity was more than 70,000 megawatts. In the energy-hungry United States, a single megawatt is enough electricity to power about 250 homes. Germany has the most installed wind energy capacity, followed by Spain, the United States, India, and Denmark. Development is also fast growing in France and China.
Industry experts predict that if this pace of growth continues, by 2050 the answer to one third of the world's electricity needs will be found blowing in the wind

Every hour the sun beams onto Earth more than enough energy to satisfy global energy needs for an entire year. Solar energy is the technology used to harness the sun's energy and make it useable. Today, the technology produces less than one tenth of one percent of global energy demand.
Many people are familiar with so-called photovoltaic cells, or solar panels, found on things like spacecraft, rooftops, and handheld calculators. The cells are made of semiconductor materials like those found in computer chips. When sunlight hits the cells, it knocks electrons loose from their atoms. As the electrons flow through the cell, they generate electricity.
On a much larger scale, solar thermal power plants employ various techniques to concentrate the sun's energy as a heat source. The heat is then used to boil water to drive a steam turbine that generates electricity in much the same fashion as coal and nuclear power plants, supplying electricity for thousands of people.
In one technique, long troughs of U-shaped mirrors focus sunlight on a pipe of oil that runs through the middle. The hot oil then boils water for electricity generation. Another technique uses moveable mirrors to focus the sun's rays on a collector tower, where a receiver sits. Molten salt flowing through the receiver is heated to run a generator.
Other solar technologies are passive. For example, big windows placed on the sunny side of a building allow sunlight to heat-absorbent materials on the floor and walls. These surfaces then release the heat at night to keep the building warm. Similarly, absorbent plates on a roof can heat liquid in tubes that supply a house with hot water.
Solar energy is lauded as an inexhaustible fuel source that is pollution and often noise free. The technology is also versatile. For example, solar cells generate energy for far-out places like satellites in Earth orbit and cabins deep in the Rocky Mountains as easily as they can power downtown buildings and futuristic cars.
But solar energy doesn't work at night without a storage device such as a battery, and cloudy weather can make the technology unreliable during the day. Solar technologies are also very expensive and require a lot of land area to collect the sun's energy at rates useful to lots of people.
Despite the drawbacks, solar energy use has surged at about 20 percent a year over the past 15 years, thanks to rapidly falling prices and gains in efficiency. Japan, Germany, and the United States are major markets for solar cells. With tax incentives, solar electricity can often pay for itself in five to ten years.

LED Bulbs(Energy Savers)


17 February 2010

Simple LED Panel Lamp

LED panel lamp:

LED panel lamp
Well, yesterday I got attack of conscience and decide to make something local to improve world power shortage. I rebuild old halogen 12W lamp in my table lamp on modern LED light.

Halogen 10W:

Halogen 10W
This was reason which push me to work. I have electric power AC220V and it's transformed to plain AC12V which drive this lamp with 1A...o light and heat with 12 W. Very hot!


You can see what heat of 12W do... plastic is melted. Well light color is warm about 3000K. Its not day light which we like much.

Eternal plexi:

Eternal plexi
I took 5mm Plexiglas 120x80mm, drill holes for LED foots. Also I shall rectify AC to DC current.


LED China. 11mm dia. 3.6V and 20mA. Each.And I have 12VDC output.What to do?
I shall connect LED serial !-so 4 leds shall easy hold 14.4V! In my case 12V divide with 4 is 3V. Less. OK.


I took alcoholic silver paint pen to paint bottom of each led. Do you know why?  To increase light in front direction. My moto: keep it simple.


White wires are AC12V from transformer. AC is converted in DC with Graetz diodes, full wave, After that is 8 lines of 4 LEDs ...voila!

Be light!

Be light!
Switch on! ...nothing burned...it works! Nothing is warm...even 4 Graetz-diodes (they are 1A) are cold.
Instead 2W I have output 12VX15mAX8=cca 1,5W... or so.. 

And light is on!

And light is on!
I put on back(glued) 6mm sponge. (roll sponge for laying on pool...). YOU can put a tape. Or card paper. Or something else --- what you like. Just for better look and some protection. And
of course: WARNING. Do this on own risk. Not mine.

13 February 2010

BloodBowl Turn Counter Using 7-segment LEDs

BloodBowl Turn Counter Using 7-segment LEDs:

BloodBowl Turn Counter Using 7-segment LEDs
This project was for a BloodBowl game turn counter using six Charlieplexed 7-segment LEDs.


A friend of mine asked me about ideas for building Bloodbowl Turn counter for his boardgame. Not knowing what this was, and what he wanted, it took awhile to decide on if and how I was going to do this.

I first had to have an idea of what he wanted, so I started with concept art (picture). The basic idea is to have 3 push buttons, controlling 3 LED's each and it would be placed inside a custom built tower.

The only big request was to have the top 4 displays count up from 0 to 8 and reset, and have the lower 2 displays count down from 8 to 0 and cycle back.

I would complete the circuit, and he would complete the tower.

Design & Parts List:

Design & Parts List
Since the concept called for 6 7-segment LED's, and I had some 8-bit Microchip PICs handy, I researched ways of using the PICs to control LEDs.

" Up to 6 displays can be accessed like this without the brightness of each display being affected." I considered this a challenge and something to investigate as part of my project.

The First thing I did, was grab some incandescent 7-segment displays from my box and see how they would work. Bad news. The particular parts I selected were not behaving like I wanted. The segment would light when needed, on the breadboard, but leakage current was distributed to the other 6 segments. I realized incandescent displays may not be the way to go, or I needed to use them in a different way. So for simplicity I verified the 7-segment LEDs I had on hand would work for breadboarding, and ordered some common anode displays.

The Second thing I needed to do was layout my design and start work on the code. Pictured is my circuit. Not much to it, as the code in the PIC takes care of the multiplexing...errr Charlieplexing. Note: ALL 6 displays have the SAME lines from the driver IC. The selector IC enables each display, 1 at a time, and the 7-segment lines are updated by the PIC accordingly. Very simple idea.

After that, code and hardware completion is all that was needed.

Parts List
After 3 small orders from Digi-Key while deciding on specific components, I had everything I needed (with some stuff on hand);
1 ~3"x4" PCB
6 small push button switches (N.O.)
1 74LS47 , 7-segment display IC
1 PIC16F627
1 CD4028 , 1 of 10 selector IC
6 10KOhm resistors
1 470Ohm resistor
1 spool of wire. I used various colors and guages, but that was just me.
1 78L05 5V regulator
1 9V battery clip
1 9V battery
1 small switch (for power on/off)

I consider this a moderately complex project, due to;
1) Microprocessor code required
2) Soldering and breadboarding
3) Design optimization.

None of these issues by themselves are overly complicated, but taking them all on without any experience can be abit much for the beginner. A hardware programmer is required to burn the device, soldering station, etc...

The FIRST thng someone might notice is that the 7-segment LED's DO NOT have series (current limiting) resistors! Let me address that quickly, by stating my original design had them in...but read the next step for explanation!

PCB Soldering:

Everytime I get to this point in my project I delay abit. At first I was going to wire wrap this thing, but quickly dropped that idea.

At first I think "A few wires to solder, no big deal"...then, by the time my project is ready to be soldered I am thinking, " I should have either sent out to have a proto board made, or etched my own board".

I am not into PCB etching (yet), and didn't want to pay $$ to have a board made, so....

Yeah.....I spent about 3 hrs soldering this thing. It's about 150 wires, so that's 300 solder points, plus touch-ups for solder bridges.

Anyway, here's the back side of the board pictured....yeah...abit of a mess, but when it was all done I only had 1 solder short. Took 20 mins of thinking since the display showed the wrong #'s being displayed in a logical pattern that I had to decipher. After that, I located the short, and bam! It worked perfectly.
PCB Soldering



This project took about;
~2 weeks to think about and email fine points to requestor,
~3 hrs of code completion and debug,
~4 hrs of breadboarding and debug,
~3 hrs of soldering

Using just 3 IC's it is possible to Charlieplex 6 7-segment LEDs.
Power draw is at about 30mA with this design, which isn't bad if I do say so myself.

I suspect more 7-segment LED's could be used, but have not pushed the envelope.

This idea could be applied to almost ANY application using 7-segment LEDs; thermometer, clock, text display, etc. With some tricky code, you could have a moving display, or pictures...maybe even a base for a POV (persistence of vision) project.

The final implementation is left for my friend to build his Tower and place the board in, as he see's fit. If/When that is done, I will get a picture uploaded. But as for the circuit, this appears to be built to order!

12 February 2010

LED Flasher Circuit

Intro41 LED Flasher Circuit using 555 IC :

41 LED Flasher Circuit using 555 IC
I made this as a quick project I made to use a lot of the LEDs I recently got. It basically connects via a 555 8 pin IC and allows for adjusting the time between the flashings by changing the resistor or capacitor values. It provides for a cool looking effect in a dark room. Use your favorite color LEDs and enjoy Now!

Step 1Parts Required:

Okay, so let's beign.
Here is what you will need:

1 - Timer 555 8 pin IC
1 - 2N3905 PNP general switching transistor
1 - 2N3053 NPN general purpose amplifier (I dont know if this is a switching or amplifer but I used MPSA2222A instead and it worked fine, I also tried 2N3904 and it worked, but a littler worse than the one I used)
20 - red LEDs (although you can use any color choice you want)
20 - blue LEDs (although you can use any color choice you want)
1 - LED (this is used to verify if your circuit works, can be any color)
1 - 1uF Electrolytic capacitor
1 - .1uF disc capacitor
1 - 150k resistor
1 - 4.7k resistor
1 - 160 ohm resistor
1 - 220 ohm resistor (although I used a 160 ohm)
20 - 100 ohm resistors (I did not have 20 so I used 10 of 100 ohm and 10 of 120 ohm)
1 - 6V source wire

Step 2Make the circuit:

Build the circuit
So here is the layout of the circuit. It is pretty basic. Here are some notes so you can better understand what is going on.

The notes were taken from the page mentioned later.
"Two sets of 20 LEDs will alternately flash at approximately 4.7 cycles per second using RC values shown (4.7K for R1, 150K for R2 and a 1uF capacitor). Time intervals for the two lamps are about 107 milliseconds (T1, upper LEDs) and 104 milliseconds (T2 lower LEDs). Two transistors are used to provide additional current beyond the 200 mA limit of the 555 timer. A single LED is placed in series with the base of the PNP transistor so that the lower 20 LEDs turn off when the 555 output goes high during the T1 time interval. The high output level of the 555 timer is 1.7 volts less than the supply voltage. Adding the LED increases the forward voltage required for the PNP transistor to about 2.7 volts so that the 1.7 volt difference from supply to the output is insufficient to turn on the transistor. Each LED is supplied with about 20 mA of current for a total of 220 mA. The circuit should work with additional LEDs up to about 40 for each group, or 81 total. The circuit will also work with fewer LEDs so it could be assembled and tested with just 5 LEDs (two groups of two plus one) before adding the others."

Step 3Lets Start Modify the project:

Modify the final project
Here is the final product of mine. I have added a switch but that is not mandatory.
Okay, now editing the flashing rate is easy. You only need to change 1 or more of the 3 values of R1, R2, and C.
Use these equations to do so.
Positive Time Interval (T1) = 0.693 * (R1+R2) * C (time first set is flashing)
Negative Time Interval (T2) = 0.693 * R2 * C (time other set is flashing)
Frequency = 1.44 / ( (R1+R2+R2) * C) (flashes per second)

Signal Generator Circuit

                                                  :SIGNAL GENERATOR :
This is a lovely project for home constructors keen on building their own test and measurement (T&M) equipment. It makes use of the ICL8038 signal generator chip, manufactured by Intersil. An improved version, made by Exar corp. is available (XR8038A). It can be used to produce three types of waveforms, sine, square and triangle. The frequency, amplitude and duty cycle can be varied, and selection of waveform is done digitally. To further reduce the complexity, a 3-to-1 switch may be used in place of the digital selection circuitry. I made use of the digital selection mechanism because switches available on the market are prone to dirt accumulation and poor contact quality. Besides, the digital method is a lot cooler!

The circuit shown below can be roughly divided into three parts: the oscillator based around the ICL8038 chip, the selection logic based on the CD4017 and CD4066 and the offset generation and output buffers, based on the LF412. Apologies for the cramped schematic, I had to keep the image size small, and the width under 640 pixels!
The oscillator is a standard 8038-based oscillator circuit, taken from the ICL8038 datasheet. The timing resistor chosen is rather small, to give a wide range of frequencies. This range might be a little too large, making precise frequency setting difficult. In that case, the freqency range may be split into two parts, using two capacitors which can be switched using an SPDT switch. Note that the 8038 is powered from a split supply, not a single supply, to generate a symmetrical waveform without the need for capacitor coupling. Two sine wave adjustment terminals (Pins 1 and 12) are provided, however only one is used. This gives a sinewave distortion of about 1%. To achieve better distortion figures, the circuit shown in Figure 4 of the ICL8038 datasheet may be used. The 8038 is powered from slightly less than +8V to allow the tuning voltage to go above the supply rail. This allows for maximum sweep range (1000:1), however the output waveform tends to be slightly asymmetric because of this. This may be compensated using the offset control R10. R2 controls the duty cycle of the oscillator. R7 acts as the sinewave distortion adjustment. The square wave output of the 8038 is an open-collector output. Hence, a 1k pullup resistor is provided. The sine and triangle outputs are about 5Vpp, while the square wave is 16Vpp. Hence to equalize the different outputs, the square wave is attenuated using a fixed attenuator formed by R6 and R9. A 47k pot may be substituted to make the attenuation level adjustable.

One of the three outputs are then selected using the digital selection circuit. This is based on the CD4066 quad analog bilateral switch and the CD4017 decade counter with fully decoded outputs. I havent used the 4051-series Analog MUXes here because getting a counter to drive this is a bit of a pain. The 4066 essentially acts like four switches under electronic control. The voltages switched must not be above or below the supply voltages. Hence, the 4066 is given both +8 and -8. Now, however, the selection inputs to the 4066 cannot be with reference to ground, and since the selection inputs are driven by the 4017, it too must have a split supply. Hence, note that both chips have ground connected to -8V and not 0V. The 4017 is wired so that it resets itself when the fifth state is selected. The clock to the 4017 comes from a pushbutton switch with a pullup to +8. A 0.1μ capacitor across the switch serves as a contact debouncer. This prevents multiple triggering of the 4017 from a single keypress. Hence, each output is successively chosen when the switch is pressed. The fourth state corresponds to the output being off (DC). On the fifth press, the counter resets itself and selects sine wave again.
The final stage is the offset adjust, amplitude adjust and output driver stage. The offset adjust is an inverting amplifier whose reference pin is not at ground. It is instead attached to a 1k pot's wiper, connected across the supply rails. Hence, when the reference pin is taken off ground, the amplifier introduces a DC shift corresponding to the product of the gain and the reference voltage. The gain of this amplifier is chosen to be about 1.5, to raise the amplitude of the sine and triange waves (but not high enough to cause distortion). The output is passed through an attenuator R12 before going to the output driver voltage follower. This stage uses an LF412 opamp, since the risetime of the square wave imposes a high slew rate on the opamp. Cheaper opamps such as the LM358 have poor slew rates compared to the LF412. Also, the LF412 has a robust output stage. A mistake in this design is putting the attenuator after the stage introducing a voltage offset. This means that even the offset is attenuated along with the signal... not desireable behaviour. The solution would either require a pot with a high resistance (100k or more) in place of the 10k unit here, connected before the offset-introducing inverting amplifier U3B. No pot is required between U3B and U3A.

The power supply is a regular split-supply design based on a 78L08 and 79L08 linear regulators. I made use of a 9-0-9 transformer, which was a bit risky (very low headroom voltage for the regulators), but then my dealer didnt have a stock of 12-0-12 transformers. The output of the transformer is rectified and smoothed using 100uF capacitors. The float-voltage measured here for the 9-0-9 transformer was +/- 13V. The unregulated DC is regulated by the 78L08 and 79L08 to give the regulated supply rails for the circuit.

Again, the usual breadboard/veroboard. I used Relimate connectors for all the connections to and from the board. As a retrofit, I also added a connector to make available the +/- 8V rails to outside projects, since I do not own a bench split supply. The whole works was shoved into a small plastic chocolate box (!!!), but I haven't got down to drilling holes for the pots, switch, etc. etc. Performance was quite good, I measured frequencies up to 60kHz, with a rather clean looking sine wave. The square wave was attenuated too much, but that was OK by me. The rise time for the square wave was very good. As a test, replacing the LF412 with an LM358 resulted in very poor square wave output (it almost looked like a sine!), and other waveforms seemed to have crossover distortion (*shrug*... crossover distortion in an opamp!!!) Amplitude could be adjusted from 0V to about 7Vpp for all waveforms. Contact bounce was still a bit of a problem for the counter, sometimes the counter moved two places rather than one. A higher value capacitor will help a bit. I attached three LEDs through 1k resistors to each of the three used outputs of the 4017 to indicate which waveform is selected. This wasnt included in the schematic due to space constraints.
It would be wise to go through the excellently written "Everything You Always Wanted to Know About the ICL8038" (AN013.1) by Intersil. It is available at Intersil's website, along with the ICL8038 datasheet.

As I mentioned, move the attenuator to before the offset stage. I'm planning to make a frequency synthesizer based on a similar circuit with a PLL, using the 8038 as the VCO. It should be under computer control or use a microcontroller to form a stand-alone instrument.

10 February 2010

Solar Cell Phone Charger

Solar Cell Phone Charger

This little gadget uses a small 3 volt solar cell to charge a 6 volt NiCad battery pack which, in turn, may be used to charge many models of cell phones and other portable devices. The circuit "scavenges" energy from the solar cell by keeping it loaded near 1.5 volts (maximum energy transfer value) and trickle charges the internal battery pack with current pulses. The simple circuit isn't the most efficient possible but it manages a respectable 70% at 100 mA from the cell and 30% when the cell is providing only 25 mA which is actually pretty good without going to a lot more trouble or using more exotic components.
Ref. Description
PC1 3 volt solar cell from a sidewalk solar light
C1 22 uF, 10 volt (values not critical)
C2 100 pF, any voltage or type, typically ceramic
C3 10 uF, 16 volt or more for higher voltage battery
R1 1.5 k, any type
R2 3.9k, any type
R3 10k, any type
R4 180 ohm, any type
R5 4.7k, any type
R6 10 ohm PTC (see text).
L1 50 to 300 uH (see text)
D1 1N5818 schottky rectifier, just about any will do.
Q1 2N4403, or similar
Q2 2N4401, or similar
J1 output jack
B1 6 volt NiCad battery w/fuse
Here is how it works:
When the voltage on the emitter of Q1 rises a little over 1.5 volts, both transistors turn on quickly, snapping on due to the positive feedback through R5 and C2. The current increases in L1 through Q2 until the voltage across the cell drops somewhat below 1.5 volts. The circuit then switches off quickly and the voltage on the collector of Q2 jumps up, turning on D1, allowing the inductor current to flow into the battery. Once the inductor has discharged into the battery, the process starts over. The circuit can charge higher voltage batteries without any circuit changes since the voltage will jump up quite high on the collector when the transistors turn off. The circuit should not be operated without a battery attached. For a little more efficiency, increase R5 in proportion to the voltage increase on the battery. (For example, double R5 for charging a 12 volt battery.) A NiCad battery was chosen because they are particularly forgiving of overcharging, simply converting the excess current into heat.
The photocell was salvaged from an inexpensive solar sidewalk illuminator and it has an open-circuit voltage of about 3 volts and supplies about 100 mA in bright sunlight.  The circuit can handle more current but avoid cells that supply more than 250 mA. The inductor should have a low resistance winding but a surprising number of cores will work fairly well. The core in the prototype is actually a piece of ferrite antenna rod chosen simply to fit in the extremely limited confines of the package. Another unlikely inductor that worked well was 10 turns on one of those 1" long, 1/2" diameter large ferrite beads often used for power line baluns! The value of inductance isn't critical, perhaps between 40 and 300 uH and during proper operation there will be a pulse waveform on the collector of Q2 with several 10s of microseconds period. This prototype operates at about 40 uS as shown and the inductance measures about 50 uH.
For experimenting with cores or other circuit values, replace the NiCad battery with a zener of the same voltage and replace the solar cell with a 3 volt power supply with a series resistor, about 22 ohms to simulate moderate sun.  Measure the current in the zener and compare that power (zener current times zener voltage) to the power coming from the power supply (3 volts times power supply current) to see how the circuit is doing. When the power in the zener is over half the power from the supply, the inductor is good enough.
It is mandatory that a fuse be added near one of the terminals of the battery! (See the little green 2 amp fuse along the bottom edge of the battery.) Battery packs can supply dangerous current levels! Keep the lead from the fuse to the battery terminal as short as practical. I had to change this fuse; I'm glad it was there!
In addition to the fuse a 10 ohm PTC was added in series with the output to limit the available power but also to allow the unit to charge my Nokia phone which doesn't like a very low impedance battery as a charging source. (The phone simply displays "battery not charging".)  I have a few thousand of those, if you need a couple.  The PTC is actually soldered directly to the copper board and one end of the fuse connects directly to the top side.
Don't copy my assembly technique! First of all, I had to cut all the mounting posts out of the case to get the battery to fit and it is held in by glue. Notice the silver nuts soldered onto the PCB for securing the cover! Secondly, there is very little height for the circuitry so everything is pressed down flat against a piece of copper clad board using little bits of board for the connections. That's a fine technique but this prototype was just too tight for comfort. Third, I had to search a while to find an inductor that would fit! All the room was used up before I got to one of the larger parts! Having said all this, the final unit is very compact and solid but there was too much luck involved!
It works great! I simply leave it on my dash until I need it. I've charged several Nokia phones without a problem. It is actually more convenient than a cigarette lighter adapter because it can travel with the phone and it doesn't need sunlight to charge the phone. I will say that the thing charges my phone suspiciously fast and I wonder if I should increase the output resistance. Fast charging cell phone batteries shortens their life, if I understand correctly. Most phones have sophisticated internal charging circuits but I suspect the manufacturers sacrifice battery life for fast charging.  It might simply be that my phone hasn't been significantly discharged since I built the charger.

Automatic Trickle Charger

Here is the schematic for the automatic charger I have been using for my kids' battery cars. The charger is a small molded unit that probably doesn't supply more than an amp and this circuit would have trouble with much more. No current limit is provided by this circuit - it relies on the charger for that. The circuit could be modified to provide more current by lowering the 470 and 330 ohm resistors in the 5195's base circuit and the 10k in the collector of the 4401. A relay could also be used in place of the pass transistor.

Here is how it works: When the battery voltage is low, the voltage at the base of the first 2N4401 (on the right) is not sufficient to turn it on and the second 2N4401 is biased on by the 10k resistor. The power transistor is turned on and the LED lights. When the battery is fully charged the voltage will exceed a somewhat arbitrary "over-voltage" value slightly below 14 volts and the regulator will switch off. The 470k feedback resistor gives the circuit some hysteresis so that it will not turn back on until the battery voltage drops below about 13.5 volts. When the battery is nearing full charge the light will begin to flash on and off and after a few hours the light will only come on occasionally. This occasional over-voltage jolt sure seems to keep the batteries in great shape.

Experimental Alternator Controller

Here is an experimental (and simple!) regulator for alternator chargers. Q1 and Q2 are medium-power transistors and Q3 is a high-power type. The zener, D1 is chosen to set the charged voltage and will be about 10 volts for a 12 volt battery. The 0.1 ohm resistor sets the maximum field current. Not shown is the connection from the output of the alternator to the battery. An AC alternator will need a diode rectifier but most car types have the rectifier built in.
When the battery is low, current flows through the 0.1 ohm resistor and Q3 to the field coil. The voltage across the 0.1 turns on Q1 which limits the current in Q3 (about 0.7 / R or 7 amps in this case). When the battery is charged to about 14 volts, Q2 turns on and turns off Q3, stopping the charging.

12 V Gel Cell Charger Circuit

Recently, a fellow amateur was looking for a gel cell charger which would first charge at a fixed rate and then later switch to a trickle charge when the cell was fully charged. After reviewing several catalogs and web sites, the MAX712 IC was discovered. This IC meets all the requirements for almost any type of battery charging system. The circuit in Figure 1 was designed specifically for 12 volt gel cells. 
When a discharged gel cell is connected, the charger goes into a fast charge mode at a fixed rate of 400 ma. After the chip detects the voltage leveling off or when 4 1/2 hours has elapsed. (which ever happens first.) the fast charge will stop. After the fast charge has ended, the IC goes into a trickle charge rate of about 50 ma. This trickle charge continues until 13.8 volts is reached which will stop all charging current since the cell is now fully charged. If the cell voltage should drop for any reason, either a fast charge or trickle charge (IC will detect what is needed) will start again. 
When constructing this circuit, be sure to attach a small heat sink to Q1. Apply a DC (partially filtered) voltage of at least 15.3 volts. The voltage must never go below this level even under load conditions. Many of the DC wall transformers available will work just fine as long as they meet the minimum voltage requirement. The input voltage can be as high as 24 volts. If the input voltage must be in the 30 volt range, increase R1 to about 820 ohms. 
The output voltage must be aligned prior to use. Disconnect the battery from the circuit and apply power. Connect a digital volt meter or other accurate volt meter to pin 2 (positive lead) and to pin 12 (negative lead). Adjust R7 until exactly 13.8 volts is read. 
Because this circuit will not overcharge a gel cell, the battery can be connected indefinitely. This circuit is designed primarily as a 12 backup system and can be connected to the load provided the device to be powered only draws current during power line interruptions. Use a diode from the battery to load if needed. This circuit makes an excellent battery backup to an amateur transceiver.
The MAX712 IC and the .62 ohm resistor are available from Digi-Key, 701 Brooks Ave, Thief River Falls, MN 56701 (1-800-344-4539). Order part numbers MAX712CPE-ND and 0.62W-1-ND respectively. All other parts are available at Radio Shack. 
C1 MAX712 Battery Fast-Charge Controller IC (Cost is $6.27 from Digi-Key)
R1 680 ohm 1/2 watt resistor (Blue Gray Brown)
R2 150 ohm resistor (Brown Green Brown)
R3 68K resistor (Blue Gray Orange)
R4 22K resistor (Red Red Orange)
R5 .62 ohm 1 watt resistor (Blue Red Silver) (Cost is 27 cents from Digi-Key)
R6 1.8K resistor (Brown Gray Red)
R7 10K PCB trimmer resistor (103)
R8 470 ohm resistor (Yellow Violet Brown)
C1 1 microfarad tantalum capacitor (observe polarity)
C2,C4 .01 microfarad capacitor (103)
C3,C5 10 microfarad electrolytic capacitor (observe polarity)
Q1 TIP42 PNP transistor or similar (attach heat sink)
D1 1N4001 Diode (observe polarity)
LED1,LED2 2 volt standard LED (observe polarity)