User:Jstapko/Personal/ElecProjects/JouleThief

=Joule Thief=

Conceive
This project was inspired by a desire to run an antique AshFlash Stratobeam flashlight. Years before, I had read at the Surefire Flashlight website of a circuit that varied the voltage coming from the battery, including increasing it to keep lamp brightness up even as the battery voltage tapers off at the end of the battery's life. Initial internet searches led me to the Wikipedia voltage boosters page, which then led me to the Joule Thief.

First Attempt
I disassembled a Kayton transistor radio to salvage some transistors. The transistors did not act electrically like transistors at all, and internet search results such as vintage-radio.com suggested that this may have been because they were very old, and transistors of that time period (1960's) could be unreliable. This did not go very far, so I salvaged a 2SA608 transistor and moved on to attempt 2 (below).

Second Attempt
For this attempt, I didn't exactly copy the conventional joule thief schematic, which calls for:
 * a toroid core transformer
 * a 2n2222A transistor or equivalent
 * a 1k resistor in series with the base

Major differences between my circuit and the typical Joule Thief were:
 * I used a potentiometer to set the base bias voltage, because I wanted to see the effect of bias point on circuit operation.
 * I used a 2k resistor (not 1k) between the pot wiper and the base of the transistor to limit base current to 5mA.
 * I used a SA608 transistor instead of a 2222a, because that's what I was able to find on scrap equipment.
 * I used a conventional E-I core transformer, and then a ferrite loopstick antenna salvaged from a radio, but with custom windings, instead of a toroid
 * I used a variable power supply for initial tests instead of a dead battery

This attempt resulted in several problems:
 * could not properly solder a joint, the solder would crystallize
 * resistors would overheat and burn up from excess power
 * many parts were crowded onto a small prototyping board, which caused their leads to short together, preventing proper circuit operation.
 * the potentiometer moved around, which could cause its leads to short together or to other leads. It was also hard to turn.

To correct these problems, I made the following adjustments in Attempt 3:

Third Attempt

 * mounted the breadboard on a piece of wood, and a piece of plastic on the front of the board to hold panel mount components such as:
 * an 11 contact rotary switch
 * several potentiometers for setting bias
 * screws with solder lugs and wires connected to them for connecting oscilloscope and voltmeter probes to breadboard points


 * put electrical insulation (salvaged from wire) on bare component leats to prevent them from shorting together
 * did an extensive internet search/You Tube session to figure out the solder crystallization problem. Major results of this search were:
 * use the largest soldering tip that can fit well on the solder joint, so that the tip stays as hot as it can while soldering
 * clean the joint and solder with isopropyl alcohol before attempting to solder
 * properly tin the soldering iron tip and wires before attempting to solder
 * things that can cause crystallization include an improper temperature-time curve profile, old or dirty solder, bad flux, dirty joints, and moving the joint while the solder is liquid
 * 63/37 "eutectic" solder is liquid over only a very small temperature range, which means that the solder is liquid for a shorter time. This means it is easier to keep the joint still while the solder is liquid, so if stable mounting devices are not available, eutectic solder may help.


 * Based on advice from allaboutcircuits.com, I used high power (2-5W) resistors instead of the 1/4 watt versions to prevent burnout. I also measured them with an ohmmeter to make sure they were within spec prior to use.

Next Steps

 * find a toroidal transformer core and wind it with equal numbers of turns
 * find a 2N2222A or similarly high gain transistor and wire the circuit per the schematic on the Wikipedia Joule Thief page page
 * build an improved version of the breadboard with front panel, less cluttered with wires
 * build a circuit based on the jfet version described on the Wikipedia Joule Thief page
 * put a capacitor across the output, as shown and described here
 * the ideal output waveform seems to be a square wave

This Page Needs

 * a link to a soldering tutorial based on the crystallization, which needs to be created
 * links to pictures of my attempt and videos of the waveforms
 * pictures of attempt 3
 * improved organization
 * schematics