Joule Thief

[drodenbe]

hello    conradelektro

I find that strange that you don't get anything from the 125t coil.  One thing I would do is to put a 2.2 k resistor in series with the led or take it out all together until you get working at least that way.  I have a digital power supply to first measure my current draw before I use a battery and I noticed that I did not get any reading on the digital meter in the ma position.  I usually install a 1 ohm resistor in my boards to measure the current but didn't in this case.  I noticed that if you are not using a 10 -20 turn pot I don't think you will find the sweat spot.  I know I used a standard pot when I first ran the circuit and kept going right by where it needed to be.  The adjustments with the 20 turn pot when you approach the sweat spot is only very fine tweaking like only 1 degree or less.  I still have mine running and I noticed that the time of day does make a difference.  It did not start working again correctly  until after 4:00 pm my time. Did you look at the wave form with a scope on the 125t coil.  I got a good strong signal.  I also put a diode on one end of the winding and brought that back to the positive battery lead then connected the other end to earth ground.  That improved the process even though it was working b4 I did that.  I have always been able to adjust the pot so I could at least charge the run battery but depending on the time of day seam to determine if my charge battery would charge or just set there with the same voltage.

I don't want to change the subject or drift off of it completely but I did while I was letting the circuit run, hooked up one of my kacher circuits with a kacher coil with 182 turns of 28g wire and 10t 19g wire the form is 5 inches long and 4.5 inches in diameter I put 4 nicad batteries in series to run the circuit and put it inside the coil and it was pulling 18-20ma at 5.3volts and charged up the batteries to 5.4125 volts and holding steady.  I am also running 3 led's in series through a diode plug.  But this is for a different group.  I just wanted to mention this because what ever is going on in the joule thief circuit all so happens in the kacher but only better.

[drodenbe]

one last thing You guys are probably wondering why I keep using the 19g wire.  In reality it is really 19.5 gage and the 28 is 28.5 which sounds strange but I have about 4000 feet of each and so I use it unless I really need some other size.  We have a old motor rebuilding company in town and I go there to get my wire and old motors they scrap out.  I found that they along time ago use to use this size wire in motor rewind but no longer use it .  They have thousands of feet of odd gage wire on hand and no longer use it.  I told them that I would take off there hands if they would sell it for a reduced rate.  I get it at 3-5 dollars a pound which is great.  They also have the cloth coated wire by the 5-10 thousand feet rolls.  I bought a couple thousand feet of that for the stubblefield coil work. works great.  So you guys might try going to your local motor rebuild shop if you have one and maybe if they are old enough they might have some of this stuff on the shelf that you can get cheap!

[conradelektro]

hello    conradelektro

I find that strange that you don't get anything from the 125t coil.  One thing I would do is to put a 2.2 k resistor in series with the led or take it out all together until you get working at least that way.  I have a digital power supply to first measure my current draw before I use a battery and I noticed that I did not get any reading on the digital meter in the ma position.  I usually install a 1 ohm resistor in my boards to measure the current but didn't in this case.  I noticed that if you are not using a 10 -20 turn pot I don't think you will find the sweat spot.  I know I used a standard pot when I first ran the circuit and kept going right by where it needed to be.  The adjustments with the 20 turn pot when you approach the sweat spot is only very fine tweaking like only 1 degree or less.  I still have mine running and I noticed that the time of day does make a difference.  It did not start working again correctly  until after 4:00 pm my time. Did you look at the wave form with a scope on the 125t coil.  I got a good strong signal.  I also put a diode on one end of the winding and brought that back to the positive battery lead then connected the other end to earth ground.  That improved the process even though it was working b4 I did that.  I have always been able to adjust the pot so I could at least charge the run battery but depending on the time of day seam to determine if my charge battery would charge or just set there with the same voltage.

I don't want to change the subject or drift off of it completely but I did while I was letting the circuit run, hooked up one of my kacher circuits with a kacher coil with 182 turns of 28g wire and 10t 19g wire the form is 5 inches long and 4.5 inches in diameter I put 4 nicad batteries in series to run the circuit and put it inside the coil and it was pulling 18-20ma at 5.3volts and charged up the batteries to 5.4125 volts and holding steady.  I am also running 3 led's in series through a diode plug.  But this is for a different group.  I just wanted to mention this because what ever is going on in the joule thief circuit all so happens in the kacher but only better.

I will do further experiments with a better potentiometer. Yes, I see nothing with a Multimeter at the third coil. Over a 100 Ohm resitor I could see something with an oscilloscope. A Sinus type signal with a bump in one halve at about 250 KHz, at a potentiometer setting with about 8 mA current draw. But the oscilloscope had trouble seeing it consistently. I lost the signal at lower than 6 mA (supply current) potentiometer settings.

Do you think a current draw of more than 3 mA can be compensated by whatever the coil induces into the AA NICad?

I also have two Kacher coils somewhere and will do tests in the coming days. The Kacher coil in the photo has 65 turns to get at 1 MHz, the other 200 turns swinging at around 400 KHz, primary was 2 turns. (See photo of a test in Kacher style some time ago). You are right, it should be more pronounced in a Kacher, because it radiates like mad. But my Kachers used Watts (2 to 7 Watts). That would be a lot of power to be given back to a battery? But may be I can set up a Kacher that uses only some mA. If I remember correctly I had a Kacher running at some 100 mA.

Greetings, Conrad

[Rfacts]

drodenbe:
Thanks for sharing your efforts and results of this project and for answering questions about it.  I appreciate the years of knowledge and experience that you have gained and the approach and equipment that you use.  To check for same results it is very important to replicate as closely as possible so I have transcribed your notes to make sure I got them right.  Please review and correct or clarify them as needed to help me and maybe others. 

Including the coil inductances, waveform, voltages and frequencies on your schematic was very helpful.  The very touchy adjustment that you have noted to find the sweet spot indicates that the operating frequency is critical to replicate your results.  Significant deviation from the 19.5 gauge magnet wire, number of turns, winding direction, 20 turn potentiometer, and tape between windings (interwinding capacitance) will introduce other factors that may prevent the confirmation of your results.  The more exact the replication the more accurate the confirmation whether the results are the same or not.  So please verify/clarify/add to the following and on a separate post I will address some coil winding details and questions that I have.  Thanks in advance.

Free Runner Schematic Notes 01-21-13 DJR:
1. Could use coil form and slip battery inside.
2. If you keep coil inductance close to stated values you should have proper frequency range.
3. Option 1 waveform: Positive pulses, 3.29V RMS, 5.28V Peak to Peak, 320KHz
4. Some transistors work better than others. Had best success with 2N3904 (NPN).
5. Glued insulated washers to top and bottom of battery (to use battery as coil form, wire connects to battery through hole in washer).  One layer of double stick carpet tape and wound coils on prototype (used carpet tape between layers to hold coil windings in place, also increased interwinding capacitance).
6. Set potentiometer for frequency from 42KHz to 400KHz.
7. With no load on 125 turn coil, battery inside coil will charge and keep Joule Thief running at around 52KHz.
8. Ran test for 3 days with Option 2 and jumbo LED in circuit. Kept battery charged. Started with battery voltage 1.02 volts ended with 1.2288 volts.
9. Have to watch both battery voltages (run battery and charge battery) and adjust potentiometer so both charge. Very touchy adjustment.

Resistor in series with LED: 2.2K ohm
20K ohm potentiometer: 10 to 20 turns to allow fine adjustments

[conradelektro]

My first test of drodenbe's JT circuit (see my Reply #16094 above) showed a 40 mV loss of battery Voltage in about 24 hours.

I set up a new test: the third coil (130 turns) is shortened via a 1K resistor. And I adjusted the potentiometer at the base of the transistor till a nice Sinus curve appeared over the 1K resistor. This happens reliably at a power consumption of about 5 mA (about 10K).

But the scope probe has a strange effect: the frequency of the signal from the third coil over the 1K resistor depends on which way I connect the scope probe (see the photo which contains two screen shots from the scope). The power consumption measured over a 1 Ohm shunt (at battery plus pole) is about 5 mA but changes slightly when I connect the scope probe over the 1K resistor shortening the third coil. It will also go up slightly when I remove the 1 Ohm shunt.

So, a new test is running at a power consumption of about 5 mA and a nice Sinus signal over a 1K resistor shortening the third coil. We will see tomorrow how that goes.

I found my 68 turn Kacher (oscillates at about 1 MHz, see the photo in my last post) and will do tests over the weekend with four Ni-MH AA accumulators in series (about 4.8 to 55.2 Volt).

Greetings, Conrad