Joule Thief 101

[picowatt]

As far as i can tell,it only increases in amplitude.

Tried some uf4007 diodes,and result's are as TK's-to a point. If i drop the 1k resistor down to a 100 ohm resistor,i get the reverse current spike back across the CVR,but not enough to light the LED.

I tried a few of those little inductors that look like resistor's(what are they called ),and even with the fast diode's in there,by raising the frequency,the LED will light brightly-different inductor of course =different frequency.

So we have learned something here
But i am a little confused about this turn off time,as the 1n4007 data sheet's all show a test frequency of 1MHz,while here i was using only 10KHz. If it is suited to frequencies of 1MHz,then how can it not switch off fast enough for 10KHz ?.

The test frequency you mention is for testing junction capacitance which is done with the diode maintained in a reverse biased (off) condition (it never turns on while that 1MHz is applied).  Note that junction capacitance varies as the reverse voltage varies (a phenomenon exploited in varicap diodes).  The data sheet often has a graph showing Cj versus reverse voltage.

You will have to look at a lot of data sheets from different manufacturers before you will find one even willing to state the reverse recovery time of a 4000 series diode.  They are generally intended for line frequency use or a tad faster, but their speed is rarely bragged about.  I think I had to look at 6 or more data sheets before I found one stating 2us, and that was probably because that manufacturer was proud of that speed (as likely compared to other slower units).


With regard to your "10KHz" waveform, I'll try one more time:

Unless the waveform is a pure sine wave, the repetition rate of a waveform has only a little to do with its actual frequency content.  Every waveform such as a square wave, triangle wave, nasty looking pulse, or any other arbitrary waveform, consists of pure sine waves of various frequencies summed together at various amplitudes.  A decent looking square wave with a relatively slow rise/fall time and a bit of ripple seen on its top and bottom "flat" surfaces will contain a fundamental and at least 4 or 5 harmonics.  A sharp looking square wave can have many more harmonics.  Decreasing the duty cycle requires the addition of even more harmonics to produce the narrow portion of the duty cycle.

A fast edge/rise time transitioning in 1us is going to need at least a 500KHz component, as a half wave at that frequency is 1us.  If that edge is even faster, an even higher frequency component is required.  So, even though you may be looking at a 1KHz square wave, it is the rise and fall time (as well as over/undershoot, ringing) that determine its actual frequency content.  A 1KHz square wave with a 30ns rise/fall time can have significant frequency content into the tens of megacycles.  Slew rate limiting (rise/fall time limiting) is often used with digital data to reduce unwanted higher frequency harmonics which may cause RF interference or unnecessary power consumption.

I suggest at least looking at the images and gifs in this Wiki:

https://en.wikipedia.org/wiki/Fourier_series

I also suggest that you connect your FG to your scope, set the scope to display an FFT and then play around with different waveforms, duty cycles, and frequencies as you watch the FFT on the scope.  This will allow you to visualize the changes to a waveform's frequency content as you modify rise/fall time, duty cycle, waveshape and frequency using the FG's controls. 

This gif from the Wiki provides a really great visualization of what the FFT display on your scope is showing you:

https://en.wikipedia.org/wiki/Fourier_series#/media/File:Fourier_series_and_transform.gif

Below is the circuit (modified version,that excludes the charge battery),and scope shot from points depicted in circuit diagrams.-->Please note in the second,i have no channel offset,and have switched channels around--i do this often to make sure one channel is not playing funny buggers.

1-So how has L1 and 1.5v battery(collector/emitter junction) got around 4.7 volts across it before the transistor has switched off?.
2- How is there below 0v across L1 and battery(collector/emitter junction) before transistor has switched on?.<--This one could be explained by some sort of overshoot of L1,and where i think coil capacitance may come into it?.


Brad

I think you are going to have to look at current waveforms (at least base current) to perform a proper analysis of this circuit.

I assume you are probing directly at the transistor...

Rather than use any of these JT circuits, I would look into the several energy harvesting IC's available from various manufacturers.  Some can charge a lithium ion battery from a 45mv source (TE gen, electrosmog, etc).

Perhaps I am missing the point...

PW

[sm0ky2]

In today's money that 22 cents becomes $4.40 -
many of us cared.

I must have missed the point in time when they let us exchange our 22 cents for $4.40
my money is still the same,
only difference is gas costs more...

If "today's" money is 20x more, why aren't we all making $50-60 an hour as min wage?

[sm0ky2]

I would look into the several energy harvesting IC's available from various manufacturers.  Some can charge a lithium ion battery from a 45mv source (TE gen, electrosmog, etc).

PW

yes, many of these are functionally similar to what we do with the JT.
rather than using an inductor to step up the voltages,
they use transistors with a low-threshold, or "zero"-threshold.
basically, they can turn on with ANY voltage, and as small as a couple hundred nano-Amps of current.
Most of them work with AC or DC inputs, or a combination of both.
and a single unit can use multiple inputs.
This is usually stored in some sort of supercap/ultracap, and a resistive circuit to deliver 3-6v
for small-scale electronics, or transferred to batteries for larger storage.


There are a lot of experiments going on with these recently,
involving areal wires, for atmospheric harvesting.

[tinman]

Why not just use a J/FET,and have a high winding ration between the primary and secondary,where the(high turn) secondary would turn the J/FET off.


Brad

[picowatt]

Why not just use a J/FET,and have a high winding ration between the primary and secondary,where the(high turn) secondary would turn the J/FET off.


Brad

Actually, some of the Linear Technology harvesters do use a 1:100 xfmr on the front end to get down to 20mv.  Most others just use a simple inductor in a boost converter arrangement.

If all you want to do is light an LED to extinction using a "dead" battery, a JT may suit your needs.

However, if you are wanting to harvest ambient energy for wireless remote sensing, remote control, the "internet of things"(will it ever end...), etc, an energy harvester IC may be more suitable.  With a commercial harvester IC you often get guaranteed start-up, battery/reservoir over/under voltage protection, charge control, MPPT, output voltage regulation, UPC flags/control, improved efficiency, etc, that may simplify production of an actual product (and you get to take advantage of thousands of man hours of engineering and application testing for next to nothing).

Linear Technologies, Texas Instruments, Maxim IC, ST Microelectronics and others all produce IC's dedicated to energy harvesting (amazing).

Although it would likely not be suitable for my EDC (I use it too often), an "at the ready" emergency use flashlight that at least trickle charges itself from ambient sources could be made using these harvester IC's.

For further reading:


http://www.ti.com/lit/ds/symlink/bq25504.pdf

http://cds.linear.com/docs/en/design-note/DN483.pdf

http://www.linear.com/parametric/energy_harvesting

https://datasheets.maximintegrated.com/en/ds/MAX17710.pdf

PW