Joule Thief 101

[tinman]

Now we get into the meat of the matter.   

When the 1K resistor across the LED is removed, the LED goes out.

Yes indeed.

Now,below is the next test.
Start off including just one diode in the circuit-D1.
Check current flow reverse peak(when LED is lit by the 3% reverse spike.
then move your FG red lead to include the next diode,so as you now have two diodes included in the circuit. You should see a higher value reverse current spike during the LED on time.
Then include the 3rd diode in the circuit. Once again,you should see an increase in the reverse current spike value.
So as you include another diode into the circuit,the reverse current spike should go up-not down.
This would seem to be the reverse of what we should see,as we decrease the capacitance value as we place more diodes in series.
The second thing that says we should see a decrease in reverse current flow through the LED and 1k resistor,is because we also increase the voltage drop across the diodes as we add more of them into the circuit.

It is also odd that if we reduce the V/in by way of the FG,the LED gets duller,and the reverse current spike becomes less. But if we decrease the voltage by way of diode voltage drop,the LED gets brighter,and the reverse current spike increases.

But when I replace the resistor with a 2.2 uH inductor (DC resistance about 0.8 ohm) , the LED comes back on.

I did exactly this,and first thought that maybe the resistor was also inductive. But further test have shown that not to be the case,nor can i read any sort of inductive value of the resistor with the meter.


Brad

[tinman]

Anyway....

Was looking into the battery oscillation more.  With the new circuit a relatively fresh AA battery is ringing at about 15mhz. The voltage swings seem quite high, so I put a white led across the battery. Circuit is running without load leds. Only the led across the battery.

Mags

What polarity is the LED in relation to the battery Mag's ?.

At 15 mhz,and being a new battery,i would think the LED is lit due to skin effect.


Brad

[TinselKoala]

How about them apples:

When I replace the 3 x 1n4007 diodes with 3 x 1n4148 diodes, and have the 1k resistor across the LED, the LED does _not_ light.

Now if I replace the 1k resistor with the 2.2 uH inductor the LED _does_ light (rather dimly). And as I increase the frequency from 10 kHz, the LED gets brighter and brighter, until brightness peaks at around 1.2 MHz and then begins to decrease.

With just the resistor, or nothing, across the LED, the LED stays off as I increase frequency from 10 kHz.

[picowatt]

Your thinking is too recent and too modern.

I've been around a while...

All emissions control is "modern" (as is the whole ecology awareness movement).  You do realize that prior to the early 1970's, and most definitely prior to the 1960's, nobody considered or worried about auto emissions or fuel economy.  With 22 cents a gallon gasoline, nobody cared. 

Some techniques used for emissions control do have their beginnings with non-emission control related uses (as I inferred regarding EGR).  Some of the methods used for WWII aircraft being a good case in point.  But used as they are today for emission control, with the fuels used today, I'll stand by my previous statements.

If you want to talk different fuels (or believe 1987 is ancient history as per Gunnerman) that is a completely different subject.

Given the fuel available at the pump, and the methods used to control emissions in modern ICE engines, emission control comes at the cost of performance or fuel economy reduction. 

But this is likely more than enough off topic discussion on this JT thread...

PW

[picowatt]

Yes indeed.

Now,below is the next test.
Start off including just one diode in the circuit-D1.
Check current flow reverse peak(when LED is lit by the 3% reverse spike.
then move your FG red lead to include the next diode,so as you now have two diodes included in the circuit. You should see a higher value reverse current spike during the LED on time.
Then include the 3rd diode in the circuit. Once again,you should see an increase in the reverse current spike value.
So as you include another diode into the circuit,the reverse current spike should go up-not down.
This would seem to be the reverse of what we should see,as we decrease the capacitance value as we place more diodes in series.

You are dealing more with turn off time than junction capacitance.  I suspect the 4007's are turning off one after another.  Is you positive going current pulse/spike also wider with all 3 diodes in the circuit as compared to just one?