Joule Thief 101

[Magluvin]

Brad:

Presumably two different LEDs won't make much of a difference.
Assuming the cores are about the same size and both are high-permeability, not much difference.
Ditto for the transistors.

The number of turns for each coil will indeed make a difference.

There is no reason for your Joule Thief to run so fast and I think something is amiss.  You clearly have a serious design problem because you have the Joule Thief death spike.

Please go ahead and explain why it is happening.

MileHigh

"Presumably two different LEDs won't make much of a difference."

Well, a red led may come on at 2.1v and a white maybe at close to 4v. Some vary.  The difference does affect things.

So presumably isnt the way to go on this one.


Mags

[tinman]

Brad:

Presumably two different LEDs won't make much of a difference.
Assuming the cores are about the same size and both are high-permeability, not much difference.
Ditto for the transistors.

The number of turns for each coil will indeed make a difference.

There is no reason for your Joule Thief to run so fast and I think something is amiss.  You clearly have a serious design problem because you have the Joule Thief death spike.

Please go ahead and explain why it is happening.

MileHigh

Mh
There is no death spike-you need to understand the circuit, and why there is a spike across L2 when the transistor becomes open. It all has to do with the number of turns on each coil-the more turns, the higher that L2 spike will be-there is nothing out of the ordinary with that spike.

In regards to the difference in LEDs-this is something you should clearly understand, and how different LEDs will effect the width of the kickback pulse. I have a larger core than Mags, and that means that I can build a larger magnetic field-along with a higher current draw. As we now have a larger amount of energy stored in the way of the magnetic field, then dose it not make sense that we have a longer time period during the LED on time.

The other question you have, I will answer once I am in front of my computer, and not writing from my phone, but if you look at the base current trace in my last video, then I think you should be able to work it out.

Brad

[tinman]

Tinman,

Moving the scope ground to the transistor base is not a very good way measure base current.  Any time the base voltage is less than the Vbe turn on voltage or is reverse biased, the base appears mostly capacitive (combined with a small amount of leakage current).  Moving the scope ground lead to the base changes the capacitance at the base.

A better way would have been to keep all scope grounds on the battery minus and use both scope channels and probes to measure both ends of the 10R base resistor.  With both scope channel vertical gains set identically (ca 20mv/div) invert one channel and then add them together (or use a math function to subtract one channel from the other).  This allows you to make a differential measurement across the 10R with minimal capacitive loading (particularly if you use 10X probes).

Also, you only have two channels so you should be taking advantage of the external trigger input.  It has limited "viewing" function, but if used smartly, it can be used to keep a marker on the screen at all times that indicates a particular time point in the waveform being observed.

That said, I do not believe the base current flowing thru L2 is acting in the manner you surmise.  In fact, I would think that current flowing thru the base (and L2) would generate a flux in opposition to that generated by current flowing thru L1.

You should consider measuring the Vce and Ic at the two base current extremes you are using.  You may find that Vce or Ic varies with Ib.  To do so, instead of making videos, consider creating six scope captures.

I would add a 1R resistor between the transistor collector and the L1/LED junction and connect the scope grounds to the emitter/battery junction.  I would then choose two base resistances to measure operation with such as 1K and 500R.  I would connect a third probe to the external trigger input and connect that probe to the L1/LED junction and trigger on the rising edge (which is the straightest edge in your waveform).

The first capture would be with the CH1 probe at the pot/10R junction in the base leg, and the CH2 probe at the collector of the transistor with the pot set to 1K (or using a fixed 1K).  The next capture would be identical, but with the pot set to 500R.  These two captures would let you see any change in Vce as the base current is changed between the two base currents.

The next capture would be with the CH1 probe moved to the L2/LED junction (i.e., the other end of the collector's 1R).  Using 1K as a base resistance, set both scope channels so that there is a good signal, and then invert and add (or subtract using math) to measure the differential voltage across the 1R, which is the collector current Ic.  The next capture would repeat this test using the 500R base resistance.

At this point you would have the collector to emitter voltage (Vce) and the collector current (Ic) when using the two base currents with 1K and 500R base resistors. 

We already know the base current is increased when the base resistor is changed from 1K to 500R, but it would be useful for further analysis of turn on, etc, to measure the base current under the same two operating conditions.

As was done to measure the collector current, make a capture using both probes to measure the difference voltage across the 10R resistor at the base using first the the 1K base resistance and then repeat this test making a second similar capture using the 500R base resistance.

You would now have a set of six scope captures.  Because the external trigger input was used, they are fairly closely time aligned.  The first pair of captures gives you Vce data, the second pair gives you Ic data, and the third pair gives you Ib data. Each pair provides data when the base resistance is set to 1K and 500R.

You might consider doing all these tests using your bench supply set to a known voltage as the battery voltage may change over the time period needed to make these six captures.

Although its just a joule thief, making these captures would be good measurement/scopology practice..

Just my 2 cents...
PW

Will get to them ASAP

cheers

Brad

[MileHigh]

"Presumably two different LEDs won't make much of a difference."

Well, a red led may come on at 2.1v and a white maybe at close to 4v. Some vary.  The difference does affect things.

So presumably isnt the way to go on this one.


Mags

No kidding.  So you will have a somewhat higher amplitude feedback signal to switch the transistor ON, and a somewhat higher amplitude feedback signal to switch the transistor OFF.  The energy burn to illuminate the LED will be slightly shorter.

I don't view this as a game changer at all.  Anybody that knows their stuff should be able to pop a different LED into their Joule Thief circuit and then poke around with their scope probe and ensure that everything is fine or if required take some slight corrective actions as needed.   

[MileHigh]

Brad:

There is no death spike-you need to understand the circuit, and why there is a spike across L2 when the transistor becomes open. It all has to do with the number of turns on each coil-the more turns, the higher that L2 spike will be-there is nothing out of the ordinary with that spike.

So you are a comedian now?  Shaking my head.  Do you know how an NPN transistor is supposed to work?

You have a major problem with your Joule Thief and the root cause for that is that is that 90% of the time you just fly by the seat of your pants and just plow forward without thinking.  Now you find yourself trying to understand how a Joule Thief works like you have never done before and that's a good thing.

For example, your "adding the L1 and L2 energy together for a brighter LED" business is dead in the water.  It never even made the slightest bit of sense in the first place but you did not even bother to think it through.  That's you flying by the seat of your pants and plowing forward without thinking.

MileHigh