Joule Thief 101

[MileHigh]

Double facepalm?

I can't speak for TK but let me give you my take on this.

Is the "double facepalm" because TK showed a solar cell powering his Joule Thief and you linked to a Lidmotor Joule Thief clip also powered by a solar cell?  "So there!"

If so, then really?  Lidmotor did not make a single claim at all about his clip so what the heck?  I would call that a "double facepalm FAIL" on your part.

Still waiting on your comments about the starting of the car and if you say nothing that would be another FAIL.

Just keeping it real.

[MileHigh]

Bill:

Two questions for you about Joule Thieves.

1. I am under the impression that for all these years you were believing that when a Joule Thief was up and running and driving a LED that it was in resonance.  Is this correct?  I thought I saw a comment from you about that.

If that's the case then I am assuming that now you have a much better understanding about how one works.  Is this true?

2.  I am also pretty sure that when you started stringing 10 or 20 LEDs together in series, that you have stated many times that each individual LED was the same apparent brightness as when you had just a single LED in the circuit.  It was only when you got to something like 40-plus LEDs in series that you started to notice that the individual LEDs were starting to get dimmer.

Is the above statement true?

MileHigh

[MileHigh]

Well I am going to plow ahead and assume that for a typical Joule Thief that each of the 10 LEDs in series as the load is just as bright as a single LED as the load.

Well, that suggests that you are putting way too much power into the LED.  With a string of 10 LEDs as the example, then you should be able to reduce your LED power requirements for a single LED by 90%.

So how do you get there?

For starters, you need to measure the average power-in for the Joule Thief and compare it to to the LED power-out only to get the base level of efficiency for your Joule Thief.  And of course we know one important fact:  The average power consumption of a single LED will be approximately the same as the average power consumption of 10 LEDs in series.  We also know that the operating frequency will be approximately the same for the two configurations.

Supposing for the sake of argument you measure that the power consumption of the Joule Thief is 100 milliwatts.  Then you measure the power consumption of the LED alone and supposing it is 60 milliwatts.

That means that the LED power is 60 milliwatts, and the overhead for everything else is 40 miliwatts.

So that presents you with a design challenge:  Drop the LED power by a factor of 10 to six milliwatts and still have the same apparent brightness and have a total power consumption of 46 milliwatts.

Again, if your Joule Thief will light a string of 10 LEDs in series with the same apparent brightness for each LED as when it drives just a single LED, then that implies you can reduce the LED power consumption for a single LED by 90%.  So that means you reduce the power consumption of your Joule Thief from 100 milliwatts to 46 milliwatts.  That should give you more than twice the running time.

So that's an interesting design challenge to give yourself.

And for all you wankers that give me shit for not being a "doer" on the bench, bring on the stony silence and the blank stares.  I state that from past experience.

MileHigh

[MileHigh]

Here is a totally radical Joule Thief "concept car" that I designed "paper napkin style" in my head today.  The chances of it outperforming a regular Joule Thief are almost zero but that's not the point.  The point is to think outside of the box and try something new.  If I had a bench and I was genuinely interested in this stuff I would build it myself.

There are three switches, and they are all programmed by a 555 timers or perhaps a super low power microcontrller.  The premise is that the power for the switching would be provided by an external fresh battery and you assume that that battery would last long enough to drain up to 100 batteries that actually power the Joule Thief.  Thus it is perfectly valid to have an external battery to power the timing system.

The basic principle of operation is that current is always flowing through the coil.  The current is occasionally given a boost (like boosting the International Space station in its orbit) by connecting the battery with S1.  The LED is occasionally lit by opening the bypass switch S3.  S2 is there to complete the current loop, and the diode is there to give the current a bypass when S2 switches.

It's radical, but it does give you 100% full control over how much current is flowing in the loop and thus the brightness of the LED when it is switched ON.  It also gives you full control over the ON/OFF duty cycle of the LED.

Note that the issue of the sloping voltage/current waveform for a regular Joule Thief is eliminated.  The assumption is that the inductance is quite large and the current through the LED when it is ON is nearly flat and unchanging.

Think outside of the bloody box.

MileHigh

[Magluvin]

@ mags

looking forward to your results

I never experimented with anything as low as a 1:1.

I started with 22 windings, which I guess would be an 11:11, considering the center tap?

many replicators carelessly wound these "bifilar",
while I myself preferred to wind them, each wire at a time, in the same fashion.
This decreased the imperfections in the coils that may lead off the "true slope"

greater perfection resulted in a cleaner signal.
once I got the sine as clean as I could, within the primary coil:

all other advancements were done using a secondary winding on top of the JT primary, on the same ferrite torroid.
even my experiments that daisy-chained multiple torroids, were all driven by a secondary in this manner.
the secondary was not electrically connected to the JT, and only driven by the inductive coupling through the ferrite.
The signal on the secondary was usually "cleaner" than the primary, as it was a function of the inductors resonance,
not including the destructive feedback from the primary circuit.

everything I did was built upon this baseline.

It is nice to see someone going beyond that to make it simpler, one to one winding and observe the limitations of induction.

Yeah, the 1 to 1 was turn ratios.  I will try separate windings like you said and wind them neat. My last pic with the larger toroid was separate windings, but not neat anymore as it was in a box with some stuff for a while and was a lil banged up. And many more turns. Ill try the 11 and 11 and see what we get.

Finishing up this gti sound system this week, as I do it on the side at my shop, but work at another shop during the day. So will have more time to fiddle.

Thanks for the help.

Mags