Is joule thief circuit gets overunity?

[gyulasun]

Try to use the first icon on the left above the smiley line when you are in Reply, it is "insert hyperlink"  (includes a small blue Earth globe). Maybe it makes inserting that image possible, I have not tried.

Gyula

[Void]

Thanks Gyula. It looks like the hyperlink feature doesn't work for inline images. It seems that maybe the only way to do it is to attach the images.

Ok, I have run some power tests with the 'standard' joule thief circuit. I used whatever parts I happened to have handy. I didn't have a smaller ferrite toroid, so I used a larger one I had available. Each winding on the ferrite core has 28 turns.

I am attaching images of the schematic of the circuit I used, my proto board layout, and scope shots of my voltage and current measurements taken for average input and output power measurement.

Since the scope waveforms were all pretty stable and quite consistent over time, I am using the RMS measurement feature on my scope to determine average input and output power. It may not be super accurate, but should be close enough to actual, I would think.

I used 1 ohm, 5%, 2W resistors in series to make input and output current measurements.

Here are my measured results:

Average Input Power:
Vin = 508mV
Iin = 8.55mA
Input Power = 4.34mW

Average Output Power:
Vout = 897mV
Iout = 3.66mA
Output Power = 3.28mW

Efficiency = 3.28mw/4.34mW x 100 = 75.58%

My next step will be to try to improve on the efficiency of this circuit.
Tinselkoala has suggested a 70nF capacitor in parallel with the 1k base resistor, so I will try that.
Does anyone have any other suggestions for improving efficiency while still keeping close to this circuit design?

Lawrence, do you have any suggestions for me to make my circuit closer to your boards? My circuit should be fairly close to your circuit except for the ferrite core I used, and possibly my winding method. Do you think that the ferrite core is an important factor here, or do you think my coil winding method is making a difference here? If I am correctly interpreting the scope shots you have been posting recently (it would help if you mentioned what each scope trace is representing in your scope shots), your input current waveforms look a lot different than mine?

- void -

In the scope shots below, yellow are the voltage traces, and green are the current traces.

[Void]

I tried another test with the exact same Joule Thief circuit that I used in my previous test (see me previous post above), except this time I placed a capacitor in parallel to the 1k ohm base resistor. I didn't have all that many capacitor values in the low nF range to try with, but I tried 100nF, 20nF, and 10nF. The 10nF nominal value (12.5nF measured value) capacitor seemed to give the highest efficiency of these three values. With the 12.5nF capacitor in parallel to the 1k ohm base resistor, efficiency increased to 95.2%, however the LED glows dimmer for about the same input voltage due to less current flowing in the circuit. At any rate, adding a parallel capacitor across the 1k ohm base resistor appears to have increased efficiency a fair bit with my circuit setup. Thanks Tinselkoala for the suggestion. If you compare the scope shots from the previous test to the scope shots from this new test, it appears that the parallel base capacitor limits the amplitude of the current spike peaks quite a bit, which for some reason seems to increase efficiency a fair bit. This also caused the frequency of oscillation to increase a fair bit as well. I am attaching the scope shots which I used to make the input and output measurements with the 12.5nF cap across the base resistor.

Average Input Power:
Vin = 510mV
Iin = 1.84mA
Input Power = 938uW

Average Output Power:
Vout = 921mV
Iout = 970uA
Output Power = 893uW

Efficiency = 893uW/938uW x 100 = 95.2%

- void -

In the scope shots below, yellow are the voltage traces, and green are the current traces.

[TinselKoala]

@void: Thanks for performing your tests. I think that the toroid that LTseung is using has more turns on the "secondary" winding; this could make a difference in your basic results (and also your result with the extra capacitor across the base resistor.) May I suggest that you try 25 and 50 turns? Also, when a core is assembled from pieces there can be small gaps at the mating surfaces that affect the magnetic field in the core. While JTs will work with almost any transformer core design, even straight rods, to give Lawrence's system a fair trial you really should consider using a one-piece toroid for the transformer. Old PC power supplies will have usable toroids in them. Toroids aren't that hard to wind if you use a long skinny spindle-type bobbin that will hold your wire and fit through the hole in the toroid. Wind the wire onto the spindle then "unwind" the spindle through the hole in the toroid, it makes things a lot easier.

I'm not sure about your calculations using RMS values. I'd like to hear PW and .99 weigh in on that topic. You didn't include your current-viewing resistors or your probe locations in your diagram, so I'm assuming you are using them in the same positions that Lawrence showed on his last posted schematic. Looking forward to more results, thanks! Is your scope capable of integrating?

[TinselKoala]

@Lawrence: How can you be absolutely sure that your timer device is not supplying any power at all to your circuit under test? How can you be sure that you have indeed accounted for all power inputs to your circuit? As I've shown, the circuit you and void and I are using is capable of picking up power from the environment very easily, and can produce bright light in the LED without even being connected to anything with any wires at all, if there's sufficient ambient power floating around. And power supplies and instrument probe leads can trick you by supplying power through ground loops and other wiring infecilities.