How do I measure the efficiency of a Joule Thief?

[MileHigh]

It might be somewhat easier and you could make some approximations even with an analog scope.  TinselKoala often does manual power or energy calculations from digital pictures of scope displays.  It would remain a significant challenge.

One trick would be to substitute the charging battery with a largish cap in parallel with a variable resistor.  The higher the near-DC voltage across the cap, on first inspection you would expect to get a slightly higher power output reading.  That's because the higher the cap voltage the shorter the output current pulse and therefore the less energy lost in the diode.  Note if you switch to a battery or virtual battery done with a cap and a variable resistor, then the diode does not count as part of the output.  The load is on the other side of the diode.

If you use a fresh source battery where the voltage does not droop too much then you could probably get a decent input power reading.

None the less, these measurements would all have to be very carefully made.  You really and truly need to know what you are doing.

MileHigh

[Dark Alchemist]

It might be somewhat easier and you could make some approximations even with an analog scope.  TinselKoala often does manual power or energy calculations from digital pictures of scope displays.  It would remain a significant challenge.

One trick would be to substitute the charging battery with a largish cap in parallel with a variable resistor.  The higher the near-DC voltage across the cap, on first inspection you would expect to get a slightly higher power output reading.  That's because the higher the cap voltage the shorter the output current pulse and therefore the less energy lost in the diode.  Note if you switch to a battery or virtual battery done with a cap and a variable resistor, then the diode does not count as part of the output.  The load is on the other side of the diode.

If you use a fresh source battery where the voltage does not droop too much then you could probably get a decent input power reading.

None the less, these measurements would all have to be very carefully made.  You really and truly need to know what you are doing.

MileHigh

From everything I have read people are putting .1uf to 22uf in parallel with the battery to help stabilize the circuit from a battery that would be constantly draining.  For me I would kill the battery and use a benchtop power supply as the source voltage and current then go from there.

You are correct you need the tools and the know how, or a lot of reading and patience, to get this done but the tools are mandatory.

[TinselKoala]

You measure the input power like this:
You charge a good quality capacitor, like 2 Farads, to the top running voltage of your JT, like 1.500 V. Then you monitor the voltage on the cap with an accurate voltmeter, and you monitor the time with a stopwatch. You turn on the JT and let the thing run until the cap is down to, say, 1.200 volts, and you record the time this took, say it was 20 seconds (or whatever).
The energy you put in is the difference between the energy of the cap charged to 1.500 v and 1.200 v. The average power is this energy divided by the time.

You measure the output power like this:
Say your JT is powering an LED. You measure the LED's light output in any of several ways: photoresistor, phototransistor, intensity-frequency converter sensor chip, etc. You have previously calibrated your sensor using DC power to the LED. So now you can get instantaneous power output readings during your input power measurement tests for comparison.
Or, you can use common methods to charge an output capacitor through a diode. Again, the time it takes to charge an output cap from one voltage up to another voltage will give you a total energy out and the average power during the time interval.

The efficiency is then the ratio of the output power to the input power. Or, the ratios of the energies in and out during a specific time period.

You don't even need a scope for this, just a couple of accurate voltmeters, and some accurately known capacitances, and a good timing method.

[Pirate88179]

You measure the input power like this:
You charge a good quality capacitor, like 2 Farads, to the top running voltage of your JT, like 1.500 V. Then you monitor the voltage on the cap with an accurate voltmeter, and you monitor the time with a stopwatch. You turn on the JT and let the thing run until the cap is down to, say, 1.200 volts, and you record the time this took, say it was 20 seconds (or whatever).
The energy you put in is the difference between the energy of the cap charged to 1.500 v and 1.200 v. The average power is this energy divided by the time.

You measure the output power like this:
Say your JT is powering an LED. You measure the LED's light output in any of several ways: photoresistor, phototransistor, intensity-frequency converter sensor chip, etc. You have previously calibrated your sensor using DC power to the LED. So now you can get instantaneous power output readings during your input power measurement tests for comparison.
Or, you can use common methods to charge an output capacitor through a diode. Again, the time it takes to charge an output cap from one voltage up to another voltage will give you a total energy out and the average power during the time interval.

The efficiency is then the ratio of the output power to the input power. Or, the ratios of the energies in and out during a specific time period.

You don't even need a scope for this, just a couple of accurate voltmeters, and some accurately known capacitances, and a good timing method.

I think this post should be preserved in a thread of it's own so folks can reference it.  This method is simple but yet accurate and all of us playing with JT's should use this.  (In my opinion)  I always thought that using caps was a good idea but, I never really knew how to do it...until now.

Thanks TK.

Bill

[MileHigh]

Using a cap for the load is fine for a JT-type circuit because the output comes from a discharging inductor.  However, be careful and don't take it for granted that a capacitor will work as a load in a generic sense for other circuits and measurements.  Note that how much power goes into a load is determined by the impedance of the source compared to the impedance of the load.  Capacitors are not a constant impedance so you could be in a situation that's fraught with problems.

How you make your measurements is in a way related to how far you want to "stray" from the original JT circuit.

Anyway, the fundamental thing at play in a Joule Thief is that you charge an inductor and then discharge it into a LED and repeat.  There are losses when you charge the inductor due to wire resistance and similarly you have the same wire resistance losses when you discharge it.  There are also losses due to the hysteresis loop of the core.  The other thing is the quasi "golden rule" where you don't want to energize the inductor for too long to minimize the resistive losses.   A very similar technology is used in computer switching power supplies and small DC-to-DC converters for printed circuit board use.  They are like a high frequency JT charging an output capacitor, and there is an output voltage sensing circuit that is used in a feedback loop to control the switching and keep the voltage output constant.  They can be greater than 95% efficient.

Just look at a modern motherboard where the processor socket is like this special throne with various voltages ready and waiting for the chip.  It's surrounded by capacitors to give the CPU chip a happy "nest" where the voltage is stabilized for "the beast."  Most PC hobbyists are familiar with tweaking the various motherboard voltages to squeeze out the maximum performance.  So it's probably fair to say that a motherboard has a whole bunch of "smart" JT-like circuits to keep the processor beast happy.  The JT chips are probably on the "I-squared C" bus and wait for their commands from the BIOS.  A modern motherboard is quite an amazing piece of technology and one we (happily) take for granted.

MileHigh