Another joule thief.

[confounded]

this was the 1st thing i've ever soldered together elctronics wise, amazed it worked 1st time too. If i stick the capacitor in, how can i measure the efficieny improvement?

 

[Sceadwian]

You have to measure the RMS current and voltage in and out, which is a little on the sticky side because the frequency is 300khz or higher. (Mine was 360khz on a fresh AA) How Colin qualified his efficiency gain is unknown, I will certifiably guarantee that a single cap did not increase total efficiency 300 percent.

 

[colin55]

Measure the current before and after fitting the 22n capacitor.

 

[Hero999]

Colin how did you measure the current?

Did you use a bog-standard DVM or did you use an expensive true sinewave DVM with a bandwidth of 1MHz or a scope?

If you cheap meter or even a true sinewave meter designed for 50/60Hz operation the chances are you won't get an accurate measurement.

Because the waveform time varying the average or peak reading (what a normal meter gives you) isn't good enough, you need a true sinewave meter with a high enough bandwidth to give you the real RMS current being drawn.

The fact that the unfiltered current might read higher than the filtered current on a DC meter means absolutely nothing. Suppose the meter is just giving the peak value? That would explain why, but it wouldn't mean the efficiency is higher.

 

[audioguru]

Make two circuits without the capacitor.
Compare their brightness and how long a battery cell lasts.
Add a capacitor to one circuit and compare them again.
The one with the capacitor might look brighter and its battery might last longer.

 

[Hero999]

Or you can look at the current waveforms on a scope and work out the RMS values to calculate the efficiency.

 

[colin55]

Place a 10R resistor in one lead with 100u across the resistor and measure the voltage across it. You cannot place a multimeter in series with a lead as the frequency is so high that the leads acts as antenna and upsets the reading.

You are comparing one reading against the other.

 

[Mike_2545]

So sensitive you can't measure it?

 

[tcmtech]

Your sucking the very last bit of energy out of an almost dead battery! Does it really matter how fast you get to completely dead?

 

[Hero999]

Place a 10R resistor in one lead with 100u across the resistor and measure the voltage across it. You cannot place a multimeter in series with a lead as the frequency is so high that the leads acts as antenna and upsets the reading.

You are comparing one reading against the other.

That won't give you an accurate RMS current reading.

 

[duffy]

All these posts and nobody asked Sceadwian what his minimum discharge voltage was or the theoretical efficiency of his circuit?

The Joule Theif is an art form.

 

[audioguru]

My solar garden lights use a voltage stepup circuit. But without voltage regulation the LED is very bright when the battery is fully charged then gets dimmer as the battery discharges. When the battery voltage is 0.8V then the LED is very dim.

Didn't Colin post a circuit that has current regulation?

 

[Sceadwian]

I haven't a clue duffy =) The frequency is too high to measure on my meter and the wave form is irregular so I'm not sure how to figure it out from my analog scope. I don't really care, I was just questioning colins two unsubstantiated claims on his torch page. The first one being 300% efficiency increase with a single cap and the obviously false claim that LED's are more effcient on pulsed current. If he only took a single ended current reading for his effciency claim then that number is just made up because he would have to measure the ratio of power in to power out both with and witout the cap. He only said he measured current at one point in the circuit which can not be used to determine effciency.

 

[colin55]

Sceadwian:

false claim that LED's are more effcient on pulsed current.

If you ony did a bit more reading and a bit more research you would find that LEDs are much more efficient when being pulsed. One document rated the increase at 15 times. This is because your eye is sensitive to the peaks rather than the average.

I am not saying the LED produces more illumination when being pulsed, just the output can reach as high as 95%, with as little as 33% of the energy, when being pulsed.

 

[audioguru]

If you ony did a bit more reading and a bit more research you would find that LEDs are much more efficient when being pulsed. One document rated the increase at 15 times. This is because your eye is sensitive to the peaks rather than the average.

No.
My LED Chaser projects and a soil moisture project use PWM for dimming.
The peak current remains the same but as the width of the pulses are reduced then the visible brightness is reduced. Therefore your vision does not peak detect, it average detects.

A multplexed display is very dim when the average current in an LED is low even if the peak current is normal.

My LED flashlights (torches over there?) are very bright because they blink very high current pulses slow enough that your vision can see the peaks. 5 per 1/3rd of a second. They are turned on for 33.3ms then turned off for 33.3ms five times then a pause until the next second.

 

[Sceadwian]

White LED lumen testing - CandlePowerForums

Those are typical lumen per watt curves for standard white LED's based on the current they're driven at. Pulsing the LED's at a higher than normal current drops their light output efficiency dramatically, it will also shorten their lifespan vs constant current because of pulse heating at the P/N junction. I would also like to mention that at higher pulsed currents increase the I^2R loss due to heating, which is basic electronics 101 knowledge.

I see people quote average current/heating constantly but this is simply not true for PN junctions, if the current pulse is high enough even briefly the PN junction will be very close to it's maximum limit, even for a short period of time which will cause the PN junction to diffuse much faster than if the junction was held at a lower temperature for a longer period of time. This increased degradation at low duty cycles isn't that big of deal, you're talking about a 5 year lifespan over a 10 year one, especially in things like TV remotes which are pulsed at high currents for low duty cycles, because they're used so infrequently that you never realize this shortens their life.

I'm sorry colin but your entire precept is simply wrong, you've posted not one single piece of verifiable data to support your theory. Your theory also violates basic I^2R heat loss basics and the efficiency curve of white LED's.
Whatever math you used to come up with your efficiency claim was obviously not done with proper current measurement techniques or even the most basic understanding of how an LED PN junction works.

 

[Sceadwian]

Just as an extra little tid bit, apparently my joule thief has started to behave like a relaxation oscillator. It drew the C sized carbon cell down to about .4 volts, after 4 days it finally stopped oscillating. But the battery voltage is recovering over the period of about an hour to the point where it starts oscillating again as the voltage rises. It's done this a half dozen times so far. I can't even put scope probes on it anymore without causing it to stop oscillating.

 

[sPuDd]

I made two units - was a lot of fun. Also looked at the waveforms & behaviour on a PicoScope 5204.

The first one was 20:20 turns on a ferrite I think that was acting as a filter choke on a wire loom somewhere. Used the fine primary winding wire from a burnt out 240V plug pack (wall wart). It ran at ~16khz (1.5V) and is still draining one of the cells that’s also leaking acid. I used a 1k base resistor to limit the maximum brightness to what looked like about 20mA. It will run down to about 0.4V before falling over.

The second one was a powdered iron core from a small battery charger I think. Removed the existing wire, folded it in half and wound it back on for 40:40 turns. It runs at about 40khz (1.5V). It too is dragging down a leaking cell. I used a 5k base resistor to limit the brightness to the same as the first one. It will run down to about 0.6V before falling over.

Both were fun to make, and I suspect they would work using any old junk cores, wire, resistor & transistor laying about. I highly recommend everyone make one of these just for the fun of it. Maybe even give the garden some night lights using old cells.

Oh, I pushed the second unit to about 3V supply and it pulled ~500mA from my bench PSU before the LED changed colours and blew

sPuDd..

 

[duffy]

.4V is excellent - this is why it's called a "joule theif". It seems to steal extra joules of energy from a dead battery. Battery manufacturers consider it "dead" at around.7V loaded, but most consumer electronic junk doesn't even come close to this discharge point.

 

[Hero999]

Don't use rechargable batteries for a Joule thief, once discharged to this level they can't be recharged.