Lasersaber strikes again. A joule thief king ?

[xee2]

I saw that my electrolytic capacitors (one 4700 µF and 50 V, the other 4700 µF and 25 V) have a leakage current of about 30 µA. This means that I can never go below a 30 µA power draw when using these caps.

The 10,000 uF 25 volt electrolytic capacitors I used were definitely below 1 uA leakage, so maybe yours are not as bad as you think.
This a Joule thief using less than 2uA that runs for over 1.5 hours from a charge of 1.36 volts:
http://www.youtube.com/watch?v=8T9HQkDnIuU
It could not do that if the capacitor leakage was over 1 uA.

[conradelektro]

The 10,000 uF 25 volt electrolytic capacitors I used were definitely below 1 uA leakage, so maybe yours are not as bad as you think.
This a Joule thief using less than 2uA that runs for over 1.5 hours from a charge of 1.36 volts:
http://www.youtube.com/watch?v=8T9HQkDnIuU
It could not do that if the capacitor leakage was over 1 uA.

@xee2:

Thank you for posting your interesting circuit. I saw it in 2011 and even filed it in my circuit collection, but I have completely forgotten.

Like always, everything has been done before.

Greetings, Conrad

[conradelektro]

Astable multivibrator with a MAX931 comparator drives a red LED with 12 µA at 2.5 Volt (LED rather dim):

The circuit without a LED needs just about 5 µA at 2.5 V to 3 V to create a nice square wave signal. Frequency and duty cycle can be adjusted by two resistors. The 1 M resistor defines frequency and the 5 K resistor duty cycle.

I have chosen a frequency of about 120 Hz and a duty cycle of only about 1 %. With this narrow square wave from the comparator output a red LED is driven through a 1 K resistor to reach this low power draw.

By removing the 5 K resistor between diode and 10 nF capacitor the duty cycle can be changed to 50%. The LED becomes of course brighter and power draw goes up ten fold. (See the measurement stated on the circuit diagram).

I found a similar circuit on the internet http://www.discovercircuits.com/DJ-Circuits/1HZOSC1.htm, but I prefer the MAX931 because it has a bigger footprint which makes it easier to handle than this awfully small components like the LMC7215.

This is not useful as a LED driver (a Joule Thief would drive a LED with 1 Volt instead of 2.5 V) but might be a useful circuit to generate a square wave with very low power draw (5 µA at 2.5 Volt). With this very "cheap" square wave one can then switch a MOSFET which in turn drives the primary of a transformer to light a 220 V LED lamp. One can select an efficient frequency and duty cycle. I will try that tomorrow.

Greetings, Conrad

P.S.: Some might want to learn more about multivibrators based on an OpAmp:

http://www.expertsmind.com/topic/operational-amplifiers-and-their-applications/astable-multivibrator-using-op-amp-comparator-918025.aspx

http://www.daenotes.com/electronics/digital-electronics/astable-multivibrators-working-construction-types

http://www.electronics-tutorials.ws/opamp/op-amp-multivibrator.html

[Vortex1]

Thank you Conrad, nice to see power input, output and duty cycle data.

R-C relaxation oscillator circuits such as the Max930 are necessarily dissipative i.e a part of the energy used to charge the capacitor through the resistor is never recovered. 1 Meg and 10 nF RC is not much dissipation. You also have the wasted bias current of the comparator to deal with.

A different approach is to charge an inductor directly from a voltage source, then dump the inductor energy into the LED. The inductor will act as a near pure current source, so LED threshold voltage is easily exceeded, and current to the LED is limited only by the winding resistance resulting in a high peak current for a very short time. LED's appear very bright to the eye, although average current is low.

Most blocking oscillators such as the JT circuit attempt to do this, however the core must begin to saturate for the circuit to switch off and this dissipates power in the resistance of the winding.

A better way is to switch off before saturation and control the cycle by cycle charge into the inductor.

The Lasersaber circuit is unique in that it uses the small interwinding capacitance of the first and second windings to couple the switch on / switch off current to the base of the transistor allowing for a fast charge / discharge of base current, hence efficient switching of the transistor. This is less dissipative and provides stiffer peak current base drive than most JT circuits which use just a resistor. The interwinding capacitance also effectively limits "on time" by discharging, thus preventing saturation, hence the high switching efficiency.

The XEE2 circuit adds a small capacitor across the bias resistor to effect fast switching. This is commonly called a speedup network in the art. There are even better ways to effect base drive, depends on the goal, which is never carefully or clearly defined.

There is more, if anyone is interested.

[conradelektro]

A better way is to switch off before saturation and control the cycle by cycle charge into the inductor.

Well, nice idea, how can one do this? Do you have a circuit?

I think I want to do that with the adjustable narrow square wave from the MAX931 circuit. It is true that the 5 µA are lost, but it is very little in comparison to the energy needed to drive a 220 V 10 Watt LED lamp.

Keep the circuits coming, idle talk is useless.

Greetings, Conrad