Blocking Ringer

[Slider2732]

After blowing several LED's on 1.5V using Lasersaber's newest circuit, I decided to try a few things out, that didn't use the pot coil method and with the primary aim of minimizing power needs.
One idea was a simple change to a regular blocking oscillator...to run the LED's in a similar series way as LS does and to incorporate the distanced coil geometry that TinMan showed with his Cool Joule. A pair of 1000uH inductors suited the idea well.
The whole idea of this is to create a very effective scavenging circuit, rather than a flashlight.
Perhaps using several sources, such as heat, vibration, sound, RF and other normal ambient environmental energies.
The result so far, is what I call the Blocking Ringer.

Performance specs:
Using a 1.5V battery and a 100uF cap, the light output run is up to 53 seconds of constant light.
Using a 3300uF cap, the run is 23 minutes.
Using a 'calculator' type solar panel a foot or so away from the bench CFL to charge the 3300uF over 5 minutes, the run is similarly 23 minutes.
(free light energy to charge it up is also appealing)

Here is an 'Unlisted' YouTube video uploaded just now showing such a test run, using a panel from a 'Dancing Flower'.
The measured voltage from the panel is 1.326V.

I intend to try an MPSA18 on the circuit and then probably do a public video...got some on order at the moment.
With such a solar panel connected, the circuit kicks on easily in ambient room light. It will fire up fine using the light from a PC monitor and even the light from a single LED solar garden light from several inches away.

Quick video to demonstrate the garden light powering it.

Attached is the circuit diagram, plus a list of various transistors tried so far using a 100uF capacitor and their run times.
The C1815 'GR' has put in the best results.


All help or comments appreciated

[Slider2732]

For clarity:
C1 is the run capacitor of choice, from 1uF upwards, though it will run without one.
C2 is 0.012uF, '123' written on it. Higher and there is flicker present, lower and the scavenging abilities suffer.
R1 is 1 Meg Ohm, same as C2 the increase or decrease affects running properties.
L1/L2 are 1000uH axial inductors.
D1/D2 are the LED's, red and green here due to using what's left on the bench.  
TR1 is a C1815 of 'GR' current gain rating.

[Slider2732]

The MPSA18's have arrived.
Kinda mixed thoughts really. Sure, the current gain is probably 5x the C1815, but the run time is far less.
I had presumed that the high current gain was LS's reason for using them and that at low levels of remaining power they would trundle on for a lot longer.
But, the upshot in this circuit, is that they put in a 32 seconds run at 100uF using the same AA as earlier runs. There is far more flickering earlier on by far than when using a C1815 GR.
...making them equal 4th place with the common as muck C945 !

Well, at least that is known now

[Farmhand]

Hi Slider, I think the main reason for using the MPS18 transistor, at least for me (when they arrive) is that the turn on voltage is very low, so they could be run down to a lower voltage. And when we have a voltage that rebuilds, like in a depleted AAA battery that means the voltage won't need to rise much after being drawn down before the transistor triggers again.

Here's the deal as I see it, unless there is more energy wasted as heat in one transistor over another, then input is relevant to output, one transistor might work different with the same other parts as another one but work much the same when tuned to do so with other value parts.

With my experiments I measure the power consumed by the LED's and the input power then work out efficiency, run times are just a matter of reducing input with a corresponding reduction in output.

I use a two transistor single coil oscillator one PNP and one NPN, the NPN being the main switch, but i have parts on the way to replace the PNP transistor with a single logic gate chip that will work down to under 0.5 volts together with a MPSA18 or a small SOC mosfet it should give very good efficiency, better with the mosfet.

Even with a MPSA06 NPN and a 2N2907 PNP the circuit runs the battery down to 0.76 volts under load. with the logic chip in place of the PNP and using a MPSA18 or a small mosfet it could run down to under 0.5 volts for sure. I have a Joule thief running from 0.48 volts but it flashes the 2 x 5 mm LEDs, but they do flash bright and fairly quick. I use it to run from a galvanic cell and flash light on my tomato plant under electro-horticulture experiment.

Much easier to wind a good quality coil if it only requires one winding.

Using a Logic chip and mosfet one can also set up a adjustable astable oscillator and very efficient switch rather than a feed back setup.

Using one CD4049 I set up a secondary circuit to run from the boosted 3 to 5 volts output of the "Very Low Voltage" circuit, the CD4049 is a hex buffer/inverter with six gates, I used three gates to build one oscillator, one on each side, one oscillator runs at 100 kHz and switches the transistor on for 2.5 uS and off for about 7.5 uS, the other oscillator runs at 50 Hz and gates the first signal off for 50 % of the time, so the actual duty of the oscillator driving the transistor is 12.5 %. With that I can get 10.75 volts DC across 4 x 5 mm LED's, 1/8th duty. And DC across the LED's so the output power can be measured.

Cheers

[Slider2732]

I'm delighted to see you are a member here, having enjoyed your much larger voltage experiments on another forum  

Yeah, a horses for courses type of thing. Put a world class steeple chaser on a 5 Furlong flat race and it'd have no chance against the other entrants.
I agree with the 1 winding thoughts and that brings in solar garden light circuits and their single inductors.

Speaking of the CD4049, or at least related, I built a small music synth a few years back with 4 oscillators. Several different chips were used, to hear if there were differences.....a TC4011BP, MC1455, CD4093 (2 sections).
Pic attached of the now dusty oscillator section. I'd have to say it sounded more like Scottish bagpipes with all 4 cranked up  
 
The more standard blocking oscillator circuit has a single LED between Base and Collector. Using a C1815 it will start up around 0.45V and does indeed flash. Have built hundreds of them and Lidmotor's Penny is pretty much the same thing. The big difference from my point of view is when decreasing the Base cap to around 0.01uF and having 2 or 3 LED's in series from Collector to 1 axial inductor.
30 seconds used to be a good run from 100uF when flashing, but now it's almost doubled and is a solid light output until complete fade out.

The next thing really will be to wind up some coils, these always do run better with actual wound coils but the axial inductors are a heck of a lot easier to plonk into the breadboard. Also to try LS's pot core coil type on this version