• ### Redesign

Unfortunately, my original plans with a simple joule thief just didn't want to work, so I redesigned the whole thing. I focused on a higher output voltage, allowing for a lower capacitance and less strain on the joule thief. To accomplish that, I had to make a high voltage joule thief - a joule thief with a third, high voltage coil.

I wound the transformer and, sure enough, it worked like a charm, so the only thing that was left was to shoot the circuit in action and document the process of making it.

The circuit now produces an output voltage of 22 Volts with the capacity of 10 000 uF, which makes for 220 mJ of energy, which isn't only enough to burn LEDs, but is also enough to weakly spot-weld wires to each other, as I have accidentaly found out.

You may notice a few updates: an updated components list and building instructions. There should also be a link to my youtube demo of the circuit.

Good luck to the other contestants for the Coin Cell Challenge!

• ### New cells - a whole new dimension

Today I went to my local hardware store to fetch some new cells. I was surprised at the price of the them (3.5€ for 2 CR2032), but it was definitely worth it. As I'm writing this, a new coin cell is not only pushing the capacitor well over the 2.44V barrier of yesterday, it can manage 6V without any trouble! If I change the coin cells in the middle of the charging process, I might be able to achieve even 12V!

However, I realized I have another problem: at 6V, he circuit isn't powerful enough to explode just any LED. I hope that it will be able to handle at least the small red ones, while needing at least 9 or 12V for the big ones.

• ### The circuit and its operation

After I decided what would be my entry to the challenge, I had to plan and choose the way to achieve my goal.

To blow up a LED, my circuit has to:

1. Accumulate energy from the coin cell
2. Convert the energy to a higher voltage
3. Release the energy into the LED

To satisfy the first point, I decided on a 120 000 uF Powerlytic capacitor from Sprague, which can pack a lot of energy. My initial tests with it have shown that, when charged to 5V, it can easily destroy a yellow LED. On the other hand, I barely believe that I can charge it that much with only a coin cell.

The second point calls for a simple, yet efficient switched mode power supply. After exploring many options, I've decided that an integrated circuit is much too complicated and will be very inefficient because of the weak power source. The next option is a circuit made from discrete elements. I've also come to the "too complicated and inefficient" conclusion, so at the end, I've settled on a simple Joule thief circuit with nothing but a 10:10 toroid transformer, a 1k resistor and a 2N3904 NPN Transistor. It's not the best choice, but it won't be a problem for my poor electronics construction skills. It also won't be hard to repair if I fry any part of it (I already did, in fact, destroy one transistor and I am happy that the circuit is easy to repair). Because joule thieves are very noisy and kind of ACish in terms of output waveform, I added a diode at the output to prevent polarity reversal on the output capacitor.

To release the energy into the LED, I used a simple momentary switch I found in my component box, coupled with a 2 pin female solderless terminal. The LED goes into the terminal (It's the kind that grabs the leads you put into it) and gets fried at the push of the button. Simple, huh?

Today, I finally built the circuit in the "rat's nest" technique and gave it some tries. The result were neither good nor bad. The joule thief that happily output 40 V peak-to-peak when unloaded, behaved quite differently under the load of a HUGE capacitor. The voltage was rising, of course, but not as fast is I have hoped. It has taken about 15 minutes to get to the previous 5V. The fact that I burnt the main transistor ind had to replace it wasn't helping, either.

Results were even worse when i switched from my bench power supply tuned to 3.6V to a real coin cell: even as I'm writing this, the voltage hasn't rised from 2.44V in the capacitor for half an hour!

The only light at the end of the tunnel at the moment is buying new cells, as the ones I used today weren't new nor full, so tomorrow I'm going to buy some at my local hardware store and try again. Expect another log entry tomorrow describing my second test.