The charger is quite simply a Mazzilli driver with a small voltage detecting attachment to turn it off when some terminal voltage is reached. And instead of the usual flyback transformer, err, the transformer isn't much different either. When the capacitor voltage is zero or below the threshold the Mazzilli driver works as normal, driving the transformer. The capacitor bank voltage is sampled much like in my boost converter, except a TL431 voltage reference is used to set the reference. This gives a stable voltage reference over a wide supply range. In fact the charger should work fine up 30V with some of the resistor values (for the LEDs and transistors.) changed, giving much greater output power. When the comparator decides the voltage is great enough the 2N3904 turns on, and the gates of the IRFP250s are held low through the diodes. This effectively inhibits the Mazzilli driver, and the capacitors cease charging. The other comparator is simply used to make the bi-LED function. It can be omitted along with the bi-LED and accompanying transistors without penalty.

The transformer was wound on a monitor flyback tranformer core, four layers x 60 turns per layer giving 240 turns for the secondary. Each layer was insulated with masking tape simply to hold them together and provide rudimentary voltage stand-off. The secondary wire was scavenged from a MOT and is assumed to be 0.3mm. The primary was made with two 18AWG wires wound first in bifilar, then around the core 7 times, and finally centertapped by joining the "front" of one wire the the "back" of the other. Just remember that you want to make one long winding with these two, so they must be in serial and in phase. Where they connect is your centertap. It makes sense once you've done it, and gives near perfect centertapped primaries. At any rate I wound seven turns, which gives 7+7 primary turns. The output voltage is rectified by four fast 1A diodes. I'm unsure of the required current rating but 1A seems sufficent, and at only 600V they're cheap as well.


The charge time to 430V on a 3.29mF capacitor bank was clocked to 6.3s. So 304J in 6.3s is roughly 50W of average charging power at only 12V. That's a massive improvement over the boost converter which was running in the 10-15W range. Not only is it more powerful than the boost converter, but it's much more efficient too. While the boost converter was in the 30-40% range I estimate this charger to be 70-80% efficient. That means more mileage with batteries and less heating. Nothing in my charger is heatsinked, nor does it seem required after multiple consecutive charges. If there are problems with power the size of the resonant capacitor has a large impact. If it's too small the frequency is high and the charging power is decreased. The efficiency is about the same however, so this can be use to regulated current draw. Using a large resonant capacitor decreases the frequency and significantly increases output power, to a certain point at least. With a maximun supply voltage of up to 30V, the power can be increased much greater than 50W.