CCFL used as a Flyback Transformer in a HV DC to DC Converter

Today, I've been trying to create a HV DC supply using a CCFL transformer - FL3209-1L_ from coil craft. I want to create 500V from a 3V battery. I've been using a bench supply of 9V. I used a power N-channel MOSFET, and a flyback configuration.

Before I explain what happened with the CCFL transformer, let me digress into a background on flyback converters.

In a flyback topology, a diode on the secondary prevents current flow and so during charging, the primary looks like an inductor. If the voltage on the primary is a pulse, then the current looks like a ramp. The slope of the ramp is governed by the inductance. The current increases linearly. When the transistor switches off, the current flips direction, and so the rectifier on the secondary winding allows current to flow. The current decreases linearly in a descending ramp until it reaches zero. If the the turns ration is unity, then the two current triangles are equal. If the the turns ratio is greater than one, then the secondary current triangle is smaller. However the voltage pulse is proportionally taller.

The duty cycle should be controlled so that the primary current never saturates the coil. Most transformer specs tell you the maximum primary current, and primary inductance. So given the primary current and inductance of the primary, it's a simple matter to compute the maximum length of the on pulse, likewise the off pulse time, assuming you want the transformer to run in discontinuous mode, is equal to the 1/N times the pulse width of the primary. From the two pulse lengths you can calculate the maximum frequency, and the duty cycle at the maximum frequency.

I= TV/L

Ton(max)= IL/V

Example: Primary of 25uH, Voltage of 3V and Max Primary Current of 1A, N=12

Ton(max)= 1*0.000025/3= 8.3us

Toff(min)= Ton(max)/N

Toff(min)= 8.3us/12= .7us.

Fmax= 1/(8.3+.7)us = 111KHz with a 92% duty cycle. Or use a 50% duty cycle which would have a frequency of 1/(8.3+8.3)us= 60KHz.

Transformer saturation is problematic because it lowers efficiency. The transistor sources extra current, yet this isn't converted to stored energy in the coil. The transistor wastes this as heat. Excessive current can also damage the transistor.

The flyback generates a short HV pulse when the transistor shuts off, before the diode starts conducting. This pulse can be seen in both the primary and secondary sides. On the secondary it can be desirable because the secondary circuit is essentially a peak detector, and so the high voltage peaks raise the voltage. On the input, however, it can be a nuisance, because it can damage the driving transistor due to excessive voltage. Thus, a Zener diode in series with a diode is often placed across the primary to limit this kickback voltage.

A common application of HV supplies is the flash charger in most cameras. These convert 1.5 V to over 300V. The transformer in these supplies is called a photoflash transformer. There are several companies that make standard off the shelf transformers PCA, and TDK being two. These transformers are small, and have a 1:12 turns ratio typically.

Now back to my experiment with the CCFL transformer. It essentially failed. The MOSFET overheated. This means I was probably saturating the transformer. I tried using a smaller duty cycle at about 100Kz and this helped reduce the MOSFET from overheating, but I still didn't get a good HV output. I speculate that there wasn't good coupling between the windings. I tried reversing the polarity of the transformer secondary, to make sure it was in flyback mode, and this didn't help. I concluded that this CCFL transformer wasn't good as a low power DC to DC converter.

I next tried a DC to DC transformer from CoilCraft with about a 1:1 ratio. This generated a very large spike on the secondary, and I was able to get about 300VDC out of it. I'm ordering in photoflash transformers, and will try these out next.