CCFL transformer

Can you use a C-W multiplier?

For low current, you can do resonant tricks too.

It's easier to use a voltage doubler or tripler that it is to find a multi-section former off-the shelf. The occasional high voltage power supply that I've dismantled clearly used proprietary formers, as do the Coilcraft parts

I suppose one could use self-bonding wire to make a series of self-supporting pancake windings, but I've never heard of anybody doing it.

The Baxandall configuration is definitely a resonant trick, and copes with the interwinding capacitance by resonating it with the winding inductance.

There's nothing "low current" about it, but if you are working at higher currents and powers you can justify even more elaborate switching arrangements.

Jim Williams talked about it a lot - application notes AN45, AN49, AN51, AN55, AN61, AN65 - but described it as a "a current driven Royer inverter" which is simply wrong.

MOSFETs work better as switches than bipolar transistors, and don't seem to "squeg".

When there are a lot of turns, 3xFEP is not the way to go. Too bulky. The three layers is basically a trick to avoid creepage requirements for pri-sec safety. That's not what you need here.

Anything off the shelf will be a great time saver.

RL

I sometimes use an autoflyback stage with a DRQ-series dual inductor, followed by a c-w multiplier using sot-23 dual HV diodes. That's cheap and easy, given a reasonable supply voltage, like 24.

You can also just buy a potted HV supply and move on to design something else.

There are also potted c-w bricks, but they are a lot more expensive than buying the diodes and caps.

Custom magnetics only makes sense at high volume, or for real exotica like transmission-line transformers.

I'm about to embark on a custom tapped inductor and I'm not looking forward to it. Drawings, quotes, revised drawings, more quotes, samples, tests, released drawings, MOQs, all that.

A C-W multiplier gives DC. I think CCFLs should use AC not DC, to prevent fast blackening of one end.

Arie

I'd look into Pi Winding HV transformers.

.

Lots of construction details from the Radio Amateur community.

Joe Gwinn

Yes, that minimizes the volts per layer, which in turn lowers the capacitance effect. You can still get custom bobbins to stack windings like that if you choose to get higher voltages.

Cheers

The Coilcraft data sheets don't say anything much about the resonant frequencies of their transformers - except "The FL Series of transformers is designed for use in cold cathode fluorescent lamp (CCFL) power supplies at operating frequencies up to 100 kHz" where the "up to

100kHz" gives them a lot of wriggle room.

A primary inductance of around 50uH with a 100:1 step-up implies a 0.5H secondary inductance. 10pF parallel capacitance would give a 71kHz resonant frequency, which is less than 100kHz.

Of course once you have one of the Coilcraft parts you can measure the resonant frequency.

I am using it for pulse generation, so cannot use a capacitive doubler

Yeah, sounds like difficult to do in production

I have ordered some parts, will arrive today and will measure the resonance frequency. I would expect it to be a lot higher than 100kHz, but we will see :-)

Yes, probably better to just wind it with distance between the turns

You could print a bunch of them - pairs would be nice for both legs of the transformer - mount them on a mother board, and feed the U core halves through the array.

Mechanical stability would be dire, but you could glue on a supporting structure.

Still messy to put together, but 3.5kV is always difficult to handle.

Interesting article, thanks.

I cannot use capacitive doubler thingies, there's no output DC voltage.

I need it cheap, so cannot buy COTS

I have almost never used standard magnetics. Most projects has requirements that dictate custom, or in higher volumes price is the key.

Yeah, boring work.

Stress between turns is limited by v/n limit of core. It's layer stress and section stress that you have to deal with. That's what the multisection bobbin and pancake windings do.

If the CCFL transformer will allow only 1600V, imagine the precautions required for 3x that stress. I'm not sure you can avoid vacuum impregnation / potting in anything 'small'.

RL

What input do you have, and what output do you want?

I have 24VDC in, and 3.5kV peak pulse with 10% duty cycle at 100kHz. I have a working circuit, but just a little too much loss due to the transformer parasitics, so working to change the construction to get the optimum transformer design

Yes, but there will loss associated with the distributed capacitance between turns, that's why I am trying to reduce that one also

I am actually working on an alternative idea, using 2 CCFL transformers, since as you write they are normally rated for 1600V. Incidentially, the

1600V is the start voltage, runs steady state at 600V.

The idea is to parallel 2 CCFL transformers primary winding, and series connect the secondary windings. Then connect the center tap to GND, that way I get -1600V and +1600V, total 3200V without violating the ratings of the transformer

I see trandformers claimed for 15kV with 6 sections of loose wound magnet wire. This is however Chinese consumer marketing kilovolts.

Given the ~2mm spark gap it suggests something less than 15kV