DIC - Current fed double inductor push-pull and other current fed topologies

[Miyuki]

Hi folks,

I want to ask is you have experience with current fed converters.

I want to build 24V to 350V 1-2kW converter and current fed look more suitable for this high current low voltage inputs than classic voltage fed
Im currently reading some papers about this topologies and double inductor push pull look most suitable for my power level and demand at i have just two switches and simple transformer without center tap at high current side

[jbb]

Prasimix has been working on a step down converter using this topology here.

What sort of voltage adjustment do you need? (Eg to handle battery voltage changes?). I ask because a resonant converter (eg full bridge LLC) could offer a higher frequency and thus smaller converter. But resonant converters generally don’t do well with widely varying voltages.

[Miyuki]

Voltage vary by 8s LiFePO4 cells so 19-28V input

[prasimix]

Hi Miyuki, CF-DIC topology is inherently good for step-up applications. You can easily achieve needed 1-2 kW and goes way beyond that if you are good in power electronic design. Possibly a good starting point for research is PhD thesis High Efficiency Power Converter for Low Voltage High Power Applications.

Also I'd like to highly recommend usage of SiC diode on the output. As controller you can use LM5041 or MCU that provide overlapped gate signals. You don't need buck input stage (as is done in my converter) if it always has load connected. You can also define which is min. load below which one it will be stopped. Voltage regulation can be achieve it that case with variable overlap, that will appear on the secondary side as PWM.

[Miyuki]

I also wonder if when I use voltage doubler at output will I need any way to avoid imbalance like modulate each side duty separately or implement some peak current limitation
Or better to avoid it

[David Hess]

Tektronix used off-line current fed switching power supplies in their 24xx series of oscilloscopes with a buck converter (sort of) providing a current to a center tapped transformer with two switches.

[jbb]

It looks like a reasonable converter for your application. You’ll have some pretty big currents but not crazy high.

I guess - guess - that peak current mode control could be good for this topology, using A CT for each MOSFET. You’ll have to check for literature.

On the output side, SiC diodes are really good. They would drop around 1V under load. For a full bridge rectifier that’s 2V out of 350V, or about 0.6% efficiency drop.

Again you’ll have to check the literature, but I don’t think there are big issues using a voltage doubler output here from a control perspective.  However, leakage inductance in the transformer, MOSFET stage and output rectifier with be critical, so tight layout is needed.

[prasimix]

I can confirm what @jbb mentioned about importance of proper PCB layout. Also transformer inductance should be as high as possible for this topology with smallest possible number of turns (to keep leaked inductance low that rise with the square of number of turns) and tight/hard coupling between primary and secondaries. That asked for core with high permeability but side effect is its sensitivity to DC component/bias. That's the reason why I've been instructed to use high performance VAC Vitroperm 500 F core.

[Miyuki]

So Im now doing some simulations at SPICE and what I see:

Output rectifier when in bridge mode dont see any rapid current or voltage swing due to primary leakage inductance, so common diodes should be fine
I need some way to limit output power as duty cycle cannot be lower than 0.5 so Input BUCK will be needed
Must do some tuning, if I put it in like transformer have lprim = 186uH and lleak = 150nH then I have something like 12W at snubber burning at 900W output tho it is not so bad if I take it as percents 1.3%

//edit:
And when Im looking for suitable components Why are current transformers for higher current (like 100A) so expensive 
And why are no cheap high side drivers with integrated charge pump (no I dont want to pay 5$ for it)

[jbb]

Regarding diodes: the SiC diodes are good for reverse recovery - which may cause nasty current spikes during MOSFET turn on. I’m not familiar with the details of this topology so maybe I’m paranoid.

Regarding a buck stage: that’s a lot of extra stuffing about. Why do you really need the buck stage? Do you want to be able to output a big voltage range? Or is it just about charging up the output bulk capacitor to 350 V at startup?  If it’s about startup, I have an idea that might work out.

Regarding leakage: it’s not just in the transformer, it’s also in the MOSFETs, diodes, output capacitor and PCB layout.  The principle effects are to make voltage spikes on the MOSFETs during turn off (bad), reduce current dI/dt (not bad) and ring against stray capacitance in the circuit (can sometimes be bad).

On gate drivers: usually they don’t have charge pumps on because it’s often possible to use the main power switches and a diode to make a bootstrap circuit.

[Miyuki]

Buck stage will be needed not only as soft start but also to regulate overload and shut down as input inductors have relative high stored energy

I must do some layout and at least simulate it from layout expected values, but now it looks like nice simple converter for my purposes with relative low loses at semiconductors

[T3sl4co1l]

Note that the transistors explode if they both turn off at the same time.  A current fed inverter is in its off state when that current is held at 0V (both switches on)!

This is obviously in conflict with the voltage source shown, so be careful.  TVSs to absorb the inductor energy in shutdown (fault), or a current source (usually a buck stage), are needed.  (To use a single buck, use one inductor to supply current to the CT of the transformer.  A TVS or snubber may still be needed to deal with transformer leakage, but this is considerably less than the main series inductance, no big deal.)

Tim

[jbb]

Well, my precharge idea didn’t work out nicely. (It involved a secondary pair of switching FETs with series resistors but dissipated way too much power.)

I still think it’s a shame to add a whole buck stage - for high current too! - to precharge the output.  Are you planning to use a precharge resistor for the input capacitor?  Maybe one 24V inrush limiter can be used to precharge both input and output cap? (Also there should be a fuse somewhere in the battery circuit.)

Thinking about shutdown, we see that we need to absorb all the inductor energy, but not very often. So an overvoltage snubber circuit (2 diodes, capacitor, substantial TVS) may be sufficient.