Is a high power boost converter suitable for my needs?

[treerobber64]

I am an electrical engineer, but I never took any power electronics courses. I just graduated and I am trying to do some projects to round out my understanding of electrical engineering.

I am trying to charge a bank of 400V capacitors as fast as I can. (Eventually that bank of capacitors will be connected to a linear mass driver in a later project).

For my input I will use a LiPo battery, this means I have some choices for what voltage I input depending on the number of cells I put in the battery. I plan to choose one of the following input levels:
Vin: 11.1V, 14.8V, 18.5V, or 22.2V.
Win: 3kW

Vout: 400v
Wout: 3kW – inefficiencies

My questions are:
1) Is this crazy? I understand this is both quite a high ratio, and high power requirement.
2) Is this appropriate for my needs? I have a very large purely capacitive load, everything I read assumes a resistive load. Will I cause instabilities if I try to charge a capacitor bank with a boost converter?
3) Does anyone have any information that could help me design such a circuit? (or in the case that this circuit is crazy, information about a more appropriate one).

Thanks,
Andrew

Edit: This http://www.dsce.fee.unicamp.br/~antenor/pdffiles/High_Step_Up_Converters_Part_II.pdf makes me optimistic that I can do this. Although it seems like a very complicated first project.

[mzzj]

You probably get better results with transformer coupled topology, ie full-bridge or 2-transistor forward converter.  Even then with 3kw power I would select at least 48v battery voltage.

Anyways this is not a good beginner project for power electronis, i would recommend starting with something that doesn't require such power and voltage and power levels.

[T3sl4co1l]

How much capacitance and how fast?  Are you really going to draw 3kW from that poor battery!?

Boost is good but with the caveat: huge ratios in a single stage (about 20:1 in your highest battery voltage listed) are impractical.  It's called flyback if you have an isolated winding, which can be any ratio -- let a transformer do your work.  Or if you stack the windings in series (you don't need isolation, right?), you can save a few turns and still have the same sort of behavior (it's slightly between boost and flyback, but who cares, they're close enough) without having such a huge demand on the circuit components.

An example, from a low current supply:



The transistors approximate a peak current mode controller, like a UC3842.  Except the 2N4401 "switch" is operated in an active current limit mode, not a hard switching mode, necessary to charge the series 4700pF capacitor (you don't just dump transistors full-hard into capacitors, unless you want a magic smoke generator).

The tapped inductor provides half the step-up ratio (5:1), and the voltage doubler covers the rest (10:1, so 120V output is comfortable from a 12V input).

You wouldn't want to use a voltage doubler, because it's inefficient (you don't want switches being operated in the linear range, not at 3kW!), so your inductor ratio will be much higher.

Tim

[treerobber64]

Thanks for the info T3sl4co1l !

Regarding the battery for $32 this one can provide a constant output of 3.3kW , a 10% tolerance without even relying on the burst current rating. Should be fine!

Regarding the capacitors, I have not seen them. A friend bought a box full from an industrial surplus store. He says they are roughly the same size as a 23 oz can.

I will read more about flyback converters.

[Richard Head]

treerobber64

You have a couple of options, but if you go the flyback route then bear the following in mind:

A fixed frequency flyback converter cannot detect when the core has emptied itself completey of energy and this is a problem when the capacitor bank is completely discharged. The result is that the rate of energy transfer is very slow initially and gradually increases as the voltage climbs. Ideally you want a constant energy transfer rate to minimise charge time.
The reason is the di/dt of the secondary winding of the flyback transformer is limited by the very low voltage that its pumping into. This is why a self oscillating topology has an advantage over the fixed frequency type as it ensures that the choke empties itself completey before the next cycle.
If you use a fixed frequency flyback then the MOSFET off time is way too short when the caps are discharged resulting in the MOSFET switching on and then off again almost immediately.
This is all academic anyway as the charge time is probably limited by the maximum allowable capacitor current.

For an absolute all out minimum charge time design the best approach is probably to charge the caps at a constant (large) current. This will minimise the capacitor stress.

Dick

To get the absolutely shortest charge time for the cap bank the best option is probably to charge the capacitors at a constant (large) current.

[Artraze]

Realistically, the only topology usable at such power levels is a Full Bridge or a variant thereof like a Phase Shift ZVT.  Indeed, common recommendations are <100W for a flyback and full bridge should be used at >500W.

There are two reasons for this.  First, both boost and flyback operate using energy storage.  That means you charge up a magnetic field from the input and then dump that energy in to the output.  Obviously, then, the power is limited by how much energy the magnetics can store and how quickly it can charge/discharge.   The second reason is that the full bridge topology is the only one that fully and evenly utilizes its components.  Others will add additional stresses on the switches and/or only partially use the magnetics.  For example, in a boost converter the switch must be able to handle both the low-side current and the high-side voltage.  Additionally, the inductor is only ever charged one way, meaning it must operate from one side of its hysteresis curve to saturation.  At low powers, such inefficiencies aren't a big deal.  At higher powers, however, they makes things difficult if not impossible.

Anyways, a multi-kilowatt power supply is not for the faint of heart.  It requires a lot of design work and has lots of traps for young players, as Dave would say.  It you go this route, you'll need to do a lot of learning, be it via app notes or a book.  Generally magnetic companies will be a good source of the former.  Here is one I pulled up after a quick search that looks like a good place to start.

For this application there are other options, however.  The easiest, and more efficient besides, way would be to forget the SMPS and just stack batteries until you have >400V and then charge the caps directly from that.  (Well, probably not directly and instead via a current limiting PWM, but that's a fairly simply design.)  You won't need 3kW batteries then either, as each would only need to handle its own share of the power.

If that's not an option, you can boost it a little with a voltage multiplier / charge pump.  They aren't good at regulated power handling, but since you're just charging a capacitor bank they would actually work fairly easily and efficiently.  Still, you will need to start with 4x 22.2V batteries.  By boosting the initial voltage, you can cut back on the number of stages and ripple current within each stage to something reasonable.

Ultimately, handling kilowatts is neither cheap nor easy.  You will definitely need to decide what your budget is in both effort and money before you know what exactly 'possible' means for charging your caps 'as fast as possible'.  My recommendation, however, would be to start with a 100-200W push-pull converter to give yourself an idea of what you're getting into.  It should be fairly simple and be able to use a modest sized scrap ferrite core without too many issues.