Very wide range input power supply

[NiHaoMike]

Here's an interesting design puzzle of a power supply:
Input: Up to 520V DC, 320-400V typical operating, minimum as low as practically attainable. Minimum startup voltage should also be as low as practically attainable.
Output: CVCC 9-12V or 12-16V for charging 3S or 4S battery, maximum current up to 20-30A if input voltage 300V or higher, below that output current drops accordingly. Both voltage and current limits would be programmable by external controller.
Efficiency: 85% or better when input voltage in typical operating range and output current is 4A or more up to 80% of maximum.
Other considerations: If input power is insufficient for sustained operation, repeated startup attempts must not damage supply.

The application is an auxiliary power supply for a solar power control system. I'm thinking it would make sense to modify an ATX PSU (with active PFC) for that application, but extending the  maximum input voltage to 520V seems a bit problematic.

[Neomys Sapiens]

Also for you: Microchip LR8K4. 

see here:
https://www.eevblog.com/forum/projects/high-voltagelow-current-power-supply/msg3006370/#msg3006370
Sorry - did not see the 4A requirement. Proposal without sense.

[David Hess]

Universal input ATX power supplies can operate from 340 to 130 volts DC so in theory adding a simple inverter to convert 520 volts DC to 340 volts DC could work and allow operation down to 200 volts DC however I would not trust any ATX power supply to operate under adverse conditions like low input voltage or current limiting.

That leaves designing something like an ATX power supply without the active power factor correction stage and including both input and output current limiting.  I would also include shutdown for high and low input voltages.  For safety, I would include galvanic isolation from input to output in excess of that required for 240 volt AC applications; DC can be nasty.  Maybe it would be better to convert down to a standard low voltage like 48 volts DC and then implement or use an existing charger.

[SiliconWizard]

Not sure what kind of solution you're looking for exactly. Full design? Well, if you're thinking of reusing/modding an ATX PSU, I guess power modules could be acceptable too.

Vicor comes to mind. http://www.vicorpower.com/dc-dc/isolated-regulated

They have modules with 9V – 420V input.
520V, couldn't find anyting, but you could think of adding a front-end to one of those, or something.

[Marco]

I doubt you're going to save any time trying to modify something existing. First you have to DIY most of the high power DC input, then you have to do the CVCC control part. Between those you could probably DIY it entirely.

No experience with it, but since you don't need isolation how about a resonant switched capacitor step down converter? No transformer needed.

[NiHaoMike]

No experience with it, but since you don't need isolation how about a resonant switched capacitor step down converter? No transformer needed.

It does have to be isolated, hence why I'm thinking modifying an ATX PSU might be better than trying to build it from scratch.

[jbb]

Hmmm

For big conversion ranges, something with buck-boost capability seems like a good idea.

How about a buck converter feeding a current-fed isolated boost converter? Prasimix was working in one here

Some nice features include the ability to soft start and good limiting of the  output current even in difficult situations.

[David Hess]

For big conversion ranges, something with buck-boost capability seems like a good idea.

The conventional solution is a transformer fed buck converter in the form of a half bridge, bridge, or forward converter.  The transformer ratio is adjusted so that at the minimum input voltage, the maximum acceptable duty cycle produces the output voltage and then the duty cycle decreases as the input voltage rises.  The input voltage range is only limited by the acceptable change in duty cycle.

Besides extremes in duty cycle, the problem with wide input range designs is that the power switches have to withstand high current at low voltage and high voltage at low currents making them larger than would otherwise be required.

[voltsandjolts]

Input: Up to 520V DC, 320-400V typical operating, minimum as low as practically attainable.

Where does that 520V come from? I mean, what is the power source?

[NiHaoMike]

Where does that 520V come from? I mean, what is the power source?

A string of solar panels, the 520V figure is the design max of the main inverter module (the worst case open circuit voltage of the string must be less than that), so I would prefer the auxiliary power supply be able to handle that.

Besides extremes in duty cycle, the problem with wide input range designs is that the power switches have to withstand high current at low voltage and high voltage at low currents making them larger than would otherwise be required.

Hence why the output power available decreases when the input voltage decreases below a certain point.

[Marco]

Personally I'd just drop the voltage to 140 and then use an off the shelf mains fed high power balance charger. Do you really want to rely on battery protection circuits while charging at 30A?

This allows you to work all the way down to 140 volt, just need to design a non isolated high power step down converter.

PS. well I guess the balance charger should have some kind of programming input so you can drop the charging current or restart it when the power drops out.

[NiHaoMike]

Some BMS boards, most notably ones designed to allow a LiFePO4 pack to be used as a drop in replacement for lead acid, are in fact designed such that the charger can be a simple CVCC type.

[jbb]

A string of solar panels, the 520V figure is the design max of the main inverter module...

...hang on.  These batteries you want to charge... are they used to feed an inverter in this solar system?

If this is some kind of grid tie + battery situation, how much control do you have?  There might be interesting ways to address the whole problem (e.g. panels connected to DC link via boost MPPT.  DC link range would be somewhat constrained (e.g. 400 - 520V) and reduce voltage span requirement for auxiliary converter.  Auxiliary converter could be bidirectional to support both charging and discharging modes.

[NiHaoMike]

...hang on.  These batteries you want to charge... are they used to feed an inverter in this solar system?

If this is some kind of grid tie + battery situation, how much control do you have?  There might be interesting ways to address the whole problem (e.g. panels connected to DC link via boost MPPT.  DC link range would be somewhat constrained (e.g. 400 - 520V) and reduce voltage span requirement for auxiliary converter.  Auxiliary converter could be bidirectional to support both charging and discharging modes.

System architecture is solar panels supplying the main bus that can go up to 520V with no load current (could do per string or per panel MPPT later on but planning to just have blocking diodes at first), on that bus there's a high voltage DC/DC converter to interface with a high voltage battery (that is optional), as well as an inverter for a 3 phase compressor motor, another inverter for grid tie, and yet another inverter for backup. Basically, I'm repurposing a cheap Prius inverter module into something similar to a Sol-Ark inverter (in particular, the zero export feature that supports external sensors), but at 1/10th the cost after including all the other parts needed to build the complete inverter.

I can't just power the auxiliary supply from the high voltage battery interface since (without the high voltage battery) there would be nothing to get the DC/DC started in the first place without an auxiliary supply. I could bootstrap the whole system from the low voltage battery but I would prefer to design it so that failure of that battery would not make the system unable to start.