Converter topology selection

[Rookie_poweran]

Hello, I want to design 150v-420v dc input and 48v output converter about 550w output power. Unit cost of converter is important. Which converter topology best fits with this kind of application?

Thanks

 

[DIY TechFusion]

If you are going to use an LLC - design it to cover the range - not too tricky - and saves a heap of parts !

Abstract:​

In server and telecommunication applications, LLC resonant converter featuring high efficiency and high power density becomes an excellent candidate for the frontend DC/DC stage. But, for conventional LLC converter, the performance will be severely deteriorated once the converter is designed for accommodating regulation over wide input voltage range.

 

[Easy peasy]

Indeed - not written by an experienced design engineer - who knows all the tricks though . . .

 

[cupoftea]

Just do a full bridge. Primary side: Do a little PCB with 4 DPAKs on it with thermal vias. Add the two gate drive transformers to it aswell.
Then screw it to a piece of metal via insulating pad.....then screw that to your PCB.
Have an overhang bit where you can couple power and gate drives to it via a suitable header type "thing" connector.
Easiest way to do a Full bridge.
Also, if you want a really easy two coil transformer which doesnt even need interleave winding, do the full bridge as a current doubler output full bridge...with sync rects on the output if you want. Youll only have some 5 amps in each of the output inductors....so you can probably find two series offtheshelf inductors for each "inductor".
Laugh your head off that you didnt even need a custom wound output inductor in the process.
Laugh even more at the amount of internet information where they say "two offtheshelf power inductors in series" is unworkable.
Spend the money you save on a nice meal out.

A full-bridge-with-current-doubler-output is effectively two interleaved two-transistor-forwards, but with just one controller.

Alternatively,
Just design it as a Full Bridge.....then swap in a phase shift full bridge controller. Then your gate drive transformers (two of them) can be really cheap ones with bipolar output and no need for turn off PNP. Give it 20vin for +/-10V out on the gate drives

 

[DIY TechFusion]

Indeed - not written by an experienced design engineer - who knows all the tricks though . . .

LLC SMPS is a very good topology, but have it's own limitations, tuning an LLC to work in very wide range is a challenge, try reserching any working application design that work in that range. That's why the main focus of LLC research no is how to achieve a control compensation to manage wide input voltage range without sacrificing output regulation and efficiency.
Been working with Power Electronics for years, knowing some trick is not enough .

--- Updated Aug 24, 2024 ---

Indeed - not written by an experienced design engineer - who knows all the tricks though . . .

Been there there, done that, experienced it, tricks will not usually lead to the best results.

 

[Rookie_poweran]

Also, , is you power source the mains...if so...then why not do Boost PFC stage up front?...then you only have vin=400vdc.
Also, you get lower current in the wires to it.
Then you may do 2 tran forward.
But it depends to some extent on your enclosure details, space constraints, and ambeint temperature where it will be run, and whether you are allowed to use fans to cool.
The UC3856 is very good but no overload shutdown with hiccuping. A simple current mode controller eg NCP1239 can give you this.....it only has one output for fet drive, but you can use a d type cct to give you the two outputs. Please ask if you want me to send sim of this in free LTspice.

You can do your own hiccup cct, and use it with uc3856, but its more components than the d type thingy.
Also, NCP1239 has UVLO.

It is solar panel output. Thank you for idea, I must check ltspice.

 

[Rookie_poweran]

Adding Vin feed forward to the half bridge ramp - is very much like adding a current ramp ( mathematically identical ) and gives instant correction for Vin ripple

It allows volt loop correction to be very good indeed - and you can add a conventional current limit circuit to the output . . .

--- Updated Aug 24, 2024 ---

The OP does not state where the DC is coming from - although it's a fair guess it's from 90 - 275Vac mains ....

The PF can be cleaned up with a choke in the AC line - or just ignored - just like millions of other 500W converters out there.

Hi, dc is output of solar panel.

 

[Easy peasy]

LLC SMPS is a very good topology, but have it's own limitations, tuning an LLC to work in very wide range is a challenge, try reserching any working application design that work in that range. That's why the main focus of LLC research no is how to achieve a control compensation to manage wide input voltage range without sacrificing output regulation and efficiency.
Been working with Power Electronics for years, knowing some trick is not enough

We routinely design converters from 80 - 400Vin - fully resonant, low rms currents - fully ZVS - good control characteristics - it's just a know how thing.

--- Updated Aug 25, 2024 ---

Hi, dc is output of solar panel.

Good to know - are you intending to incorporate PPT into the control for the converter ?

 

[DIY TechFusion]

Hi, dc is output of solar panel.

Since it's from Solar, I suggest to have a BucK/Boost MPPT Converter in the Front end - Then feed the Output to an LLC converter. 150VDC-450VDC Solar MPPT Input , 300VDC MPPT Output Then Feed to a Half Bridge LLC converter with 48V Output..

 

[cupoftea]

Also, you could just do a single Full Bridge straight off the panels, to a 48V output. Just incorporate the MPPT into the Full Bridge SMPS.

Have a micro in there to gradually increase the demand of the Full Bridge, then peg it back when the Vin starts to decrease significantly as you draw current.
The micro will continuously monitor the vin of the Full Bridge, and the input current of it....and make alteration to the power throughput thereby.

You basically have an error amplifier closed on input current, and keep adjusting the reference of this to suit. You also have an error amplifier closed on vout of course. You diode-OR the output of the error amps. and whoever pulls down more...wins, so to speak.

As you know, MPPT means..you just keep drawing more and more current from the panels, until the panel voltage starts to collapse as you draw that current.