Experiment on a 3.5KW 100KHz Buck converter

[aneesmohammed]

Hello guys..
I am currently trying to build a 3.5KW basic buck converter as an experiment. I calculated the inductors, and capacitors , then simulated in MATLAB ...I am not planning any closed loop control for the time being(may be later)...So i need to chop the input voltage into half(dutycycle=50%).....The losses due to hard switching or absence of isolation is not a problem in this experiment. Can you suggest me a power electronic switch and gate driver for it to fullfill this experiment.(Does anyone having experience with siC mosfet and any associated gate driver )

Ratings:

Vin =650V
Vout=325V
power rating 3.5KW

(I have attached the closed loop buck converter which i simulated after open loop buck converter...Simulation is okay....But need advice for hardware)

 

[cupoftea]

650 to 325 at 3kw ayk, would usually be an LLC or a phase shift full bridge, to reduce the switching losses.
But yes, a buck at low frequency would theoretically be possible.
Or maybe a few in parallel to ease the component choice.

For a buck , you wouLD want a gate drive transformer drive for the FET.
You woudl also need a current sense transformer to see the high side current.

Basically, i woudl start the design with the inductor choice.

Go on maginc site and find a torroid that you can wind......if you cant find one big enough, then just downsize the power, and use parallelled converters till you find an inductor torroid that can be wound for your inductor and power level.

Paralleling is easy if you just add a max currnt clamp to the output of each buck. (there are other ways to pllel too)

Yes i would think you will neeed a SiC diode.
Also a sic fet woudl be helpful for the fet.

You can put sic diodes in pllel.

(Are you sure you dont want to just use a vicor module?)
remmember to balance series caps if they each dont stand the full volts.

As you know, you will need to DC fuse it
And be v careful, as 650VDC woudl kill i believe....HVDC is far more dangerous than HVAC.

 

[Easy peasy]

This is not really a build project for anyone who does not have a healthy respect for higher DC voltages ( > 200VDC )

and some good experience with power electronics at this power and voltage

switching at 30kHz say at 650VDC input generates a lot of RF noise that can get into control chips and upset things

so a moderate turn on of the switching devices is a good idea - as is appropriate snubbers to limit dv/dt on the sw node

At 3kW the currents are around 10A, so the source has to provide 10A pulses and the bucking choke has to handle 10A + ripple

it also has to handle 650V from end to end at start up - so the windings must be properly insulated from each other and the core

+/- 2A of ripple at 30kHz = 1.34mH approx, energy stored in the choke = 97mJ, on a 470uF cap this will raise the Vout by 0.63V at sudden turn off

so 470uF is quite acceptable - also a good quality electro can handle the 1.4 A rms ripple current.

Good luck - you'll need lots of it.

 

[D.A.(Tony)Stewart]

The Q of your switched LC will significantly overshoot 50% without a soft start.

 

[Easy peasy]

Start up at 50% D 650 Vin:

--- Updated Jul 12, 2024 ---

Circuit:

--- Updated Jul 12, 2024 ---

At full load attached - it still gets to 530VDC pk - so soft start a pretty good idea !

 

[aneesmohammed]

Thank you for your responses...Let me reveal the actual area where this is to be used.

I am currently working as a part-time hardware intern at an underwater Remotely Operated Vehicle(ROV) manufacturer in India...So for their next project, the requirement is to extend the power cable(Tether cable) up to 2 KM range (This vehicle is not battery operated instead it transmits power for propulsion through underwater cables). So my task is to find some power architecture by searching for SMPS and DC-DC Converters of some standard manufacturers.
My design criteria are as follows:
1) Total load at ROV is about 3KW
2) So to reduce the conductor size the transmission voltage should be above 500
3) Thrusters are operated at 24V DC
4) Power supplies Inside the ROV should be compact sized

One of the identified power architectures is attached here. Later I thought that, if replacing a bulky DC-DC Converter from ABSOPULSE(marked in the figure) with a basic buck topology the size of power inside the ROV further reduced.....That's why I arrived at a 3KW buck converter ( Input side and output side use isolated converters from standard manufacturers )

From the above responses, I accept that a basic buck converter of such a higher rating is a poor option. Then if you guys have any alternatives you are welcome.

 

[Easy peasy]

We have done this project for many ROV systems - we put from 1kV to 4.2kV down cables to ROV ( typ 1km ) and have LLC converters ( 80kHz ) inside the ROV typically 16kW to convert the HV down to 350VC at 40+ amps - we have very RF quiet LLC to reduce electrical noise.

We also design the 3 phase mains 400V L-L to 4.2kV converters

400 - 800 VDC to 325VDC at 3kW would be very easy for us.

N.B. just so you are aware - at start up and no load the full 800V would be seen at the cable end.

 

[BradtheRad]

Your scheme can work although everything rests on whether your 2 km cable can go unnoticed by sharks. Sharks are the creatures with sensory organs in their snouts that can detect galvanic (mild electrical) activity in boat hulls. If it thinks it's a living thing they might take a chomp out of the boat. Underwater communications cables have several layers of substances that discourage shark bites. Buoyant balloons every 50 feet counteract weight of the cable. All the better if they secrete shark repellant.

 

[aneesmohammed]

Your scheme can work although everything rests on whether your 2 km cable can go unnoticed by sharks. Sharks are the creatures with sensory organs in their snouts that can detect galvanic (mild electrical) activity in boat hulls. If it thinks it's a living thing they might take a chomp out of the boat. Underwater communications cables have several layers of substances that discourage shark bites. Buoyant balloons every 50 feet counteract weight of the cable. All the better if they secrete shark repellant.

That's interesting !! But I need not bother about shark bites for the time being.....Instead the Power Supply

 

[Easy peasy]

also - if your buck switch fails for any reason - the down stream electronics go boom!

 

[aneesmohammed]

I have found something new to me called a daughter card from TEXAS INSTRUMENTS !! (https://www.ti.com/tool/PMP23249#technicaldocuments)

Do anyone have knowledge about this.....From the reading, I hope that this is a GaN-Mosfet along with its gate drivers and protection circuits.....Does it means I can use this daughter card as a whole power electronic switch unit for my requirement

 

[cupoftea]

That PMP23429 offers the bootstrap circuit, but that wouldn't be adviseable with 650V+
So use the SN6505 hi side supply instead.
Its just a GAN fet with its driver, placed in the hi side......you could do that yourself.
That module they show..it doesn't have a heatsink, so can you mount a heatsink to it?

--- Updated Jul 12, 2024 ---

As EasyPeasy suggests, the best one is an LLC for your requirement.

 

[D.A.(Tony)Stewart]

Not bad but does not meet your specs

Google confirms their internet was attacked by sharks 2015

 

[Easy peasy]

The cost of failure of the buck switch is very high indeed - when you consider it carefully

this is the major reason we use our modified LLC to get 350VDc @ 45+ amps from the HVDC input

if any part of the power electronics fails - we just get less than 350VDC - not 800V ! ( or up to 4.2kV in our case )

 

[aneesmohammed]

The cost of failure of the buck switch is very high indeed - when you consider it carefully

this is the major reason we use our modified LLC to get 350VDc @ 45+ amps from the HVDC input

if any part of the power electronics fails - we just get less than 350VDC - not 800V ! ( or up to 4.2kV in our case )

Do you have any reference materials for that modified LLC converter

 

[Easy peasy]

Hi - sorry - no - it is all in house - we have been building these since 1999 - so quite refined and successful.