Post Go back to editing

High voltage buck converter design

Category: Hardware
Product Number: LT3748

I'm looking to design a high input voltage buck converter stepping a 1000V input down  to 24V nominal,  delivering about 2.5A (around 50W, potentially lower).  Although I've design a number of converters in the past, they've all been for embedded applications where the voltage levels are all "manageable", so I"m at a loss as to how to deal with the 1000V:

1) I would need to step this input voltage down prior to connecting it to any of the various  PWM/power supply controller  ICs i've found  (like the LT3748, for example).  How is this typically done? I don't expect a simple divider to work, since I wouild expect the ripple  on the Vin pin  to be pretty significant if I did that.  I figured I'd step it down to power the IC, but feed the flyback xformer directly from the 1000V source, but I need to understand how I'd lower this voltage to a range compatble with these ICs' Vin pins.

2) How do I manage the high (Vin+Vfyback) voltage at the drain of the mosfet?  I guess I'm limited to SiC mosfets? or are there other tricks to deal with this?

3) Sizing the primary inductance (of the flyback transformer) seems to also be  challlenging  --requiring a fairly large magnetizing inductance (>833uH) which, although I haven't looked into it yet, i'd expect  it to be of significant physical size.  What can be done to reduce the size of the magnetics?

Perhaps the flyback topology is wrong for this solution.  Should I be considering  something like a full or half bridge converter instead?  Perhaps a stage like that to reduce the Vin to power the IC ? 

Or better yet, perhaps,  implement a two-stage solution where I use a full bridge to convert the DC to AC, then step that down via a transformer to something more manageable that allows me reduce the step down ratio?  hmmm, as I write this, it occurs to me that if the 24V can be unregulated, I could just DC/AC convert  the 1000V,  step it down to 24VAC and just rectify it then....

I'm still reading the literature to better understand what I'm up against and how I can approach this problem, but If anyone can offer some pointers on how to approach this design, recommend a device, and point out things I should be aware of,  that would be great.  If Analog Devices has reference designs or can help with the design, I'd like to hear about that  too.

Thanks,

Jorge

Hello Jorge,

For this type of input, I think you would need something isolated. The use of transformer greatly reduces the bulk of the inductor needed and the stresses for the seconday side MOSFETS. Like a two-fet forward converter. Basically you would need a PWM controller on the secondary side, and then send the PWM signal to the input side via an isolated gate drive to control the MOSFET.

Regards,

Jhun