DC-Link Input Exploration Harry Johnson Motivation: General Specs: - - PowerPoint PPT Presentation

dc link input exploration
SMART_READER_LITE
LIVE PREVIEW

DC-Link Input Exploration Harry Johnson Motivation: General Specs: - - PowerPoint PPT Presentation

DC-Link Input Exploration Harry Johnson Motivation: General Specs: SiC low-side FET SiC high-side diode 1Mhz Switching Frequency Output cap appropriately sized. Average input current control loop. Low-Side FET: Cree


slide-1
SLIDE 1

DC-Link Input Exploration

Harry Johnson

slide-2
SLIDE 2

Motivation:

slide-3
SLIDE 3

General Specs:

  • SiC low-side FET
  • SiC high-side diode
  • 1Mhz Switching Frequency
  • Output cap appropriately sized.
  • Average input current control loop.
slide-4
SLIDE 4

Low-Side FET: Cree C2M0280120D

  • 280mR RDSon
slide-5
SLIDE 5

High-Side Diode: Cree C4D05120A

  • Output current is actually rather low
  • Optimize Qrr
slide-6
SLIDE 6

MATLAB:

  • Simple and quick to run sanity check
  • Only models cycle-by-cycle, not control

inside cycle

  • Results used earlier
slide-7
SLIDE 7

LTSpice Overview:

  • Models circuit behavior within cycle
  • Runs for a single 120Hz cycle
  • Exported setpoints from MATLAB to CSV, imported as

PWL files for current sources.

  • Rest of Littlebox (main buck, DC-link output buck)

modeled as current sources where necessary.

  • SiC devices: manufacturer-supplied models
slide-8
SLIDE 8

LTSpice Schematic:

Main Buck Emulator DC-Link Input Current Setpoint DC-Link Capacitor Discharge Current

Package Inductance

Total Current setpoint: 5A w/ soft start 450V + 10R

slide-9
SLIDE 9

Per-120Hz cycle plots.

slide-10
SLIDE 10

Turn-On

slide-11
SLIDE 11

Turn-Off

slide-12
SLIDE 12

Calculated Results:

  • Averaged over the entire charge cycle
  • Power In: 1.3195KW
  • Power Out (into cap): 1.2888 KW
  • Switching Power Loss: 30.7 Watts
  • Eff: 97.6% (only including switching loss)
slide-13
SLIDE 13

Areas for Future Work

  • Model capacitor ESR appropriately for

mega-cap array (not modelled at all for purposes of this simulation)

  • Inductor losses: DCR relatively easy to

model but I suspect primary losses will be in core.

  • Actually build the darn thing!
slide-14
SLIDE 14

Bonus: Megaderp moment

Late-night calculations:

  • Automatic power calculation in LTspice (alt-click) on

Cree NMOS model.

  • 500W !?
  • Worked for a long time trying to optimize package, gate

drive, etc.

  • Turns out I’m being autotrolled by the thermal

modelling, which is represented by voltage and current sources.