DISTRIBUTED CONVERTER CONTROL

ADVANCED POWER ELECTRONIC CONTROL

Distributed control technology for power converters

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus. Nunc magna tellus, ultrices id bibendum et, aliquam at sapien. Proin a felis dignissim diam maximus cursus.

What is distributed converter control?

In power electronics, lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus. Nunc magna tellus, ultrices id bibendum et, aliquam at sapien. Proin a felis dignissim diam maximus cursus.

One technology for Multiple objectives

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus. Nunc magna tellus, ultrices id bibendum et, aliquam at sapien. Proin a
CENTRALIZED CONTROL

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverrapien. Proin a felis diam maximus cursus.

Examples:

  • Low-complexity systems
  • Laboratory converter prototypes
  • Interleaved systems (not necessarily)
COORDINATED CONTROL

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverrapien. Proin a felis diam maximus cursus.

Examples:

  • Back-to-back applications
  • Microgrid inverters
  • Distributed motor drives (e.g. traction)
HIERARCHIZED CONTROL

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverrapien. Proin a felis diam maximus cursus.

Examples:

  • Modular Multilevel Converters
  • High-performance motion control
  • Paralleled inverters (to some extent)

Networked control solutions

Flexible control architectures

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus. Nunc magna tellus, ultrices id bibendum et, aliquam at sapien. Proin a felis dignissic viverra lectus. Cras ac lacinia nunc, et tempus lectus. Nunc magna tellus, ultrices id bibendum et, aliquam at sapien. Proin a felis dignissim diam maximus cursus.

Dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus. Nunc magna tellus, ultrices id bibendum et, aliquam at sapien. Proin a felis dignissim diam maximus cursus.

I/O extension mode
MASTER-SLAVE

Vertical data exchanges
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverrapien. Proin a felis diam maximus cursus.

Max. num. of masters1
Max. num. of slaves64
CPU-to-CPU
communication
N/A
CPU-to-FPGA
communication
Native (RealSync)
Slaves addressing mode
SynchronizationNative (RealSync)
Exchange directionVertical only
Parallel mode
Multi-masters

Horizontal data exchanges
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverrapien. Proin a felis diam maximus cursus.

Max. num. of mastersUnlimited
Max. num. of slavesN/A
CPU-to-CPU
communication
Can/Ethernet
CPU-to-FPGA
communication
N/A
SynchronizationNot available
Exchange directionHorizontal only
Combined mode
MASTER-SLAVE

Vertical and horizontal data exchanges
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverrapien. Proin a felis diam maximus cursus.

Max. num. of masters16
Max. num. of slaves16×64=1024
CPU-to-CPU
communication
SFP (RealSync)
Also ETH/CAN (slower)
CPU-to-FPGA
communication
Native (RealSync)
Slaves addressing mode
SynchronizationNative (all units)
Exchange directionVertical + Horizontal

I/O extension mode

Master + slave operation

Easy-to-use centralized control architecture

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus. Nunc magna tellus, ultrices id bibendum et, aliquam at sapien. Proin a felis dignissim diam maximus cursus.

  • Only the master has its CPU operating
  • All FPGAs are tied to the same CPU (inside the master)
  • Communication and synchronization are native (RealSync over SFP+)
CONTROL SOFTWARE
  • Only one controller exists.
  • Only one control executable is needed (e.g. Simulink file or C/C++ project).
  • CPU processing performance often sufficient, even for complex algorithms.
  • Multi-rate control already possible in this mode thanks to different interrupts.
COMMUNICATION
  • Data exchanges between master and slaves uses imperix’s RealSync protocol.
  • No configuration is needed. All data are automatically sent from/to the master.
  • Data exchanges are full-duplex and entirely deterministic.
  • Latency is extremely low (largely sub-us).
SYNCHRONIZATION
  • Slaves operate in the same clock domain as the master.
  • No configuration is needed. All slaves are automatically synchronized to the master.
  • Synchronization accuracy is ±2ns.

Parallel modE

Multi-masters operation

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus:

  • Each master has its CPU operating
  • Each FPGA is tied to its neighboring CPU
  • Only poor synchronization is available (IEEE1588)
  • Data exchanges use from/goto functions or blocks

The basic approach for coordinated control

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus. Nunc magna tellus, ultrices id bibendum et, aliquam at sapien. Proin a felis dignissim diam maximus cursus.

CONTROL SOFTWARE
  • Distinct controllers are implemented.
  • Each controller requires its own control executable (Simulink or C/C++ project).
  • Distributed control can be implemented, even though controllers are independent.
  • The overall computation capability is proportional to the number of units.
COMMUNICATION
  • Data exchanges between units are made using Ethernet or CAN.
  • Communication is user-configured using from/goto function blocks.
  • Data transmission delays depend on traffic.
  • Latency is significant (can be hundreds of us).
SYNCHRONIZATION
  • Devices operate in different clock domains.
  • RealSync is not available.
  • Synchronization can be achieved using IEEE1588, but with limited accuracy (ms-scale).

COMBINED MODE

Multi-masters operation with slaves

Conventional distributed converter control

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus:

  • Each master has its CPU operating
  • Each FPGA is tied to its neighboring CPU
  • Only poor synchronization is available (IEEE1588)
  • Data exchanges use from/goto functions or blocks

The basic approach for coordinated control

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus. Nunc magna tellus, ultrices id bibendum et, aliquam at sapien. Proin a fsectetur adipiscing elit. Mauris nec viverra lectus:

  • Each master has its CPU operating
  • FPGAs are tied to their respective masters
  • All units are natively and perfectly synchronized
  • All data exchanges use SFP+ (galvanically-isolating fibers)
CONTROL SOFTWARE
  • Distinct controllers are implemented.
  • Each master+slaves system requires acontrol executable (Simulink or C/C++).
  • Control can be distributed or hierarchized among complex controllers.
  • The overall computation capability is proportional to the number of masters.
COMMUNICATION
  • Data exchanges between masters and their slaves is made transparently (I/O extension).
  • Exchanges between masters is user-configured with from/goto function blocks.
  • SFP+ offers superior performance over CAN/Ethernet, thanks to strict determinism.
  • Latency can be very low with SFP+.
SYNCHRONIZATION
  • All devices operate in the same clock domain.
  • No configuration is needed. All slaves are automatically synchronized to the master.
  • Synchronization accuracy is ±2ns.

Combined mode

Multi-masters operation with slaves

DISTRIBUTED CONTROL OF MMC

Distributed inverter controll

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus:

  • Each master has its CPU operating
  • FPGAs are tied to their respective masters
  • All units are natively and perfectly synchronized
  • All data exchanges use SFP+ (galvanically-isolating fibers)

DISTRIBUTED DRIVES

Distributed inverter controll

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris nec viverra lectus. Cras ac lacinia nunc, et tempus lectus:

  • Each master has its CPU operating
  • FPGAs are tied to their respective masters
  • All units are natively and perfectly synchronized
  • All data exchanges use SFP+ (galvanically-isolating fibers)