TI C2000™ (or simply C2000 going forward in this post) is a family of real-time microcontrollers from Texas Instruments. I (and probably many of you) first heard of C2000 MCUs during my undergraduate study 12 years ago – specifically the DSP course. In fact, even today, C2000 MCU-based trainer kits and power electronics lab setups are commonplace in a lot of reputed universities in India as well as outside.
This microcontroller family has been around for 20+ years now and from the looks of it (so many new parts!), TI has no plans to stop investing any time soon in this popular MCU platform.
“Accelerating” real-time innovation!
Over the last two years, almost every carmaker of repute has at least two powertrains on offer for the same variant to cater to multiple driver personalities. The average commuter who needs a car for daily errands would prefer a good old-fashioned petrol engine whereas the thrill-seeker prefers the smaller but more powerful turbocharged engine. The difference in torque and power is not the fuel but how the fuel is used by the engine.
Same is the case with microcontrollers executing a real-time control loop.
For a control loop which executes at an interval of microseconds, the time available to process the feedback (often analog inputs), execute the control law (often a PI or PID controller), pre-process outputs and update the actuator (often PWM signals) is a precious commodity. The choice of the right microcontroller for such use-cases is governed by how fast the different aspects of the control loop are completed with sufficient margins while still allowing the CPU to do other tasks.
The C2000 MCU family features a number of real-time control-specific accelerators like the Trigonometric Math Unit (or simply TMU), Floating Point Unit 32 or 64 (or simply FPU32 or FPU64), Control Law Accelerator (or simply CLA), Configurable Logic Block (or simply CLB) and Fast Integer Division (or simply the FINTDIV).
There is already great content from TI to help understand these features and the value they bring – here are some resources.
- An Intro to C2000 MCU Accelerators
- An Intro to CLB
- An Intro to CLA
- A Real-time Benchmarking of C2000 MCU Accelerators
- The awesome C28x academy
Traction Inverters in EVs – the need for speed
India is currently a hotbed for Electric Vehicles (or simply EVs) – and this applies to 2-wheelers (or simply 2W-EV), 3-wheelers (or simply 3W-EV) and 4-wheelers (or simply 4W-EV). You have to be living under a rock to not notice the upswell in the EV ecosystem in India.
By far, the most critical subsystem in an EV is the traction inverter (like the engine in the ICE vehicle) – the thing that converts the battery’s DC output to pulsating waveforms driving the PMSM motor (or motors) thus generating anything from a few kW to upwards of 100 kW of power! At the heart of these systems is a real-time microcontroller executing the control loops at break-neck speeds – silently transporting you.
Here’s a good high-level view of traction inverter.
But then one would ask – if this is simply what all traction inverters do (or are supposed to do), what would differentiate one manufacturer from the other? Let’s see a few aspects and how they influence the choice of the central microcontroller.
Experience
Since the control loop implementation is owned by the manufacturer, one could tune the control response to be aggressive, normal, sluggish or even dynamic depending on the state of the battery. This heavily influences the experience and the perception of the motorist.
The ability of the microcontroller to implement an entirely configurable v/s a rigid implementation of a control loop i.e. feedback processing, dynamic control law implementation, offloaded pre-processing and flexible actuation methods are the differentiator.
Auxiliary Features
From what can it do to what else can it do? There are attempts to use the traction inverter to also handle other functionalities like simultaneously generating the DC power rail that can be used by other loads in the vehicle. Similarly, brilliant minds in the industry are racing to design a system that could drive the motor when on and charge the battery when off.
The ability of the microcontroller to crunch multiple control loops (eg: traction + DC-DC) thrown at it without compromising time margins and determinism is the differentiator.
Safety
What is trust? Sitting inside a metal container zooming on the expressway at triple-digit speeds, just above flammable batteries whose capacities are of the order of tens of kWh, trust is the feeling that we will be A-OK at the end of the trip.
In a traction inverter, you could have a short circuit in the motor windings or the battery input, a blown FET (or many) in the inverter section, a damaged position feedback sensor which sends the control algorithm into a toss, a faulty capacitor (or many) which suddenly reduces the instantaneous power going into the motor and so on!
The ability of the microcontroller to quickly enter a fail-safe state even while blazing through the control loops and being available always is the differentiator.
Communication Links
Manufacturers are increasingly packing in more code into traction inverters. The control algorithm is still the same – the enhancements are in the kinds of communication mechanisms that these systems support. But why do we need communication in a pure power-conversion system? To bring all the above aspects to fruition.
The ability of the microcontroller to execute one or more communication stacks (eg: UDS, OBD-II, some proprietary, firmware upgrade agents, etc.) while silently executing the control loops deterministically is the differentiator.
C2000 MCUs for EVs – an ideal match!
The most popular control algorithm for traction inverter systems today is the Field Oriented Control (or simply FOC) of PMSM motors. Here’s a simple diagram showing the key aspects of the FOC algorithm used to drive a PMSM motor.
Note that the position feedback could be through quadrature encoder, resolver, hall sensor array or even thru motor phase currents (referred to commonly as a sensorless implementation). However, sensorless implementation is not a common approach adopted by EV makers.
In most FOC implementations, the control law implemented is a PI controller which operates on normalized feedback inputs. The outputs are also normalized quantities which are then transformed in order to generate PWM duty cycle values. There are ample resources on field-oriented control on the internet – go fish! Here’s one to get you started – https://www.ti.com/lit/an/sprabz0a/sprabz0a.pdf
Why do the C2000 MCUs stand out against competition?
The TMU inside the C2000 MCU significantly accelerates the implementation of the Clarke, Park and Inverse Park transforms which is the crux of the FOC algorithm. Similarly, the FPU drastically reduces the time taken to compute the control law and the SVM outputs. A similar computation on a general purpose-core like a Cortex-M7 or Cortex-M4 would take at least 2.5x more time – thus wasting precious CPU cycles which could have been used to implement a different system functionality or a safety diagnostic.
Conclusion
The C2000 MCUs have always been the platform of choice for digital power applications – especially critical systems like server power supplies, industrial-scale UPS systems, solar grid installations, etc. The sheer integration of real-time functionalities and accelerators in these devices made them the first choice of architects across product generations.
With the growing adoption of EVs and the manufacturers’ need to differentiate, it was only a matter of time before C2000 MCUs became the platform of choice – especially with the integration of relevant accelerators like the TMU, FPU and the FINTDIV – which drastically increase the headroom available for software developers to pack in more functionality as well as to switch faster and step better – leading to higher power efficiencies and longer times on the road.
When are you switching to C2000 MCUs for your traction design?