How to optimise peripheral design for microcontrollers
The design of tomorrow’s embedded systems presents complex challenges given the aggressive goals of improvement in areas of performance, cost, power, size, new features and efficiency.
There is, however, an emerging design option to address these complex problems – analogue components (amplifiers, ADCs, DACs, voltage references, temperature sensors, wireless transceivers etc.) and digital peripherals smartly integrated with ARM microcontroller cores.
The difference between this and traditional analogue integration is the high level of analogue performance now being offered, and the optimisations made to solve specific system level problems.
Logically, combining parts could solve many of these embedded system goals, but simply putting several discreet components and a processor in one package is not the answer; the solution is far more complex, requiring smart integration.
In order to create the optimum microcontroller solution, a knowledge of the overall system along with the availability of the right intellectual property (IP), and expertise in that intellectual property, is required.
Increasingly, the acquisition/creation and implementation of the IP itself needs to be facilitated by the semiconductor manufacturer.
This IP then needs to be modified to meet the needs of the primary target application and work well and easily with the other IP blocks.
Finally, there needs to be business level collaboration, combining the expertise and knowledge of the system manufacturer and semiconductor manufacturer.
Solar inverter example
New technologies promise advancement for solar PV inverters, including: neutral point clamped (NPC) topology; three-level, five-level and multi-level topologies; and high frequency switching using silicon carbide (SiC) and gallium nitrite (GaN) transistors.
Figure 1 shows a two stage solar inverter. The first stage is a dc-dc conversion that raises the voltage level so it is compatible with the peak voltage on the grid. The second stage is dc-ac conversion. The area in red shows the low voltage circuit control components that can be integrated into a microcontroller.
For supporting multiple stages of conversion and high-speed control loops, a processor core with the right performance needs to be selected. In this case a 200MHz ARM Cortex-M4 will meet the need.
The filters, shown in figure 1, are used in combination with isolation ADCs. This allows for measurement of ac current on the grid and dc current injection in order to avoid saturation of transformers.
The traditional method is to use a Hall effect current transducer but this is expensive compared to the isolation ADCs. This assumes that the filters are integrated into the microcontroller avoiding an additional programmable logic chip.
The isolation ADC-SINC3 filter combination also has improved linearity than Hall effect sensors, leading to a reduction in harmonic distortion.
As the grid becomes smarter, solar inverters will need to have more intelligence to help deal with grid imbalance when more power is available from multiple sources than is needed.
For this reason, there is a focus on photovoltaic system intelligence with an eye towards grid integration, where each contributor to the grid must co-operate to stabilise the grid, rather than simply supplying power open-loop. A harmonic analysis engine designed specifically to monitor the quality of power injected into the grid helps address this need.
Motor control example
If a motor is used to convert electricity into movement, higher efficiencies can be achieved by: selecting the correct motor type, special motor design, accurate control circuitry, and an adjustable speed drive (ASD).
Permanent magnet motors are of increasing being adopted – efficiencies can be as high as 93%, which exceeds the Europe’s premium efficiency standard (IE3).
Smartly integrated analogue microcontrollers offer benefits for ASDs, and control circuitry improvements. Figure 2 shows a block diagram of a motor control system.
Control circuits can be improved by using an accurate ADC with fast conversion.
In pick-and-place machines, for example, fast control loops can increase throughput and efficiency, resulting in lower production costs.
Variables need to be measured at the rate the end machine is moving. An ADC with greater than 12bit accuracy improves the precision with which the phase currents can be controlled, but sample conversion latency cannot be sacrificed for accuracy – eliminating ADCs that average or over-samples to improve noise.
Fast conversion times complemented with a fast processor core allows a control loop to run faster resulting in better response and settling times.
Good base IP starting points were available for PWMs, trigger routing unit (TRU), multiplexing, buffering, successive approximation ADCs, and memory control (DMA) for a microcontroller targeted at motor control.
However specific design modification of these block was necessary in order to achieve the level of coordination required for timing ADC sampling within a PWM period – hence the analog-to-digital converter controller (ADCC) block for co-ordination, which is designed to fully use the speed of the two ADCs engines which convert in 380ns.
Advanced base technology is just the starting point – chip designers must not only have broad knowledge of customers’ systems, but also have expertise in the design, application, and optimisation of the precision analogue and digital components, and the associated IP and software components.
In addition, silicon manufacturers must be willing and able to directly interact and collaborate with system manufacturers to create new products.
The most appropriate components are selected, modifications are made to optimise for the end application, and IP blocks are modified to work well together.
It is only then that the optimised pieces can be integrated together.
Writers are Colin Duggan, marketing director and Denis Labrecque, business development manager with Analog Devices.
Tags: analog, ARM, embedded