MCUs vs. MPUs: Choose the right one for your industrial application

MPUs offer more functionality and faster time to market, while MCUs provide a smaller, more cost-efficient solution.


  • Microcontrollers (MCUs) feature on-chip memory and peripherals to minimize footprint and cost while optimizing power efficiency.
  • Microprocessors (MPUs) deliver high performance, high flexibility, and more memory, but place RF, analog, and memory modules off-chip, which increases size and bill of materials (BOM).
  • The line between the two is blurring–MCU vendors increasingly include simple software driver libraries, while MPUs have begun to offer more integrated peripherals.


In the industrial environment, as in the consumer electronics space, embedded designers need to deliver more for less: more performance, functionality, and flexibility for less cost, lower power consumption, and smaller footprints. At the same time, industrial components must meet stringent requirements for reliability, robustness, and lifetime, even in the face of punishing operating conditions—and they need to do it year after year.


The first step to meeting all these conflicting demands is to choose the appropriate central processing unit (CPU). Microprocessors (MPUs) deliver sophisticated, purpose-built performance, but at the cost of greater complexity, larger size, and bigger bills of materials (BOMs). Microcontrollers (MCUs) can’t do it all, but for the types of clearly constrained problems frequently found in industrial applications, the right MCU can deliver significant capabilities for a very reasonable price. Let’s take a closer look at the performance characteristics and the design trade-offs involved in each to help you choose the right component for your product.


Key differences

From a high level, we can classify the differences between an MCU and MPU in following ways:

  • MCUs have internal flash memory and are intended to operate with a minimum amount of external support ICs.  They commonly are a self-contained, system-on-chip (SoC) designs.
  • MPUs rely on external memory and sophisticated power supplies to provide higher processing ability with much greater flash and RAM capacity.


Also known as application processors or media processors, MPUs are normally very high performance compared to MCUs produced at the same time. All that speed and digital functionality comes at the expense of reducing analog functionality and cutting out nonvolatile memory entirely, though. As a result, MPUs typically perform many system-level functions off-chip using external ICs such as NOR flash, dual data rate (DDR) RAM, power management ICs, analog-to-digital converters (ADCs), codecs, and touch-sense controllers.


MPUs generally use open-source operating systems such as Linux and Android, although high-reliability applications may require proprietary operating systems like those from Green Hills Software and Wind River. These operating systems include a library of drivers, for example for codecs, Ethernet, USB, etc. Most MPU providers create and maintain both Linux and Android releases for their evaluation boards, often supporting them with large software teams. Full-fledged operating systems make the development of software much more simple and lend themselves to “plug and play” hardware.


These levels of functionality require significantly more memory. The Linux kernel alone, without application code, can be 1 to 5 MB, for example. Although it can be stored in non-volatile flash, it must be transferred to DDR memory for high-speed execution. This approach results in faster execution of the code, but requires longer boot time, consumes board space, and adds to the BOM. Because of the sophistication of the applications, OSs, and hardware involved, MPUs require file and memory management functionality like structural partitioning, wear leveling, error correction code (ECC), and bad block management (for NAND). All of these capabilities add complexity and cost.


MPU trade-offs

Although the performance of MPUs makes them effective for demanding applications like touch-screen HMIs, they are not necessarily the best solution for every problem, particularly in the industrial sphere. MPUs commonly use cores with extended pipelines, caches, speculative branching, and out-of-order execution. Although these features can greatly improve overall code execution performance, they also introduce some problems, like interrupt latency. Interrupt latency is defined as the time between when an interrupt event occurs and when code to service that interrupt starts to execute. Multi-stage pipelines can require several clock cycles to store the current pipeline information before beginning to service the interrupt, so that the process can be restarted coherently afterward.


For time-sensitive applications like motor control, consistent latency is critical for safe operation. Unfortunately, variability in user configurations prevents vendors from specifying interrupt latency for an MPU; as a result, the devices are not suitable for most motor control applications.


Variability introduces other problems. Although relying on the number of cycles for an instruction to complete is considered bad programming practice, many real-time embedded control systems are based on exactly that¾the assumption that the CPU will execute code in the exact number of cycles every time. This is known as deterministic code execution. When caches and speculative execution are enabled, as is frequently the case for MPUs, the number of cycles for a subroutine can vary. For real time applications like robotics and pick-and-place, this can be a problem.


The external PMICs required to create and monitor the supply voltage for an MPU range from switched-mode power supplies to simple, low dropout (LDO) linear devices. Switched-mode power supplies (SMPSs) have a higher conversion efficiency, but they can add $2 to $3 to the BOM per board. Some MPUs support low-power operating modes, but the leakage current of the high performance transistors and large amounts of memory makes these options less useful than in an MCU.


Many MPUs developed in the past five years have multi-core derivatives. This boosts the processing power several-fold without significantly increasing product costs, since the cores can share external resources. In most cases, the same CPU is used for all the cores, which is referred to as symmetrical multi-core processing. Some new MPUs, which border on being called MCUs, have asymmetrical, multi-core designs that combine an applications processor with a much smaller, less powerful embedded processor.


Today, the line between industrial and consumer electronics has blurred. Manufacturers use PCs to control machines, while meter readers and maintenance technicians use tablets to record readings. Meanwhile, users have come to expect the same level of convenience in their industrial tools as they find in their personal electronics, with high-level functionality and graphics-intensive interfaces. As a result, MPUs increasingly show up in industrial applications like networked industrial controls, relay controls in electric utility substations, and GPS units used to locate survey and construction equipment.


MCU solutions

If there’s one truism in engineering, it’s that there is no perfect solution, just the best solution for the problem at hand. For every high-performance industrial system that demands an MPU, there are many more that need just enough functionality to do one specific job. An MCU delivers the right balance of cost, size, efficiency, and reliability. Vendors leverage modular platforms and value pricing to create multiple part numbers from one die. As a result, designers have a range of options in terms of memory size, pin count, and peripherals to develop the most cost-effective solution.


MCUs are sometimes referred to as SoCs because of the large amount of built-in functionality. They typically integrate reset functionality, low-voltage inhibit, clock sources, interrupts, and on-chip regulators, to name a few. The CPU cores used in MCUs are designed explicitly to deliver the very low interrupt latency and deterministic code execution required for motor control in real-time applications like fans, compressors, washing machines, etc. The Cortex-M4, for example, has a worst-case interrupt latency of 12 cycles and uses a three-stage pipeline.


In addition to peripherals, the process technology used to fabricate MCUs makes it easy to include precision analog and RF blocks, as well as non-volatile memory. Such capabilities allow embedded engineers to easily implement battery-powered wireless systems and sensor monitoring systems with one IC and very few passive components.


Built-in flash memory provides many advantages. First and foremost, it provides high-performance code execution. When implemented with a pseudo-cache buffer, on-chip memory can be read in 64- or 128-bit accesses. This doubles, or even quadruples, the speed at which the system can access data compared to the nominal flash speed. On-chip flash also can enable EEPROM emulation to store and update non-volatile variables such as state of health, product life cycle, serial number, and error codes. Lastly, on-board flash allows very strong security measures that protect chips from piracy.


Putting MCUs to work

MCUs shine in simple, real-time systems for applications that demand speed and economy, with small form factor and clearly constrained functionality. There are three things always in short supply in the average industrial environment: cooling capacity, space, and power. MCUs rarely need heat sinks or fans, whereas MPUs often do. We’ve already discussed the advantages of MCUs in form factor, but they also offer power savings. While a smartphone can be charged on a daily basis, a pressure monitor in the middle of an oil refinery needs to operate on the same battery for two years or more. Almost all MCUs offer a variety of low-power modes that range from WAIT/SLEEP (CPU shuts down, select peripherals continue) to STOP (clocks shut down entirely, only leakage current is evident). More sophisticated MCUs feature additional stop modes to maintain a section of RAM or register space and allow either internal or external wake up sources. This capability is becoming increasingly important because of the inherent high leakage of deep submicron designs.


In the industrial environment, sensors are everywhere, monitoring temperature, torque, pressure, moisture, and more. Keeping track of the environment reduces unplanned downtime and increases device lifetime, both of which save money. Almost every large motor these days incorporates an accelerometer at the base, for example. When the bearings begin to fail, the accelerometer detects increasing vibration, sending an alert so that the bearing can be replaced before failure. An MCU can stand up to tough conditions, monitoring the device for years while consuming minimal power. MCUs not only support conventional sensors, they also improve the performance of the wireless sensors that are beginning to proliferate as part of the growing industrial Internet of Things.


MCUs also address a variety of smart power supply needs for industrial devices. In today’s commercial environment, portable electronics like walkie-talkies, data loggers, and tablets uses wireless battery chargers with complex power management capabilities. Other equipment requires dedicated switched-power supplies. MCUs work well in both types of applications.


Making the right choice

Given all of these trade-offs, how do you choose between MCUs and MPUs for your application? Start by determining how much functionality you need and what your specifications are for voltages, power, size, cost, environmental conditions, etc. Next, consider your options (see table 1).


Table 1: High-level list of advantages and disadvantages


MPU-based system

MCU-based system


  • Higher performance
  • Much greater memory capability
  • Lower cost per bit on RAM
  • Lower cost per bit on flash
  • Very high speed communication interfaces are standard
  • Scalable (able to increase performance and memory)
  • Simple PCB
  • Code security
  • Lower power (both run and STOP)
  • Lowest BOM count and cost
  • Lower interrupt latency
  • Deterministic execution time
  • Smaller form factor
  • Faster boot times


  • Complex board layout
  • Long boot times (Spansion HyperFlash™ can help with this on NOR based systems)
  •  Higher cost/byte because of expensive process
  • Limited CPU speed


The line between MPUs and MCUs has increasingly blurred because of the greater degree of integration seen in MPUs and the growing processing power of MCUs. Still, differences remain (see table 2). The bulk of the MCUs now operate under 150 MHz and have pin counts below 150 pins. The majority of current MPUs operate above 250 MHz and have over 200 pins. MCUs deliver advantages in terms of security, size, power, and cost. MPU-based systems have the advantages of established and free operating systems, higher compute power, scalability, standard interfaces to very high speed communications buses, and lower cost per byte of memory.


Table 2: Compare and contrast table

Peripheral or Feature



DDR memory

Almost always used

Seldom used

NOR flash

Almost always used

Seldom used

Memory management

Standard feature

Rarely used

Operating system

Almost always used

Growing in popularity, but still only used in less than 10 percent of applications


Almost always needed to provide various voltage rails

Seldom needed or used

SATA, Gigabit Ethernet, PCI express and High Speed USB


Seldom used because of the very high performance needed to handle data transfer speeds

Thermal management with heat sink or fan

Common requirement

Seldom needed

PCB metal layers

6 to 10

2 to 6

Bus frequency

100 MHz – 5 GHz

20 MHz – 200 MHz

Pin / ball count

100 to 500

8 to 208


MPUs still dominate for sophisticated applications, but for the types of simple-to- moderately-complex systems that exist everywhere in the industrial sphere, MCUs frequently offer greater long-term value. And they are only improving. MCUs increasingly support operating systems. Taking their cue from the MPU community, more and more MCU vendors are providing peripheral driver libraries that reduce the amount of non-recurring engineering hours required to field the product and allow the code to be more portable. For engineers designing embedded devices to be produced in high volumes, an MCU provides the best balance among performance, cost, and power efficiency.


Spansion®, the Spansion logo and combinations thereof, are trademarks or registered trademarks of Spansion LLC in the United States and other countries. Other names used are for informational purposes only and may be trademarks of their respective owners. ARM and Cortex are registered trademarks of ARM Limited in the EU and other countries.

Get More from Core & Code Subscribe

Leave a Reply

Your email address will not be published. Required fields are marked *

Other stories in this issue


Setting the stage for the fourth industrial revolution

The economies of scale and intelligence introduced by the interconnected industrial Internet promises to help manufacturers save money and streamline operations while satisfying consumer expectations.