Part 2 of the RZ/T1 microcontroller discussion, elaborates on how the performance of previous microcontrollers (MCU) and microprocessors (MPUs) was insufficient for the Internet of Things (IoT) and machine-to-machine communication (M2M) era. Included in this discussion are examples of how implementing servo motor control with high-performance MCUs and MPUs removes the need for an FPGA or ASIC and improves the speed and precision of industrial robots and machine tools in factories.
Autonomous M2M Through Unorthodox Servo Motor Control
In recent years, there has been an increasing demand for high-performance microcontrollers and microprocessors for a wide variety of products. The factory automation field is no exception to this trend. In this age of IoT and M2M, there is increasing demand for high-performance MCUs and MPUs to improve the speed and precision of industrial robots and machine tools, improve the performance of real-time networks in factories, and manage all factory systems. Part 2 presents a discussion with Katsuhiko Neki of Ubiquitous Computing Salon (UC Salon) and engineers from Renesas Electronics about what is different about implementing servo motor control with high-performance MCUs and MPUs without FPGAs or ASICs, with real examples.
Participants in This Discussion
Don't Use Cache in Motor Control
Tell me about the current trend for motor control.
Kojima: Motors are now used in a wide variety of applications, so there is a demand for more precise and faster motor control. For example, although the most popular refrigerant for compressors of freezers and other equipment is a chlorofluorocarbon alternative, another idea is the liquefaction and circulation of carbon dioxide. This idea requires high-speed, high-precision motor control. Thus, a microcontroller with better performance than ever before is needed. Or even in the case of an ordinary motor, if a high-performance microcontroller were used to increase the rotating speed, the amount of iron and copper used in the motor could be reduced. This is an example of better value from better performance. Furthermore, industrial robots require constant maximum servo performance in order to improve production efficiency. So this is another area where a high-speed, high-performance microcontroller is required.
Ogawa: I started being asked about three years ago whether we had a 500MHz microcontroller for servo motor control. However, with the current way of doing things everyone had their hands full with advancing to the next finer process every two or three years (because a finer process means faster speed). In terms of over 500MHz devices, there were general-purpose processors, but to be frank using a general-purpose processor for servo motor control is very problematic. This is because a general-purpose processor uses cache memory to increase average performance. However, run time varies greatly depending on the cache memory hit rate. Therefore, if you wanted to perform servo motor control, using cache memory was not an option.
Kojima: The CPU core of the RZ/T1 is an Arm® Cortex®-R4 processor with FPU. This core employs a tightly coupled memory (TCM) structure where the CPU core and memory are directly connected. This eliminates performance variation caused by cache hit rate.
Ogawa: I created a demo of an inverted pendulum where we can see the effectiveness of TCM. An inverted pendulum requires quite complex real-time control. It doesn't work well with a low-speed microcontroller, and not even with a high-speed general-purpose processor due to cache miss/hit. In this demo, the RZ/T1 is used to control a servo motor. You can turn TCM on and off to see the effect of TCM. Compared to when TCM is off, when TCM is on the time required to process the necessary calculations for control doesn't vary. Thus, the movement of the inverted pendulum can be controlled extremely smoothly.
Technologies for Multi-Protocol Support Lead to Freedom from FPGA
From Part 1, we know that the number of engineers who can handle FPGA is very small. Why is this?
Neki: FPGA designers write structural descriptions. Thus, knowledge of electrical engineering, timing design, and so on are required. Furthermore, structural descriptions are very tough for a third party to decipher. Thus, maintenance is also tough unless everything is meticulously documented. A common story I've heard from customers is that they had to redesign everything from scratch when they lost their FPGA designer.
Kojima: Since an FPGA is commonly used in the encoder part of servo motor control, we've been asked by many customers for a solution to this problem. If you want to know the position of a servo motor very precisely, a complex encoder protocol such as EnDat, BiSS®, or A-Format is required. In some cases, the customer does not manufacture the encoder themselves, and is purchasing it from an external source. In other cases, the customer's robotics manufacturer will specify the servo motor with the encoder that the customer needs to use. In such cases, the customer cannot predict which encoder protocol will be used. Therefore, they must use an FPGA or ASIC instead of building the protocol-processing hardware into the microcontroller. Renesas' RZ/T1 has made it possible to realize this encoder protocol processing without an FPGA.
Kojima: Unlike microcontrollers which require an FPGA for encoder and industrial Ethernet processing when used for servo motor control, with the RZ/T1 an FPGA is unnecessary. First of all, in terms of the encoder, the RZ/T1 has a built-in multi-protocol encoder interface that can support multiple protocols by changing a configuration file. Using this, it is possible to support complex protocols such as EnDat, BiSS®, and A-Format without an FPGA.
Ogawa: In terms of industrial Ethernet, the RZ/T1 has a built-in R-IN Engine that has multi-protocol support for industrial Ethernet. There are several industrial Ethernet protocols. Usually, protocol support is achieved using an external FPGA or ASSP for communications. The R-IN Engine, the built-in industrial Ethernet engine of the RZ/T1, allows multiple industrial Ethernet protocols such as EtherCAT®, EtherNet/IP(TM), and PROFINET to be supported without any additional LSI.
A multi-protocol encoder interface is built in. This makes it possible to support complex protocols such as EnDat, BiSS®, and A-Format without an FPGA.
Manuals for each protocol can be downloaded here:
RZ/T1 Supports Autonomous M2M
Ogawa: Recently, there has been an increasing demand for smaller AC servo amplifiers. FPGA consumes a lot of power and generates a lot of heat, so thermal design for a high-performance microcontroller and an FPGA on a small board is difficult. Due to this, it is anticipated that it will become more and more difficult to use FPGA even if you want to use it. We are considering a future RZ/T1 product with two built-in encoder channels. This would enable customers to control two axes of a multiaxial servo amplifier (which have recently entered the market) using a single chip, which would greatly help reduce costs.
Kojima: Up until now, if you wanted to support IoT/M2M with a microcontroller, the handling of communications was an excessive load. Therefore, you had to use an external LSI. The high-performance processor and communications-specialized R-IN engine of the RZ/T1 enables you to make use of the heretofore unused terabytes of sensor data. This means that it has become possible to work effectively on improving the overall efficiency of a factory and reducing defects. Renesas' RZ/T1 device provides strong support for this "autonomous M2M".
Ogawa: Integrated development environments for the RZ/T1 are the ARM® DS-5, the IAR Systems Embedded Workbench® for ARM®, and the Renesas e2 studio. Debugger and in-circuit emulator (ICE) tools are available from Kyoto Microcomputer, Yokogawa Digital Computer, Bitran, and Computex. In addition to the evaluation board made by Renesas, Alpha Project also offers a board. Sample programs for the Renesas evaluation board can be downloaded for free from the Renesas website. We have also been talking to three other companies, and we're thinking about an evaluation kit that includes a motor in the future. This rich and comprehensive development environment is one of the advantages of the RZ/T1.
Neki: I think of the RZ/T1 as the "savior" of the Japanese manufacturing industry. In the age of IoT/M2M, more new demands will continue to appear. As a new type of device that is not quite a microcontroller or a microprocessor, I fully expect the RZ/T1 to stimulate a fresh surge of activity in Japan's electronics industry.
Development of "RZ" Stands for Commitment
The RZ/T1 was developed by leading development teams working thoroughly and closely together to challenge the 500MHz operating frequency limit. The goal they aimed for was 600MHz. The device that was developed breaks through to 600MHz operation while having a structure capable of real-time control that does not rely on cache memory, dramatically differentiating itself from existing products. With such a different class of performance, it becomes possible to realize functions by software that until now could only be realized by hardware. We could call this a disruptive innovation. The RZ name stands for Renesas' commitment to continuously enrich this device lineup. It signifies Renesas' desire to provide solid support for these devices so that users can feel secure in using them for ten or twenty years going forward.