Scalability-The Best Approach to Change-The new Midrange Architecture System from Rockwell Automation, designed for ... motion control as it removes the need for a separate motion

Scalability-The Best Approach to Change

Contemporary control systems scale to match the size and complexity of an application, putting OEMs and manufacturers on a pathway to a single control and information architecture

Learn more about:

- The use of scalability to help reduce the costs and complexities of a machine - Performance you might expect from using EtherNet/IP™ in motion control

applications - Deciding between using managed and unmanaged switches - The new Midrange Architecture System from Rockwell Automation, designed for

applications with 16 axes or fewer that require the same high-performance motion control over EtherNet/IP as larger systems

2 | Scalability – The Best Approach to Managing Change

Introduction

The number of product variations continues to explode, requiring machines to handle a wider scope of product sizes, shapes and pack configurations. To stay ahead of changing production requirements, machine and equipment builders (OEMs) have two options. They can design ultra-sophisticated machines capable of more throughput, motion axes and I/O than immediately necessary. Or, to avoid the added expense of just-in-case features, OEMs can take a more contemporary control approach – an approach that effortlessly scales from small to large and from single to multidiscipline applications. A scalable control system provides the price and performance a machine needs to continuously mirror the size and complexity of an application. “With a scalable control platform, we’re able to match the control requirements of each customer,” said Bob White Jr., president, JANDA, manufacturer of resistance welding equipment. Previously, the company’s welding machinery was based on relay logic and cam timers, which were challenging to maintain, not very adaptable for new manufacturing requirements, and required large, unwieldy control panels. “This type of control system was sufficient when customers needed tolerances of one-eighth inch,” said White. “But now, customers are demanding tolerances as small as one-ten thousandth inch.” The Rockwell Automation® Integrated Architecture™ system helps JANDA and others streamline applications with multiple disciplines, such as machine and motion control, and allows users to manage tighter tolerances and other changing requirements.

Scalable Control System for OEMs and End Users

Multidiscipline control systems – managing high-performance integrated motion, process and safety control – have been around for years helping users converge large-scale applications with a common control platform. Yet OEMs have traditionally used multiple control platforms and networking architectures to accommodate the range of applications and machine sizes within their portfolios. This has created unnecessary complexity –

for both the OEM and the end user – because each platform and network has a unique design environment, user interface and vendor support model. Now, Rockwell Automation has extended its Integrated Architecture system with a full portfolio of EtherNet/IP-enabled compact controllers, I/O, servo drives, visualization and simplification tools that are ideally suited for smaller applications. This scalability is especially powerful in the area of motion, as applications with up to 16 axes of motion can be designed using the same programming environment as operations with higher axes counts, such as complex packaging and material-handling operations. With the extension of the Integrated Architecture system, OEMs have a

single development environment and a fully integrated axis portfolio. OEMs are finding that this improves design flexibility and helps quickly scale the control system up, down or across applications to meet a range of needs.

The Pathway to a Single Network Architecture

In addition to converging the control platform for small to large applications, EtherNet/IP is providing a pathway to converging the network architecture – helping OEMs and their end customers streamline information flow. Through EtherNet/IP connectivity and features such as integrated motion, safety and standard control, OEMs can take advantage of a single network architecture for their machines. This network convergence can help bring significant savings, especially for machines requiring high-speed motion control as it removes the need for a separate motion network. The time synchronization capability

“With a scalable control platform, we’re able to match the control requirements of each customer" - Bob White Jr., president, JANDA

3 | Scalability – The Best Approach to Managing Change

within EtherNet/IP provides high-precision motion control, distributing time reference across all nodes so the network does not have to be scheduled. This is ideal as scheduled networks are prone to fluctuating latency and jitter. In addition, network traffic is reduced because the size and content of data packages can be dynamically changed. “Our customers can connect the machine through one single channel, to other machines, to the entire line and even up into the business level,” said Terry Davis, chief executive officer at Production Automation Inc. (PAI) – a machine builder in Montgomery, Ala., that specializes in palletizers. “We didn’t need to consider separate network requirements and specifications when designing the motion application. Plus, for my company, replacing a multi-tier networking strategy with one standard network architecture reduced engineering, commissioning and deployment time, and integration risks.”

Scalability Brings Endless Benefits

The benefits of a scalable control approach are significant. OEMs can better accommodate changing design parameters or expansion of the end customer’s application needs, and dedicate more engineering resources to drive machine innovation. Plus, if OEMs motivate the majority of their customers to standardize on a single control and networking platform, they can streamline support and maintenance efforts. Because engineers would only need training on one platform, OEMs can provide more focused support while saving on overall training costs. With more engineers on staff that are fluent on a given control architecture, the consistency and quality of customer support also improves. Finally, standardizing on a single control platform allows users to stock fewer parts while improving overall part availability. A Germany-based packaging machine builder recently redesigned its system of stretch banding machines used to produce sleeve wrapped packages with the Rockwell Automation Integrated Architecture. The redesign allows engineers to more easily adapt the system to individual customer production requirements and eliminates the need for end users to spend resources on in-house development. It also simplifies frequent product changeovers while maintaining high machine availability for the end user. “What our customers want is modularity and flexibility so they require the best industrial network interface options,” says the machine builder’s project manager for banding machines. “This is why we selected Rockwell Automation as our supplier in this area.” The modularity and faster communication gained by utilizing EtherNet/IP also helps reduce installation costs and the number of switches, connectors and modules in the network. “For example, one add-on allows us to operate each AC drive in the network individually in multidrive mode,” explains the project manager. “We can do this within the controller using one single interface. Because the I/O unit is closer to the process, we can achieve better performance throughout the system.” Choosing a control system that connects and integrates easily to the end user’s manufacturing network makes it significantly easier to incorporate the machine into the end-user environment – leading to faster commissioning and time-to-market. Once the system is installed, the single control platform and networking architecture helps users effectively converge machine-level data with business-level data for improved management and decision-making. This drives smart, safe, sustainable manufacturing.

“What our customers want is

modularity and flexibility so

they require the best industrial

network interface options.

That is why we selected

Rockwell Automation”

4 | Scalability – The Best Approach to Managing Change

Design Considerations for Midrange Architecture Systems

To validate the capabilities of the new Midrange Architecture System from Rockwell Automation, the portfolio of solutions were put through a series of tests. System design considerations are based on these test results. Motion Test Metrics and Methodology During validation testing, various motion-centric metrics were captured and evaluated to verify that motion control would not be impacted by network traffic or switch selection (i.e. managed vs. unmanaged). These metrics included, but are not limited to, the following:

- Position Error (Following Error) - Phase Error (Registration Input Error) - Offset to Master

For more information on these metrics and the validation test procedure, please refer to the Converged Plantwide Ethernet (CPwE) Design and Implementation Guide (ENET-TD001). For network topology, the motion system was exposed to non-motion traffic in an attempt to simulate “real-world” network traffic. The traffic patterns consisted of the following:

- 20 percent Multicast / 80 percent Unicast - Packet 1…1,500 bytes

A maximum of 80 percent traffic bandwidth utilization was injected into the network and motion-related metrics were recorded. The metrics were recorded into a series of histogram data collection arrays and the results evaluated to determine if motion performance was impacted by network traffic. The maximum traffic (80 percent) represents a heavily loaded network that is typically beyond most real-world network loading situations.

Network Topology Overview

Use of Industrial Ethernet gives OEMs and end users more topology options. Benefits and costs of different designs vary based on the application requirements. The Midrange Architecture System was tested in several topologies, including the star, linear and ring topologies. Switch-Level Star Topology A switch-level star configuration is a traditional Ethernet network layout where devices are connected to a centralized network switch, point-to-point. The star configuration is most effective when the devices are located near a centrally located network switch. This configuration may require more cabling for device connections compared to a linear or ring topology, but offers the advantage that if a point-to-point connection is lost to an end device, the rest of the network will remain operational. Device-Level Linear or Ring Topology A device-level linear or ring topology differs from the traditional switch-level star, as the devices are linked together via the two-port embedded switch, instead of being connected back to a centralized network switch. Products that do not support embedded switch technology can still be integrated into a linear or ring topology with the inclusion of an EtherNet/IP network tap (1783-ETAP). The ETAP contains a single device port and two network ports to connect into a linear or ring topology. Both linear and ring topologies simplify interconnect wiring between access layer devices. A ring topology offers the same benefits as a linear topology, but is single-fault tolerant, providing network resiliency with the closure of the ring. With a linear topology, removing a node will essentially disconnect all

5 | Scalability – The Best Approach to Managing Change

downstream nodes or devices, while a ring can withstand a single connection loss. A ring cannot be connected directly through a network switch, but instead the switch can be connected into a ring via an ETAP. This is especially important to remember when using an unmanaged switch, as they do not support loop prevention. It is crucial to configure the ring supervisor before connecting a linear topology into a ring topology. For more information about applying two-port embedded switch enabled products, please refer to the EtherNet/IP Embedded Switch Technology Application Guide (ENET-AP005).

Isolated Network Topologies

The test validates the use of the Midrange Architecture system, based on the Allen-Bradley® CompactLogix™ controller, for executing both advanced motion logic and discrete I/O logic. The following diagrams show the system design using 16-, 8- and 4-axis star topologies and a 4-axis linear/ring topology. Isolated 16-Axis Star Topology

POINT I/O

RPI @ 20ms

PanelView Plus HMI

ArmorBlock I/O SOE

RPI @ 20ms

PowerFlex 40

RPI @ 20ms

Kinetix 350 – 16 axes

@ 8ms CUR

1769-L36ERM Workstation

When tested up to 80 percent network bandwidth utilization, using both managed and unmanaged switches, the system managed 16 position configured axes with an 8 ms coarse update rate.

- Continuous task: 750 rungs (By entering the continuous task size, the user can get an estimate of the execution time after the motion task, system task and period task have all executed).

- Periodic task: 40 rungs/8 ms period (The size of the periodic task and the frequency of execution help determine how quickly the continuous task can be executed).

In addition to the validation testing that was conducted, the same network layout, devices and configuration were entered into the Rockwell Automation Integrated Architecture Builder (IAB) software using the Advanced Performance Calculator (APC) to determine the estimated performance of the entire motion system. The APC allows the user to account for the additional network traffic caused by the CIP Motion™ devices, as well as HMI terminals and I/O devices. This is coupled with algorithms that help determine critical performance criteria for the processor and communication modules.

6 | Scalability – The Best Approach to Managing Change

This figure illustrates the input fields that can be entered and the results that are displayed to the user with the Advanced Performance Calculator.

These utilization factors show how much the Ethernet media, communication controller in the processor, and the CompactLogix controller are loaded. Sizing the system to these values, along with the CIP Motion Cycle Utilization will provide strong estimates for system performance, long before the machine is actually built. These metrics are defined with a three-level rating system, represented by a green, yellow and red set of indicators. This tool should be used early in the design process to help address any concerns related to the overall motion system. Using the Rockwell Automation IAB performance calculator, the color-coded metrics quickly address concerns related to the motion system. The system utilization metrics, for example, can help identify potential bottlenecks or limitations in the overall motion system during the sizing and selection stage of the design process.

7 | Scalability – The Best Approach to Managing Change

Desired results range from Level 0 (green) to mid-Level 1 (yellow) values. Results at Level 2 indicate that the system has exceeded the acceptable limits and modifications to system will be needed to reduce the demand on resources.

Utilization (%) Description Level 0 Level 1 Level 2 1 – Ethernet Media % Utilization of the total Ethernet network or wire <50 50…100 >100

2 – Comm Controller % Utilization of the Ethernet communication controller (embedded)

<50 50…100 >100

3 – Logix Controller % Utilization of the controller or logix CPU <50 50…90 >90

Utilization results for a 16-axis star system topology

Isolated 8-Axis Star Topology

POINT I/O

RPI @ 20ms

PanelView Plus HMI

ArmorBlock I/O SOE

RPI @ 20ms

PowerFlex 40

RPI @ 20ms

Kinetix 350 – 8 axes

@ 4ms CUR

1769-L33ERM Workstation

When tested up to 80 percent network bandwidth utilization, using both managed and unmanaged switches, the system managed 8-position configured axes with a 4 ms coarse update rate.

- Continuous task: 500 rungs

- Periodic task: 40 rungs/8 ms period

8 | Scalability – The Best Approach to Managing Change

Utilization results for a 8-axis star system topology

Isolated 4-Axis Star Topology

POINT I/O

RPI @ 20ms

PanelView Plus HMI

ArmorBlock I/O SOE

RPI @ 20ms

PowerFlex 40

RPI @ 20ms

Kinetix 350 – 4 axes

@ 2ms CUR

1769-L30ERM Workstation

When tested up to 80 percent network bandwidth utilization, using both managed and unmanaged switches, the system manages 4-position configured axes with a 2 ms coarse update rate.

- Continuous task: 250 rungs

- Periodic task: 40 rungs/8 ms period

Utilization results for a 4-axis star system topology

Isolated 4-Axis Linear/Ring Topology

9 | Scalability – The Best Approach to Managing Change

POINT I/O

RPI @ 20ms

PanelView Plus HMI

ArmorBlock I/O SOE

RPI @ 20ms

PowerFlex 40

RPI @ 20ms

Kinetix 350 – 4 axes

@ 2ms CUR

1769-L30ERM

WorkstationOptional Ring Connection

Requires Additional ETAP

When tested up to 80 percent network bandwidth utilization, using both managed and unmanaged switches, the system managed 4-position configured axes with a 2 ms coarse update rate.

- Continuous task: 250 rungs

- Periodic task: 40 rungs/8 ms period

Utilization results for a 4-axis linear/ring topology

Selection of Switches

The basic functionality of a switch is to forward Ethernet frames efficiently. This is accomplished by the use of a Content Addressable Memory (CAM) table. The CAM table is built in the switch by examining the source address of an Ethernet frame. Once the table is populated, frames are forwarded only to the ports needed. This additional functionality differentiates a switch from a hub, which simply forwards all data out all ports. This CAM table is the basic functionality found in all switches, managed or unmanaged. Rockwell Automation provides a family of both managed and unmanaged switches, the Allen-Bradley Stratix families. Managed vs. Unmanaged Switches An unmanaged switch will provide no additional functionality beyond basic CAM table forwarding (switching) of Ethernet frames. They can be an attractive option because of their lower cost and lack of management features to configure. However, they do not provide any additional features that may be necessary to meet the performance requirements of a machine or application.

10 | Scalability – The Best Approach to Managing Change

A managed switch, on the other hand, helps provide the means to properly segment the network – an important role as IP traffic is increasing significantly as manufacturers add new applications, such as motion and safety, to the network. The intelligence services delivered by managed switches allows more sensitive and complex applications to run securely on EtherNet/IP, while providing enterprises with easy access to manufacturing metrics for business decisions. These services include additional Layer 2 features typically found in plantwide networks: diagnostics, segmentation (VLANs), security, data prioritization or quality of service (QoS), resiliency, and Layer 2 loop prevention. Additional Layer 3 functionality such as inter VLAN routing is included in the Allen-Bradley Stratix 8300™ Layer 3 managed switch. Careful considerations need to be taken when deciding on designing around an unmanaged or managed switch. Considerations may include:

1. Will this small to mid-sized motion system be integrated within a greater plantwide system? 2. Are ISA99 Industrial Automation and Control System security standards applicable to this

system?

Managed switches should be used if the answer to either question is yes. Test results reveal that it is possible to build an isolated Midrange Motion System with an unmanaged switch. However, there will be times when it may be more beneficial to utilize a managed switch.

Selection table shows when to use a managed switch

In short, managed industrial Ethernet switches provide:

- Layer 2 loop prevention - Segmentation and prioritization to reduce latency and jitter for a converged control and information

network - Security features to help protect manufacturing assets from internal and external security threats - Diagnostics to reduce Mean Time to Repair and increase Overall Equipment Effectiveness - Resiliency options to maintain availability and integrity of control and information data

11 | Scalability – The Best Approach to Managing Change

To reduce network latency and jitter, Rockwell Automation and Cisco Systems CPwE architectures recommend segmenting and prioritizing network traffic. Segmentation reduces the impact of broadcast and multicast traffic by creating smaller Layer 2 broadcast and fault domains. Segmentation also establishes smaller domains of trust, which simplifies management of security policies. Loop Prevention As small Ethernet systems expand and become larger systems or as small systems get connected to larger systems, it is crucial to plan the system accordingly. If this is not done, a loop situation could occur. Loops can cause multiple sections of a machine to fail, by causing a broadcast storm. In short, loops cripple the communication of the machine, and potentially entire networks. A loop occurs when two connections are made to the same line or series of switches – creating a physical and logical Layer 2 loop. It could easily happen as the machine is being expanded, during commissioning, or even during maintenance of devices. This loop could also occur on one switch during maintenance. Software in some managed switches disables the port causing the loop, thus helping prevent possible lost production time. This loop prevention protocol also offers a level of resiliency at the switch level. For more information on resiliency protocols (and loop prevention protocols), please refer to the CPwE Design and Implementation Guide (ENET-TD001). Quality of Service Quality of Service (QoS) refers to a management protocol found only in managed switches which prioritizes network traffic. QoS helps to ensure control traffic is prioritized over other types of traffic to achieve deterministic data flow, along with low levels of latency and jitter. QoS in Rockwell Automation switches is a Layer 2 and Layer 3 service.

Network requirements of different types of traffic

12 | Scalability – The Best Approach to Managing Change

Prioritize traffic using cell/area zones

The IEEE 1588 Precision Time Protocol (PTP) Events and CIP Motion packets have the highest priority in QoS. Because PTP is a protocol which enables motion on CIP Motion drives, it is always some of the most crucial traffic in an Ethernet system. QoS found in managed switches helps to ensure the PTP packets are delivered with limited jitter and predictable latency. This is especially important in large motion systems. In smaller motion systems, where Rockwell Software® RSLogix™ 5000 Version 20 or higher firmware for Allen-Bradley ControlLogix® controllers is implemented, it is possible to implement a motion system without utilizing QoS. The testing done utilizing the Allen-Bradley Stratix 2000™ unmanaged switches would be an example of this type of installation. It is important to consider that as more (and different types of) traffic is introduced onto the network, the traffic will not be prioritized with an unmanaged switch. Good industrial network design, including managed switches, enables the convergence of many standard traffic types on a common network infrastructure. If mixing different types of traffic, such as voice and video, on the same network as motion, an industrial switch that prioritizes industrial traffic over commercial traffic. Commercial switches designed for the enterprise may prioritize voice or video over the industrial traffic. Security Proper segmentation and security practices must be considered when designing an Ethernet network. Because ODVA has defined EtherNet/IP on open standards, unwanted and unnecessary traffic or users could travel onto motion networks. Most security recommendations will require enabling Layer 2 and Layer 3 services. Only managed switches will be able to tap these services. Rockwell Automation recommends using a “Defense-in-Depth” method of security that starts at the physical layer with topology choices and moves through Layer 2 and 3 security services and segmentation.

13 | Scalability – The Best Approach to Managing Change

Mechanisms for improving network security

Network Loading

Ethernet media and devices have a limited amount of bandwidth. As this bandwidth is utilized, especially in systems involving switches, latency and jitter of the network can be greatly affected. In a midrange motion system, this is critical to consider, as these network attributes directly affect the protocol used in motion planning. Assuming a full 100 Mbit/s, full duplex connection, a network will be considered loaded to 20 percent when 20 Mbit/s in both directions are being utilized. This is the same for 40 percent, 80 percent, etc. This traffic, for the purpose of testing, can be generated using artificial means. This is done to simulate the network utilization of a traditional control system. For our internal testing we chose a split of 20 percent of traffic generated as multicast and 80 percent unicast. The packet sizes were randomized up to 1500 bytes. The total utilization was varied from a small “typical” system of 20 percent overhead, to an unusually heavy load of 80 percent utilization. Figure 8 shows the breakdown of traffic in the largest tested system. To translate the overhead traffic into real-world network utilization values, 80 Mbit/s is calculated to what it means to an industrial control system:

400,000 HMI Numeric Tags at a rate of 250ms, or

15,000 HMI String Tags at a rate of 250ms, or

1,600 I/O Racks at a RPI of 20ms A typical midrange motion system of up to 16 drives would rarely, if ever, have this level of overhead. The system was tested beyond the normal operating realm to ensure accurate results even in the most extreme conditions.

14 | Scalability – The Best Approach to Managing Change

Ethernet Cabling Recommendations

Selecting the appropriate Ethernet Media is critical in applying a successful Ethernet network. The following criteria should be used to select Ethernet media in order to meet CE requirements (if necessary) for your installation. Cable Lengths The length of each single span of Ethernet cable for Allen-Bradley Kinetix® 350 drives must not exceed 100m (328ft). This includes connecting drive-to-drive, drive-to-controller and drive-to-switch. Cable Types The recommended cable type is Shielded Twisted Pair (STP) cable and should be routed in a clean zone of your panel installation. Please refer to the Kinetix 350 user manual (2097-UM002) for more information on recommended panel layout and design. Rockwell Automation recommends customers use the Bul. 1585 Ethernet cable; a high voltage 600V version is available for use in cable tray with high-voltage power cables and also to address the UL 508A requirements. The cable environment must also be considered when selecting the appropriate cable. Please refer to the following white paper (ENET-WP007) for more information. For best practices to minimize the possibility of noise-related failures, please refer to the System Design for Control of Electrical Noise reference manual (GMC-RM001).

Additional resources for designing small to mid-sized motion applications:

All publications can be found online www.rockwellautomtion.com/literature

Kinetix 350 single-axis EtherNet/IP servo drive 2097-UM002

CompactLogix 5370 programmable automation controller 1769-UM021

CIP Motion Configuration and Start-up User Manual MOTION-UM003

CIP Motion Reference Manual MOTION-RM003

Kinetix Motion Control Selection Guide GMC-SG001

System Design for Control of Electrical Noise Reference Manual

GMC-RM001

CIP Motion Popular Configuration Drawing Without Notes IASIMP-QR026

CPwE Design and Implementation Guide ENET-TD001

Ethernet Design Considerations Reference Manual ENET-RM002

Guidance for Selecting Cables for EtherNet/IP Networks ENET-WP007

EtherNet/IP Embedded Switch Technology Application Guide

ENET-AP005

Allen-Bradley, CompactLogix, ControlLogix, Integrated Architecture, Kinetix, Rockwell Automation, Rockwell Software, RSLogix, Stratix 2000 and Stratix 8300 are trademarks of Rockwell Automation, Inc. CIP Motion and EtherNet/IP are trademarks of ODVA.

Publication IA-WP002A-EN-P – August 2012 Copyright ©2012 Rockwell Automation, Inc. All Rights Reserved. Printed in 2012