The application of IEEE 1588 to industrial automation

One trend in discrete part manufacturing is towards faster and higher precision part production — more parts per minute and better quality. Traditional control solutions can be stretched to their limits. By replacing traditional control solutions with time-based control, faster and higher precision goals can be realised.

The IEEE 1588 standard provides a time-based control solution that can be easily adopted by the industrial control industry to distribute precision time for time-based control on the factory floor.

The case for time-based control

In traditional sequential control systems where input sensors, output actuators and industrial controllers are distributed over a local area network, the control algorithms are typically scan-based and asynchronous, and consequently suffer from significant processing jitter. Some systems employ change-of-state or event-triggered techniques to improve performance. However, time-based control provides the best performance alternative.

Scan-based control For scan-based control, the process is as follows for a simple input, control, output sequence. Input data from sensor devices are sent to the controller at a periodic rate. The controller runs its control algorithm at a periodic rate and output results are sent to the output actuators at a periodic rate. The inputs and outputs change state asynchronously to the periodic input and output scan. Network architecture This input-process-output sequence creates a very elastic or jittery input to output delay. The delay jitter will be a function of when the input changes in relation to the asynchronous periodic scans of the input, controller and output transfer, network transport delays, and internal device delays. These delays are illustrated in Figure 1.

Figure 1: Jitter delays for the input-process-output sequence.

Event-triggered control

Event-triggered or change-of-state control can significantly reduce jitter. With change-of-state operation the input, control and output scan delays are eliminated. When an input transition is detected by the input device, it immediately sends it to the controller. The controller is interrupted when the input arrives and immediately executes its processing algorithm and sends the result to the output device. When the output message arrives at the output device it immediately actuates the output.

This approach will still incur jitter delays due to network transport. If a large number of input transitions occur at once, network congestion and packet loss may occur resulting in additional jitter delays and possibly machine failure. Also, since many I/O devices do not support event trigger mechanisms, this approach is often less viable. In practice, a traditional control system will use a combination of scan-based and event-based control mechanisms.

Time-based control

For many applications, the jitter won’t matter as long as the application response times are satisfied. However, some applications require more precision and have a low tolerance to jitter. For these applications, a time-based control system can solve these problems more effectively.

In a time-based system, an association is made between input and output events and time. Time becomes an integral function of the control system and control algorithms. All devices in the system have the same notion of time. In such a system, the input events are time-stamped and output events are scheduled. The control system precisely knows when the input was sampled and can precisely determine when the output should be actuated. The output device can schedule the output to actuate at a predetermined time.

The only jitter sources for this system are those associated with accurately time-stamping inputs and outputs.

Table 1 shows relative delta jitter delays for the three control mechanisms discussed. The delay numbers indicate the processing delays for the component. The delta jitter is the maximum minus the minimum jitter for the component. Notice how the time-based approach eliminates the jitter sources in the control system.

Table 1: Delta jitter for various control mechanisms.

Jitter or delay source	Delay	Delta jitter (max - min)
		Scan-based	Event-triggered	Time-based
Input	0.2 ms	10 ms	0	0
Input network	1 ms	1 ms	1 ms	0
Controller	10 ms	100 ms	10 ms	0
Output network	1 ms	1 ms	1 ms	0
Output	0.2 ms	10 ms	0	0
Total	12.4 ms	122 ms	12 ms	0

A real-world example

The advantages of time-based control can best be illustrated with a real-world example. In a high-speed conveyor diverter application, individually manufactured parts travel along a conveyor at a constant rate of speed.

A ‘part’ might be a chocolate bar, a nappy or any discretely manufactured product. In this system, the intent is to detect the presence of individual parts as they move down the conveyor, perform on-the-fly analysis of the part to determine if it is a defective part and then trigger actuation downstream to reject the defective part. Network architecture If the resolution of the control system does not match the speed of the conveyor system, then the wrong part or more than one part will be rejected.

Figure 2: High-speed conveyor diverter block diagram.

Traditional scan-based control operation

In this example, an input sensor such as a photoelectric sensor is mounted along the conveyor to detect the presence of a part. The ‘part detected’ status input is sent to the controller as part of the input scan and provides a registration mechanism to track the part as it moves down the conveyor. Knowing the speed of the conveyor, the controller can calculate the location of the part at any given time. An optical inspection system will also be located along the conveyor. The inspection system examines the parts as they move down the system and determines whether the part is good or defective. The controller matches a defective part detected by the inspection system with the part moving down the conveyor and at the appropriate time signals the diverter system to reject the part.

The maximum speed of the conveyor, and consequently the maximum number of parts that can be manufactured per minute, will be determined by the total input-to-output jitter. Using the numbers from the previous section for scan-based control, the maximum speed and parts per minute are calculated as follows:

Part resolution = 122 ms jitter
Maximum speed = 1/122 = ~8 parts per second
Maximum ppm = 8 * 60 = 480 parts per minute

Time-based control operation

Now consider the same system using time-based control. When the photoelectric sensor detects a part, a timestamp is recorded indicating the time that the part was detected. The controller sends the inspection system a time-stamped schedule to signal when the part should be inspected. The controller sends the diverter system a time-stamped schedule to signal when a defective part should be diverted. For this case, the maximum speed of the system is limited by the delays through the system from input to output. The maximum speed and parts per minute are calculated as follows:

Part resolution = 12.4 ms jitter
Maximum speed = 1/12.4 = ~80 parts per second
Maximum ppm = 80 * 60 = 4,800 parts per minute

As previously discussed, the jitter delays are confined to the electrical and mechanical delays for the input detection and output actuation device. Only the transport delays become a factor. The precision of the system is limited to the transport delays and the time-stamping and scheduling accuracies of the respective input and output devices.

Using a time-based control mechanism shows a tenfold improvement in performance. Jitter sources in the system are virtually eliminated. System performance is only limited by the processing speed of input, control and output devices

Application of time to industrial control

Table 2 lists several applications of time to industrial control. For example, time-stamping and scheduled outputs were described in the real-world example of the previous section. Each application of time is briefly discussed in this section, along with examples of how it may be applied to industrial control solutions.

Table 2: Applications of time-based industrial control.

Time usage	Control applications	Industry example
Time of day	General logging, scheduling	Shift management
Time-stamped inputs	Alarm and events, sequence of events, first fault detection	Power generation, pipelines
Scheduled outputs	High-speed sort and divereter (1000 now, 10,000 future parts per minute)	Discrete part manufacturing such as sweets, cigarettes, juice boxes, razor blades etc
Synchronised outputs	Motion, position and velocity control	Coordinated drives, robotics

Time of day

Time is distributed to all devices in the system to perform scheduled activities based on time of day. These activities might include shift start-up, shutdown and reporting operations. For these systems, it is also desirable to automatically distribute time zone and daylight saving time to all factory-floor devices, and to manage the annual transitions between standard and daylight saving time.

Time-stamped inputs

An input is time-stamped to record the time of the input transition on the rising edge, falling edge or both. The input may be a physical device such as a sensor or logical input such as a detected alarm condition. The timestamp is carried with the input data for detection of alarms and events. Applications use time-stamped inputs for time-based control, general logging and trending or statistical analysis. One primary application is to determine from a set of timestamps the sequence in which a set of events has occurred or when the first fault occurred among a sequence of faults.

Scheduled outputs

A time-stamped or scheduled output is used to schedule the time when the output should be applied. Applications use scheduled outputs for time-based control to precisely actuate or assert an output. The output may be physical as in the case of sorting or diverter applications, or logical such as triggering some control action.

Synchronised or coordinated control

Synchronised or coordinated control is used to simultaneously coordinate the actions in one or more devices in a time-coordinated manner. This is very prevalent in distributed motion control where actions for individual motor drives are synchronised to each other. It is also required in robotics to coordinate individual axes of motion.

Conclusion

Time-based control will satisfy the requirements for high-performance applications in industrial automation. The IEEE 1588 standard is ideal and easily adaptable for distributing a precision time-base to all devices of the control system. The challenge will be to migrate control-system devices, particularly I/O, to the IEEE 1588 standard and time-based control solutions. This task will be aided as 1588 becomes more prevalent in the industrial automation arena.

Rockwell Automation
www.rockwellautomation.com.au

References

1. IEEE 1588-2008 Standard for Precision Clock Synchronization Protocol for Networked Measurement and Control Systems.
2. Common Industrial Protocol (CIP) Volume I Edition 3.4 July 2008, ODVA. Available www.odva.org