“We don’t want people to think of the DeviceNet of Things the same way they think of DeviceNet,” said John Caspers, principal software engineer at Rockwell Automation. “We want to make this so easy that you wire up a panel and go online and hit an easy button and everything works.”
Caspers spoke about the proposed DeviceNet of Things specification enhancements at the 2017 ODVA Industry Conference in Palm Harbor, Florida. The increased popularity of EtherNet/IP and the IoT movement has established a basis for innovation of new smart things in the automation worlds, explained Caspers. “However, the costs associated with deploying EtherNet/IP prevent it from being successfully deployed on many low-end things, such as push buttons, relays and contactors,” he cautioned. “Enhancing DeviceNet as an in-panel cabling and communication solution for low-end devices will lower the cost of deployment below the levels provided by EtherNet/IP; simplify physical or mechanical connection methods, leading to lower panel-build costs; and enhance ease-of-use over existing networking solutions, including DeviceNet. If we can attack the cost of wiring, there’s greater savings there than lowering the cost of components.”
DeviceNet and EtherNet/IP share the common industrial protocol (CIP), which provides a unified architecture for network technologies supported by ODVA. “We did customer listening around the world and got key requirements for DeviceNet of Things,” explained Caspers. “It must be functional, simple to use and economical.”
It has to save panel builders time. Our intent is to eliminate all of that wiring. It must deliver both network power to power device electronics and control power to facilitate the actuation of contactors and relays, and it must support non-safety and safety devices on the same wire.
It must use a single easy-to-use media connector, must include a simple network commissioning methodology, or none at all, and must eliminate the need for media trunk and drop distance calculations.
To satisfy the economical requirement, it must use low-cost media, allow for a reduction in price and size of typical DeviceNet products and the use of commercial off-the-shelf (COTS) power supplies, resulting in a lower total cost of ownership (TCO) than hardwired solutions.
“Think of it as DeviceNet with an easy button,” said Caspers. A new seven-wire ribbon cable will be specified, and the network will be restricted to a linear topology. Because the maximum length of the network will be 50 m, suppliers will be able to sell 50-m spools of ribbon cable, and, when the spool is used up, a new network will have to be started.
“We’ve added some conductors to the ribbon cable,” explained Caspers. “This will allow selective service delivery which isolates the network power and switched power to minimize the noise from it on the line.”
The linear nodal geography of the network will allow for commissioning the network, facilitated by a new link object and new commissioning object introduced into the DeviceNet specification. These new objects will control some system behaviors by using the select line to throttle system level service delivery sequencing for the network.
“The DeviceNet of Things commissioning object initiates select line messaging; captures, stores and reports an actual nodal geography; accepts a reference or desired nodal geography; and facilitates automation node commissioning based on a reference geography,” explained Caspers. “The commissioning object may be implemented in a CIP router, a DeviceNet of Things master scanner or a commissioning tool. In all cases, the DeviceNet of Things commissioning object resides in the first or leftmost node on the network whose media access control (MAC) ID is fixed at Address 0.”
A nodal geography is a complete ordered set of device keys for all devices on a network, explained Caspers. “In a subnet system, the CIP router device will include the ability to discover the entire subnet nodal geography and deliver it to a commissioning tool or processor on the other side of the gateway for display,” he said. “Select line messages are received on the select in line. The device services the message, and, after processing is complete, the message is retransmitted on the select out line. The result is sequential message delivery from left to right.”
One way to easily support the select line message transmission and reception is to use a 1-millisecond timer interrupt. A node’s interrupt service routine (ISR) can be up to 1 ms out of phase with its neighbor’s. Each bit is transmitted three times in the ISR, which means each bit has 3 ms on the wire. When a node senses a high level on the select in pin, denoting a packet is being transmitted by a neighbor, it waits for the next interrupt and then stores 16 bits every third interrupt, guaranteeing the value read will be in the middle third of each bit’s transmission time. After all 16 bits have been sampled and stored, the validation field is verified.
“We’ve prototyped the 16-bit, select-line protocol, delivered by implementing an ISR, in the labs, and it works pretty well,” explained Caspers.
“In a subnet system, the CIP router device will also be able to receive an entire reference subnet nodal geography from a commissioning tool or processor,” he continued. “While auto-addressing sounds good in theory, a lot of OEMs we talked to want to do a base design on a machine and then order an option, but, when you add it to the line in the middle, it bumps the addresses of everything behind it, so it breaks your ladder. We’re working on something better than traditional sequential addressing to fix that.”
Also, for power supply architecture, many people have complained they didn’t like the power supply requirements for DeviceNet, so three requirements will drive the DeviceNet of Things, said Caspers. “Users can now pick a COTS power supply,” he explained. “Both network and switched power must be delivered. And it must be possible to size a single power supply for a system and use power taps to distribute power for the network. Our power taps are going to be smart.” The four power tap styles include first tap, switched-power tap, network-power tap and switched-power-and-network-power tap.
“Since this is going to be such an inexpensive network, we think we can do this on an $0.85 microprocessor,” predicted Caspers. “Improper termination of a DeviceNet network is a common problem in the field. Since the cost of simple DeviceNet of Things nodes will be small, it makes sense to provide smart terminators so tools can report.”
DeviceNet is a mature network. To extend the functionality of the various devices on it, a smart-drop device would allow traditional devices to participate in DeviceNet of Things select line and proxy identity information, acting as a DeviceNet router.