To mux or not to mux (a HART story)

In the beginning

Back in the 1980s, the ubiquitous 4-20 mA current loop became a de facto standard for instrumentation employed in process control. 

4-20 mA current loop

It had quite a few advantages over other kinds of signals, like the possibility of powering the field devices through the signal wires (loop-powered devices), the option to have a “live zero” functionality, meaning that a 0-signal value was equivalent to 4 mA.

The “live zero” feature allowed the possibility to have some advantages, like wire breakage detection diagnostic, allowing the device to remain powered even when there was no process signal output, the relative immunity to electrical noise and voltage drop, and the option to convert the current signal to a voltage signal using an adequate resistor.

Here comes the digital era.

When digital technology became available, the idea of improving the 4-20 mA signal became the goal for every major instrumentation supplier. Using newly available cheap microprocessors, manufacturers developed various techniques to include additional information to be encoded digitally into the 4-20 mA current signal loop. As far as I can remember, the three major contenders that attempted to do this were Foxboro, with its FoxCom protocol, Yokogawa with its Brain protocol and Rosemount with the HART (Highway Addressable Remote Transducer) protocol.

FSK encoding employed by the HART protocol.

All these three technologies were variants on the basic same idea: to include digital information into the 4-20 mA current signal loop by modulating a variable frequency shift key into the current signal. This method of encoding had been developed by Bell Laboratories in 1976 and was known as the Bell 202 communication standard. The Bell 202 standard encoding method was to modulate 2 frequencies in the current loop, a 1200 Hz signal meant “1” and a 2200 Hz signal meant “0”.

Digital does not always mean fast.

Since HART uses a sine wave the average value of the signal is zero, so the 4-20 mA signal is not affected. That meant that if your control system did not understand HART, and they did not offer that feature until much later, your HART capable instrument worked like a traditional 4-20 mA device, therefore ensuring backwards compatibility.

The data transmission speed of the HART protocol was 1,2 kbps in half-duplex mode, like what computer modems were capable to do in those days. If you think that 1,2 kbps is slow, you are correct, by modern standards. Industrial computers were also slow in those years so more speed implied more processing power and higher costs.

Yokogawa and Foxboro developed their own proprietary versions of the FSK encoding method. Yokogawa’s Brain protocol doubled the data transfer rate to 2400 kbps and Foxboro’s Foxcom protocol reached the blistering speed of 4800 kbps. They achieved that performance by increasing the modulation frequencies.

The Open Standards revolution.

But Rosemount did something disruptive. Instead of trying to fight a price/feature war with its faster rivals, the company declared HART an open standard and created an organization called the HART Communication Foundation to create and maintain that standard. Any instruments supplier who wanted to add digital communications capabilities to its 4-20 mA devices just had to join the HART Communication Foundation and then he would receive all the necessary information to develop a HART capable version of its devices. The HART Communication Foundation performed all the necessary interoperability tests. If successful, the device got the right to show the HART Foundation logo.

The HART Communication Foundation logo.

Today this practice has become commonplace, and it is recognized as the simplest and most efficient way to gain market share and spread the acceptance of new technologies.

But in 1986 this was unthinkable. And although both Foxboro and Yokogawa claimed that their versions were better (they were initially 2 to 4 times faster), the advantages of adding digital communications capabilities to a field device by simply using an open standard were obvious: no need to invest in R&D, no need to design new hardware and easy access to HART enabling kits from third party suppliers. Implementing HART became so much easier that nowadays, there are just a few geeks that remember what FoxCom and Brain meant.

How to HART simply? There is no simple answer.

HART is a Master/Slave protocol, the field device is the slave, and the master can be either a handheld device, a computer with a HART modem or a HART interface module or I/O card. The initial method to communicate digitally with a HART device was and still is, the use of a HART modem and a host computer, since although handheld configurators were also available from an early stage, they were quite expensive. The original goal was that the control system itself would be eventually able to communicate with the HART devices using I/O analogue cards with an embedded HART modem.

Typical commissioning setup for a HART device.

As time went by, more and more control system and Remote I/O system suppliers started to include HART capable I/O analogue cards. But the system integration of HART functionality into the control system’s software was not common until the early 2000s. 

The main obstacle against the widespread use of HART as a useful tool and not only as a configuration method was that it was complicated and cumbersome. Even control systems that offered HART capable analogue I/Os did not integrate HART functionality into their control strategies. The usual way to employ HART, other than with a handheld modem was using Asset Management Software, being Fisher Rosemount’s AMS the first one available. They did not really try to give this software an imaginative name. AMS software run in a computer, therefore an interface module that could split the 4-20 mA signal from the digitally embedded HART data was required. The 4-20 mA signal was sent to the control system and the HART data to the AMS equipped computer.

HART interface module, also known as loop converter.

Since most control systems were not HART capable yet, the use of diverse HART interface modules became commonplace. These devices offered a point-to-point connection between a computer running an AMS software and a HART device while maintaining the 4-20 mA current loop with the control system. This practice was expensive if you wanted to enable HART functionality for many instruments, mainly due to the additional cabling and hardware necessary, but the HART protocol had a hidden ace.

You do not need to HART alone...

Since the HART signal used the same physical layer as the RS-485 serial protocol, it also had support for addressable nodes. Every HART device comes from the factory with its address set to 0, and that makes it work as a 4-20 mA device with a superimposed sine wave, FSK encoded, digital signal. But if you change the address from 0 to a range from 1 to 16, the device switches to what it is called the multidrop mode and sets a constant current signal of 4 mA in the loop transmitting all the available data in digital format, including the main variable.

This mode made it possible for manufacturers to develop multivariable field devices, which send the main variable as a 4-20mA signal and the other variables as HART encoded data.

The same device, in multidrop mode, sends all the available data encoded in HART. Currently available multivariable HART devices can send up to four different variables, being possible to configure which one will go as a 4-20 mA.

HART interface module with support for multidrop mode.

If you have a multidrop compatible HART interface, then you can make a daisy chain of up to 16 HART devices, thus radically diminishing the number of required HART interface modules. The problem with this kind of setup is that a wire breakage at any point interrupts all communication activity.

…As long as you HART slow

Sounds good in theory, until you remember that HART has a data transfer speed of 1,2 kbps. Then the idea of running 16 devices at that data transfer speed makes chess look like an exciting action-packed sport.

And what could you do if you had a lot of HART devices? If 16 devices per HART interface looked too fragile and complicated, you could use a HART multiplexer. Hart multiplexers usually supported 16 or 32 HART devices, but each multiplexer also worked as an RS-485 serial addressable device. 

HART multiplexer with RS-485 interface.

That meant that you could connect in a daisy chain up to 16 HART multiplexers and then, using either an RS-485 adapter card in your AMS computer or, much later, using an RS-485 to Ethernet converter, you could handle hundreds of field devices with a single AMS equipped workstation. If you had the patience of a Buddhist priest, of course.

Also, the computers available in those days and the Operating systems were usually not reliable enough nor had the required processing power to handle such arrangements with confidence. And you had to keep your HART devices connected to the analogue I/O cards to transmit the 4-20 mA signals.

Everybody liked HART, but few were willing to HART.

The complexity and cost of implementing those solutions did not encourage many end-users to spend the time and face the trouble of using all that extra hardware just for online field device configuration. When AMS software packages started to incorporate advanced online diagnostics and multivariable HART field devices became available, the industry’s interest in HART technology increased somewhat. But even today, although HART is the most commonly available digital communication protocol in the world, it is also actually the less employed. One of the main issues of using HART as a fieldbus was it’s slow data transfer speed.

With the introduction of fully digital communication protocols for field devices, such as FOUNDATION Fieldbus and PROFIBUS PA, the concept of digital multivariable field devices with online diagnostics became popular, but its use remained in a few specialized niche markets such as soybean oil plants, biodiesel production facilities, chemical plants and the food and beverage industry. 

If HART seems too slow, maybe you would like to Fieldbus…

But in any case, field device Fieldbuses worked at an awe-inspiring speed of 31,25 kbps or 26 times faster than HART devices. They were simpler to install and cheaper to use in the long term. But they were overly sensitive to the quality of the cabling installation and grounding. And maintenance people never actually liked fieldbuses. 

The idea of having to use a laptop computer or a handheld communicator in the field was not something they liked. Anything more than a multimeter looked like overengineering for them. Fieldbus diagnostics required specialized equipment and well-trained personnel and usually implied hiring an external, and expensive, service supplier. 

Fieldbuses appeared in an era where digital natives, a.k.a. millennials, were still not in the active working force. So, the long-promised digital revolution did not actually come to the Process Control Industry with fieldbus technology, even if the marketing machinery of control system suppliers was screaming that it was already going on.

One main hurdle remained in the road to the digital plant: the unavoidable presence of these concoctions commonly referred to as gateways.

The gateways from hell

Foundation Fieldbus to Ethernet gateway.

Gateways are basically protocol translators and media converters with a big fat memory buffer in between. Since there was never a coordinated fieldbus standardization process in the industry because every market had its own niche application requirements that were fulfilled by a specific bus, and the upper levels of the automation hierarchy moved closer and closer to Ethernet, all that large variety of fieldbuses (most of them based in serial protocols), required at some point either a media conversion or a protocol conversion to be connected to the Ethernet network.

The massive introduction and acceptance of Industrial Ethernet protocols that started in the late 2000s put additional pressure on fieldbus protocols, and their respective support organizations, to adhere to a unified standard.

Here comes the catchy Ethernet.

The PROFINET protocol, developed by PI, was the first Industrial Ethernet protocol that included in its specifications the concept of a standard proxy to communicate other networks and fieldbuses with the main Industrial Ethernet network. PROFINET aspires to become the backbone of the Industrial Ethernet revolution. It provides proxies for almost any existing communication protocol. And that includes HART, of course.

PROFINET standard proxy concept.

Although since the early 2000s, most Remote I/O solutions offered the possibility of tunnelling HART signals through the communications protocol they were using to communicate with control systems, thus enabling HART over PROFIBUS, HART over Modbus, Hart over Ethernet IP, etc. functionality, they still required separate Asset Management Software to take advantage of the additional information provided by HART technology.

For a different kind of HART, just add some IP.

But two things happened in the early 2010s that changed the scenario:

1. First, due to its simplicity and lower cost, Wireless HART field devices started to spread across many plants, usually employing a Modbus TCP enabled access point. Wireless HART devices employed the 7th Revision of the HART standard, using a mesh type network to provide an industrial-grade wireless network protocol. The rival ISA100 standard, in a similar way as FOUNDATION Fieldbus and PROFIBUS PA, tried to offer a broader, more ambitious solution, but never really caught the industry interest.
2. Second, in 2012 the HART protocol was enhanced to allow the use of the TCP/IP stack to enable access to all the available HART data located in the field devices, without the need of any translation process or any loss of information. This technology is known as HART-IP.

HART-IP allows simple integration of any HART device.

This method also avoids the error-prone data mapping process required in Modbus based solutions. It also eliminates the need for a HART modem since all required communication tasks are done through Ethernet. The 4-20 mA current loop becomes unnecessary in some cases, although compatibility with traditional HART devices is maintained via HART-IP multiplexers. Finally, you can get rid of the ancient 1,2 kbps data transfer speed, because since the physical media is Ethernet, your HART-IP devices can communicate at 100 Mbps.

And since the HART-IP devices are connected through Ethernet, you can use whatever Ethernet-capable device you may want to use to interact with them, a tablet, a smartphone, a laptop, etc. There is no need to pay thousands of $ for a bulky handheld HART Communicator.

It is easy to HART-IP.

The availability of cheap powerful multithreading processors enables the development of devices such as the Phoenix Contact Ethernet HART multiplexer. This device can be connected to a PROFINET network, where it works as a modular IO Device, or to an Ethernet network using Modbus TCP, a legacy option that requires data mapping, or to an Ethernet network using HART-IP. 

Phoenix Contact modular Ethernet HART multiplexer.

It can be integrated into any control system using a GSDML generator for PROFINET integration, a Modbus driver for Modbus TCP applications, an EDD file description for integration with AMS solutions, a DTM file for integration in any FDT framework and finally can communicate directly with any OPC-UA client. And this can be done for modular configurations that allow the connection of up to 40 HART field devices. Due to its powerful processors, it can scan multiple HART devices at the same time, avoiding the long scan times required by traditional RS-485 multiplexers.

And it will get better, soon…

The incoming advances in Ethernet technology such as TSN (Time-Sensitive Networking), which will provide deterministic network behaviour using standard Ethernet interfaces and the availability of the APL (Advanced Physical Layer), which will enable to have field devices with Ethernet connectivity, using two wires cable and providing intrinsic safety and power and communications through the same cable will only reinforce the importance of HART-IP technology in the long term.

APL: one physical layer, compatible with Ethernet, for any kind of field device.

It is somewhat ironic that the first attempt at introducing digital communications into the process industry might be also the one that will finally enable the adoption of Ethernet as a universal communications solution.

If this is the case, HART technology will be present long after the other, more ambitious fieldbuses fall into oblivion. 

Mirko Torrez Contreras is a Process Automation consultant and trainer that started to HART a long time ago.. The opinions exposed in this article are strictly personal. No affiliation exists between the author and the companies mentioned. All the information required for and employed in this article is of public knowledge.