How do you bring new technology to an already extant install base?

BY MICHAEL ANDERSON
Director of
Technology, The PTR Group
www.theptrgroup.com

As the IoT continues to expand, we see a large interest in the use of
distributed sensor nodes in the industrial space. However, many industrial
applications fall into what is commonly referred to as “brownfield”
applications, which require new hardware/software solutions to co-exist with existing solutions. In addition, many of these in situ solutions may be more than 20 to 30 years old. So
how do you bring new technology to an already extant install base?

First, we need to
understand the nature of the current problem set being addressed. In the
industrial space, we are frequently talking about the online monitoring of
existing physical processes. These could range from monitoring the temperature
of the content in a crude oil pipeline to monitoring the amount of chlorine in
water as it leaves a water treatment facility. In the industrial space, they
tend to focus on simplicity of installation and reliability of the sensor
technology. In some cases, the state of the art may not have progressed very
far; that is, the good old way of doing it may be good enough. In other
applications, there may be a significant advance in the technologies that could
result in a lower cost of ownership and better reliability.

Next, we need to
understand the connectivity involved with the current sensor system. This
ranges from the type of physical connectivity (e.g., RS-232/422, RS-485, 20-mA
current loop, Ethernet, etc.) to the protocols. Some of these protocols could
be predicated on the use of synchronous serial communications à la HDLC, Bisync,
DDCMP, X.25, and others. Or they may be using other industrial standards such
as Profibus, Modbus, EtherCAT, DeviceNet, or even custom protocols.

Also, the nature
of many industrial applications has a wide variety of harsh environments that
they have to survive in, including extended temperature ranges, shock and
vibration, high humidity, unreliable power, and intermittent connectivity. And a
characteristic of the typical industrial operator is to run the equipment for
as long as they can, even though there may be a “better” solution, to reduce
costs by using already amortized equipment until it dies. Further complications
that are often encountered in industrial infrastructure is that the solution
may have been designed 30 years ago and all of the designers have moved on or
retired. This leaves the current operators with a large number of unknowns when
they start contemplating a phased rollout of a new sensor system.

Fortunately, the
sensor marketplace now offers a broad selection of alternative approaches. Many
aging analog components have now been replaced with microelectromechanical
systems (MEMS) that are much more robust, smaller, and more power-efficient
than their original counterparts. Often, these sensors are already designed to
take advantage of new communications media such as LoRaWAN, NB-IoT cellular,
Sigfox, or even Wi-Fi.

As much as the
industrial operator may want to continue to use their existing cable plant, that
may not be practical. Old industrial communications interfaces may not be
practical for the updated sensor systems. Even if they are available, they may
be cost-prohibitive. A set of trade studies will need to be undertaken to
determine the most cost-effective communications approach. If the decision is
to go with a wireless solution, then the designer will need to investigate the
practicality of the solution given the physical characteristics of the system
to be upgraded.

Before beginning
any significant update of an existing system, the designers should conduct a
simple prototype test that reproduces as many of the operational constraints as
possible. This can be accomplished using one of the many development kits that
are available on the market. These will often include a
commercial-off-the-shelf (COTS) development board, an operating system such as
Linux, and a COTS radio solution.

If we plan to
replace hardline wiring with a wireless solution, there will need to be a site
survey in the radio frequency (RF) spectrum to make sure that the frequency
bands will work for the application. Oftentimes, industrial applications will
require large motors or other sources of RF interference. Alternatively, just
the presence of large metal structures, such as those found on an oil platform,
will be enough to limit the effectiveness of a wireless solution. This sort of
survey should be performed before the radio selection so that the printed-circuit-board
(PCB) designers and software developers will have sufficient information to
make informed decisions about the wireless approach.

The process of
upgrading brownfield applications with new physical sensors will often begin
with the verification of the new sensors at the silicon designers. Referred to
as physical verification, this process involves a number of electronic design
automation (EDA) tools, including a design rules check (DRC), layout versus
schematic (LVS), electrical rule check (ERC), exclusive OR (XOR), and antenna
checks.

Fortunately,
there are many co-design suppliers that provide an automated suite of tools to
perform these checks. The verified design will then be handed off to a silicon
foundry and the circuits will be produced using an appropriate fabrication
approach to meet the requirements (e.g., extended temperature or shock and
vibration) of the original design. From here, the packaged chips go off to a PCB
designer for board-level design.

For highly
sensitive or mission-critical applications, the PCB designer must ensure that
there exists a validated and audited supply chain. In the past, there have been
documented cases of counterfeit chips, typically with lower quality and higher-than-normal
chip mortality issues, which compel developers of board-level solutions to
verify that the component suppliers are meeting the reliability requirements
for the components. This is particularly true for applications that are part of
the critical infrastructure such as power production and water processing.

Once the board-level
solution is ready, it will need to be subjected to a series of environmental
tests to validate the performance in the environments where the sensors will be
deployed. This includes thermal cycling, shock and vibration testing, RF
emissions and susceptibility testing (including FCC certification), and a host
of other tests depending on the severity of the environment where the sensor
will operate. For some applications, such as high humidity, the boards may also
need to be conformal-coated or “potted” to protect the circuitry from the
environment.

Assuming that the
sensor application will use some sort of microcontroller or other processor
(often needed for the encryption of communications or management of the
wireless solution), the software design team will need to make a broad spectrum
of decisions about whether or not an operating system will be needed, the
software design philosophy, the communications protocols, and how to handle
security. In addition, if this is a brownfield application, the software team
will need to consider the compatibility with the existing system and the data-collection endpoints.

Security is a
particularly tricky issue because we need to make allowances for software updates
in the field, provisioning the sensors/radios, use of encryption keys, etc. No
matter how many steps we take to thwart access to the device (special screws,
potting, tamper-resistant cases, etc.), we must assume that the device has been
compromised by the time that it reaches the field. This puts a special burden
on the provisioning approach to make the device as secure from a software
perspective as possible. This will require the generation of digital
certificates and/or the use of security circuitry such as smartcard chips for
secure key and parameter storage.

To summarize, the
update of a brownfield industrial application will require that the designer
takes into account a broad spectrum of design criteria. This ranges from the
desired compatibility with existing systems to
the rollout of the platform to even the supply chain if the application is part
of the critical infrastructure.

If
the plan is to replace existing cabled solutions with a wireless approach, the
designer will need to perform a number of tests to ensure that the RF
frequencies will propagate as expected at the site given the RF interference
that may be present for certain frequencies. And the designer would do well to
incorporate security mechanisms throughout the system to protect it from
would-be attackers. The application upgrade will not be a simple task. But paying
close attention to the details will result in a new solution that can both
replace aging existing equipment and provide significant cost savings moving
into the future.

Cover Image: Shutterstock