Jump to content
  • Agriculture (MEA)


    Tim Kannegieter

    Description: Three years ago MEA successfully launched an On-Farm Internet of Things system called Plexus. The system moves soil moisture and climate data across the farm and allows farmers to access their irrigation data from anywhere, at any time. The product was the result of a three-year project, a multi-disciplinary approach that combined electronic, mechanical, communications, software and web engineers plus external industrial design skills, a manufacturing link into China and all sorts of technical skills to set up the production line.

    Over the past three decades MEA has built more than 2500 systems for wind and solar monitoring, irrigated agriculture, remote weather station networks, a web-enabled radio measurement system for Australian farmers, novel soil moisture displays for African farmers, salinity sensors for farmers in Bangladesh and a new-to-world sensor for getting living plants to talk to irrigators about their state of water stress.

    Source: Based on a webinar delivered on 9 August 2016 to the Applied IOT Engineering Community of Engineers Australia.

    Presenter: Dr Andrew Skinner, Engineering Director, MEA (Measurement Engineering Australia). Dr Skinner founded MEA (Measurement Engineering Australia) in 1984 to carry out the three-year SA Wind Energy Survey. While running the company full-time, Andrew completed a Masters Degree and PhD in environmental measurements. His forty-year engineering career began in 1971 with studies in electronic engineering at the SA Institute of Technology. Andrew was awarded a Dean's Commendation for his PhD thesis from Adelaide University in 2009 and was the Engineers Australia South Australian Professional Engineer of the Year for 2015.

     

    Title: What does it take to be an IOT Engineer? An agricultural case study.

    Background

    I'm going to start today's lecture with the history of the product, explaining how a small Adelaide-based company came to be involved in the Internet of Things (IOT), and then progress to a full end-to-end description of the development of an IOT product.

    I founded MEA in 1984, after a long history in mining, because I wanted to do environmental measurements in the bush. MEA’s first major contract was to carry out the 1984-87 wind energy survey, which played a role in South Australia's rise to dominance as a renewable energy player in wind. We did that for many more decades with the purpose of finding the windiest spots in Australia.

    Back in 1984, there was no internet, no telemetry and we didn't even have hard drives. What had arrived was the first IBM PC, and so we had a place for data to be analyzed.

    10 (2).jpg

    Early measurement technologies included desktop computers and data loggers (pictured here to the right of the PC). Pictured is Dr Andrew Skinner circa mid-1980s.

    In addition to the PC, the invention of the micro controller made possible the invention of data loggers, which contained a battery, measurement devices, memory and a communication system. In other words, all of the basic elements of what is now an IOT device. However, without any way of communicating with these data loggers in far-flung places, we collected data by shuffling loggers backwards and forwards by truck, train and post, then unloaded it into fairly primitive software on these very early PCs.

    There was then a progression of technologies that allowed us to do communication with these remote measurement sites in the bush. The introduction of laptop computers around 1994 helped and we moved to cabled modems whipping along at 1200 baud but we really only got to be wirelessly connected in the 90s with the arrival of analogue cell phones.

    10 (3).jpg

    Pre IOT technologies are demonstrated in this wind monitoring site in Leigh Creek in rural South Australia. Shown is a laptop using proprietary software for collecting data on the hard-disk and trenching equipment for installing a copper landline.

    From 1993 on we had 2G cellular phone telemetry. Telstra, Optus, and others started to blanket the country with powerful transmitters and receivers, and we could now start to move wireless data across the land by GSM in the towns. What really opened up the bush to measurement was the arrival of CDMA radios. When 3G arrived we had coverage almost everywhere and this gave us the ability to make measurements and to get data easily from all sorts of remote sites.

    10 (4).jpg

    Left is a site measuring snowfall high in the Snowy Mountains, and bottom right there is a buoy measuring water stratification in a suburban reservoir.

    However, in the remote areas of South Australia, out west, and far north-west, we still have to revert to satellite telemetry. It's broadband data, which we didn’t really need, and it's expensive.

    Packet Data

    The next step in this evolution of technology was to avoid having to dial in to collect data and to automatically send packet data to ftp servers. For example, we had a wind turbine site on the southern Great Australian Bight, which had turbulence issues. We just didn't have capacity to store all of that high speed turbulence data on-site. So we arranged to have the data continually offloaded to FTP servers, where it would be stored ready for analysis.

    This has allowed us to deploy weather stations in agricultural regions providing real-time climate data to farmers and irrigators in particular. There's now something like 300 weather stations in all sorts of agricultural regions and farmers can now go to a website and get up-to-date information that's very local to them.

    That was very helpful to farmers but we then started focusing on getting data across individual farms. For this we needed a different approach, because a cellular connection at every measurement point is too power-hungry or too expensive.

    For gathering moisture data across a farm, we began to experiment with sub-gigahertz radios in the 433 megahertz range. This is a battery-powered box, sensors in the ground and a whippy antenna that pushes over in front of machinery. This data was transmitted randomly and redundantly back to a data logger, which again pushed it up to a FTP server where it could be got at.

    In wide open areas we used a star network – a single point in the middle which is collecting the data. Each of the points has to have a line of sight, or nearly line of sight, connection back to the middle. This is very robust with few data outages. Seven out of eight of these sites could fall over and the eighth would still continue to push its data through.

    These devices could operate for a year on batteries, but they had no data storage and they had no repeaters. If the aerials or any part of the system were down to allow for farm operations, that data was lost. So this was one driver in the evolution toward true IoT devices.

    Data loggers

    Our next product was the GBug, which was a really low-cost data logger measuring soil moisture, running on a PP3 9V transistor battery. It would store three weeks of soil moisture data recorded every two hours.

    10 (5).jpg

    To get the data out the GBUG datalogger, a farmer would tap it with his knuckle. That would wake up the radio. It would radio all of that data on 433MHz radio to a hand held retriever. He could then plug the retriever via cable to his computer, unload the data, and update his database.

    This system was deliberately designed to trade capital for labour. The farmer is putting his/her own time into the mix to get the cost down.

    10 (6).jpg

    With GDOTS, a farmer can pass by on a quad-bike and tell at a glance by the number of dots how full the moisture in the profile is, and can make decisions on the fly.

    The next product we developed was the GDot, which is a very simple visual display powered by two AA alkaline cells that run it for about five years. It sits over a single sensor in the soil profile measuring soil moisture.

    Drivers of the IOT

    In 2011 we decided to conduct research as to what players in that industry really needed. What came through was that gear had to be really robust, had to be “set and forget”, had to operate for five to ten years without any problems, and naturally they wanted it really cheap. But what they really wanted was to be able to get at the data any-time, any-where.

    2011 sounds like only yesterday but at that time we were uncertain about whether smartphones would be used or even accepted by all. It seems odd in this day and age where everybody has got one in their hand. Nevertheless, we had to join the link between sensors to measurement devices across the farm, up through the internet, through the cellular phone network, back to a server in the cloud where the data would be stored, and allow the farmer to take his phone and collect that data and see his soil moisture or microclimate data right then and there.

    Plexus

    We had to create such a system and what we built was a product called Plexus, after the solar plexus, the nerve centre in the human body. The field stations are solar powered from a small 400 mW solar panel, that runs the system by storing energy in a lithium-ion rechargeable battery. It has to run year-round, through darkness, through stormy periods of low charging and through winter.

    10 (7).jpg

    The Plexus field stations moved MEA’s industrial design forward

    The system included getting data from the field stations to a cellular gateway, out to a server in the cloud, and into a web application where the farmer could get it.

    The beauty of being online is that we can SMS the system and tell it to shift gears, and to download log files, and to do all sorts of useful things that help us solve problems.

    Mesh networking

    We were aiming for a world market and that meant we had to be careful about selecting our frequency of operation. The one frequency which is accepted on the ISM free-to-air band all over the world is the 2.4 gigahertz band because ZigBee, Bluetooth, and Wi-Fi are all on that band. It gave us various problems because it's a very short wave-length and travels in a straight line, which is not always possible on a farm.

    The solution was to use a mesh network (pictured), which we were able to do thanks to the development of ZigBee and the big integrated circuit manufacturers creating suitable radio products.  

    10 (8).jpg

    Because it's a mesh, each of these nodes in the ZigBee network cooperate to move each station’s data back to this Hub in the right hand corner (pictured). The Hub's job is to collect and store this data and push it up to the cloud, to our Green Brain application where it's available to the farmer.

    If one node happens to be down for farm operations or for some other failure, the network self-heals and data passes around that stuck node and arrives back. In the meantime, that node continues to log, and once it's put back up into the radio space it pushes that data back up to the hub where it's sorted out and pushed on to the web. This is a true on-farm internet of things product, based on ZigBee sensor to web technology.

    To make this all work, the first thing is that we had to increase the Zigbee range. The normal ZigBee range is about 10 to 100 meters. We needed at least 1000 meters. So we had to boost the signal and increase the reception sensitivity.

    We had to get field stations to run for 10 years on the solar power and battery. So we had to store every electron and every photon that turned into an electron had to be stored in the battery, so we had to work very hard at the power systems.

    The real challenge was that with the Zigbee industrial standard, the routers in the system are always powered on. This was never going to work within our skinny power budget, so we had to work at Zigbee’s applications layer to reknit the network for one minute in every 15-minute measurement and shut the whole thing down for the remaining 14 minutes to meet our power requirements.

    The hub within that ZigBee network controls the network. It's two-way traffic, and the hub's job is the synchronize the network, make sure each node is talking, has their measurements and other diagnostic information (such as battery voltage) ready, stores all that, and essentially just pushes it up to the cloud.

    We're don’t have real data loggers at field level here. We're just trying to move data up a pipe. We're using smart sensors connected to a bus structure called SDI-12 which has three wires being 12 volt, ground, and data.

    Sensors

    What Plexus has enabled is the ability to integrate a whole range of measuring devices we have developed over our 30-year history, and we're still adding to that list. From frost alerts, pump status, pressure and flow, obviously with rainfall, but disease pressures, spray drifts, climate records, chill units for fruit and for stone fruit particularly, evapotranspiration.

    10 (9).jpg

    The Plexus system is able to connect measuring devices across a range of applications including those listed here.

    Three are sensors that haven't been invented yet for things that I know are wanted but we are only just beginning to think about. For example, knowing when a plant is experiencing water stress and environmental issues like deep drainage and nitrogen levels in soils.

    Mounting Structures

    We also had to get the mechanical engineering right so it would survive the environment and meet the farmer's needs. One of the big challenges was the mounting structures. In tree crops you need a very tall six to eight-meter mast that could be erected by a single person.

    In very short crops we've got different problems because the spray booms and the picker carts are really low and these Plexus field stations have to fold over and get out of the way.

    10 (10).jpg

    Machinery in a vineyard pushing over a Plexus field station mounted on a pole, which springs up again.

    We had six months of engineering to build a method to allow that pole to be knocked down and to spring back up once the machinery had gone over it. We could not allow Plexus to interfere with farm operations. Farmers just don't want the aggravation.

    In row crops you can just lift the Plexus pole out, lay it along the top. Even in strawberry crops we had to invent a very springy bracket to allow the picker's carts to roll over the top of them.

    Your sales staff also have to know how to diagnose faults in Plexus, with support from guys back at headquarters. They have to know how to install these long capacitor or the gypsum block sensors, and they have to solve problems in the field. We had to invent not just electronics and firmware and software but we had to invent mechanical structures.

    Web interface

    We also had to create a web-based interface we call the “Green Brain”, so that the farmer didn't have to download apps and mess around with software. They could actually just go to a website, which would serve him the data he wanted.

    We had to hire a brand new type of software guy, specializing in web applications, unlike the PC-based software that we'd used for decades. This was a whole new class of activity and we needed a very simple front end. In the MEA’s Green Brain web application, the farmer goes on and he sees his site, and he can see at a glance which nodes are live and which ones might be having trouble.

    line.jpg

    A vineyard in South Australia. The green lit nodes are saying, "We've heard from these stations recently."

    The vineyard pictured is where we did all of our testing. We went up there and the vineyard owners were very generous letting us run around trying to make all of these nodes connect across a large 250Ha property. We needed them to cooperate and jump around tree lines, over ridges, and to cooperate to get data back to the hub over a distance that amounted to about 4.2 kilometres. We had quite a few nodes in the system, and we went back many times to trial the technology before we even let it anywhere near a customer.

    Green Brain does what you expect. It shows the soil moisture data over some time frame in different graph forms (pictured below). It allows you to pick out sites. It allows you to look at how the batteries are charging, how the soil temperature has varied, etc.

    10 (1).jpg

     

    Manufacturing

    A major shift in our business was to shift manufacturing off-shore. We wanted to create an international product which looks slick but could manufactured be at a very reasonable price. To do that we had to shift from being a custom integrating firm and learn about mass production. We had to acquire a whole raft of different engineering talents (including those pictured below).

    10 (11).jpg

    For example, we had to learn about injection moulding in order to build low-cost robust polycarbonate enclosures, with silica gel to keep the inside dry. We had to learn special circuit boards, special charging technology, and then we spent a year and a half building the test and measurement systems that would qualify every single Plexus for the bush. We made use of a 3D printer to create test jigs and fixtures, all talking to a computer and qualifying the information. As part of developing all these skills needed to create IOT products I had to learn about a whole lot of new terms.

    list2.JPG

    A list of new words Andrew Skinner needed to learn in order to understand how to build Plexus.

    Electronics

    The printed circuit boards had to be built from the ground up. If you're going to make an international product that can be sold at a reasonable price, you can't just go out and buy modules, because you're adding too much of a fixed cost into the manufacturing cost. We had to design at chip level. This board contains all the ZigBee radio and the ZigBee stack, the range extender, main processor, technology that allows us to keep the battery in top condition, the 2.4 gigahertz aerial, etc. Hubs consists of two boards.

    10 (13).jpg

    The printed circuit boards inside the Plexus field stations

    The shape of the board was part of the industrial design, so that it wouldn't snag on over row equipment. Special connectors, poles, we had to build everything, even these connector plates here.

     

    The future

    MEA has now begun the second generation of technology. We've heard at loT about Low Power Wide Area Network technologies recently. I like the look of the narrow band IoT technologies, but we may stay with ZigBee and cellular, or 6LowPan which is a variation of ZigBee.

    Satellite has been terribly expensive but in Adelaide we have a company called Myriota funded by Canadian money, coming out of one of our local universities. Myriota have done some really clever things with very low power VHF technology and satellite links to these polar orbit satellites that are the size of a small carton. This system uses a VHF modem to connect to standard sensors, pushing data up as the satellites pass over every hour or every three hours. They're inexpensive and the cost of getting the data back via the ground stations and the internet is very reasonable.

    Technology like this will open up measurement in the bush for those places not served by the cellular phone network

    What have I learned from forty years of engineering that will help in the future? I've learned that things go wrong but you can't give up. You've got to keep going back and fixing things, making it right, paving over the cracks. Change is inevitable. Good companies strive to stay relevant, which means you've got to continue to stay on the technology curve.

     

    Edited by Tim Kannegieter



    User Feedback

    Recommended Comments

    There are no comments to display.


×