Jump to content

Andrew at MEA

  • Posts

  • Joined

  • Last visited

  • Days Won


Posts posted by Andrew at MEA

  1. The IoT and the Hype Cycle for Emerging Technologies

    Yet again, I’ve received an invitation to a gathering of knowledgeable folk who will present various high-level speakers to “share insights on the future of Australian life over the next 14 years, specifically the changes that the Internet of Things and related emerging technologies will have on the world around us".

    In August 2014, the IoT reached the ‘Peak of Inflated Expectations’ (according to the Gartner report on Emerging Technologies), predicting real product in the mainstream market 5 to 10 years hence (2019-2024).

    Here’s the story:


    Our Engineers Australia IoT forum will soon be one year old.

    Can anyone provide me with a list of successful Australian IoT technologies that have reached technological maturity and have a profitable business model?

    I could do with some statistics when I enter that roomful of IoT experts…

    Or is the IoT still only hype?

  2. A recent article in Electronic Design magazine by contributing editor Lou Frenzel is an excellent summary of technologies available for Low-Power Wide Area Networks (LPWANs). The article is called “Long-Range IoT on the Road to Success”


    There’s discussion on some new long-range Wi-Fi technologies called White-Fi and HaLow, as well as a useful table comparing the nine different technologies outlined.

  3. I also read this article when it appeared in print in the IEEE magazine 'Spectrum'

    What interested me was the return to the concept of 'on-site' data compression offered by this powerful computing platform.

    I agree with Heath that the size is interesting, but that's not particularly important in my field of IoT environmental measurements.

    I suspect that the real hint being offered here is that the power and speed of computation in the Cloud is being recognised as finite, and that on-site sensor algorithms (that we've had for decades within smart sensors) will make a big difference to the speed and utility of the IoT.

  4. The IoT and Open-Source Software

    I can claim no particular moral high-ground for using an Open-Source Linux platform for our latest IoT Hub, which links our on-farm Plexus ZigBee network through to our Green Brain web app in the cloud.

    Frankly, that was a decision thrust upon us by the modem manufacturer (Sierra Wireless) when they obsoleted previous generations of cellular modem chips (running the Open-AT OS), essentially forcing us to step up to their ‘MangoOH’ (Linux) platform.

    So, it felt a bit odd to receive an email from Sierra Wireless asking MEA for a product testimonial.

    To save time explaining Plexus, I simply sent him a link to this blog, and here’s what came back:


    “Hi Andrew,

    I have read and re-read your blog posts a couple of times since you sent this mail.

    Seems like you and your team have done a great job in building/maintaining/diversifying the platform and also carrying the torch for IoT in Australia. Kudos on that!

    One aspect that I would be interested to know is how the open source technologies like mangOH are helping your product development.

    Would you be able to provide some insight into that?

    As you can imagine that is a lot of effort that we put in (and it still needs more) but would be good to get input from people like you.



    So I wrote back, as follows:


    “Thanks for your kind words.

    I’ve had a quick word with the engineer writing the new Plexus Sierra Wireless WP8548 modem code under the Legato system (which is new to MEA)

    His comments were that:

    1. It’s reassuring to be using code that has been ‘stress-tested’
    2. However, this is somewhat offset by the pain of being an early-adopter, which often requires work-arounds for pieces of incomplete code that may be only partly documented. All with the risk of having to re-work stuff later.
    3. For the most part though, the new Legato platform and all the good tools that it includes is making development quicker.
    4. The real beauty of the open-source platform is that access to the vast existing code-base written by others allows one to learn, build, share, improve, contribute to the community. As a fellow engineer commented recently, he won’t even start a software project these days unless a community is involved.
    5. So, there’s ‘safety in numbers’…
    6. The licensing arrangements are a real head-spin; we’re engineers, not lawyers. Likewise, the fee structure for the Over-Air-Programming.
    7. Support for web-socket CLIENTS written in the C-language seem a bit scarce.

    Hope this is helpful.

    It’s all just the usual fog ahead, which happens frequently to us scouts.”


    This past week at MEA, we fired up our first prototype of the next Plexus Hub running under Linux. And yes, it’s doing the same job as the old Hub.

    While all this change, expense and heart-ache is invisible to our end-users, perhaps the benefits will be felt further along the product cycle in more rapid reaction to changing demands?

  5. MEA's Green Brain

    Inside MEA, our on-farm IoT product – Plexus - looks like a whole bunch of sensors, radios, hardware and firmware. Where there seems to be pain, that seems to be the game.

    Yet the real window into the IoT world is through our web-application named Green Brain. It’s Green Brain that our customers see daily, and their visits out to the paddocks to peer at our excellent hardware are virtually nil.

    So for old hardware engineers like myself, the IoT that we’ve so painstakingly built over these past five years has almost no impact on the customer, unless it fails and stops pushing up data.

    Yet this juxtaposition between on-ground hardware and in-cloud computing is entirely appropriate. Green Brain is the farmer’s shop-front into his on-farm measurements.

    So MEA is becoming ‘the Green Brain company’.

    Green Brain has been running for four years now, and over this past weekend, passed the half-billion record mark.

    I’ll just have to seek what comfort I can from Green Brain’s reflected glory…


  6. Hi Tim

    I'd certainly enjoy listening to your ideas, and will look you up next time I'm in Melbourne.

    But beyond just myself, there is a whole community of folk here interested in all aspects of IoT; perhaps you could reach us all by presenting a webinar on your experiences?

    Contact Tim Kannegieter or Geoff Sizer through this site to take this further.

    All the best



    The IoT in the Outback

    I’m in sheep country in far northern South Australia, walking the dirt tracks across the saltbush plains through Yalpara, Minburra and Menton stations to the Waukaringa ruins and back.

    Out here life is at its most elemental. The IoT seems to exist on a different planet. There is no mobile phone coverage and therefore no Internet, power is generated at the homesteads by solar or diesel generators and all communications happen by UHF radio or satellite.

    Yet the measurement problems found out here are, if anything, even more pressing, exacerbated by the tyranny of distance and the difficulties that arise in servicing and maintaining equipment and stock.

    I pass windmills and more modern solar systems pumping underground bore water into storage tanks connected to stock watering troughs. Sheep trails vector in on these life-sustaining sources of water, making grazing of these vast areas possible. Knowing that water is available to remote flocks of sheep and cattle is a quintessentially Australian measurement problem.

    I carry no backpack out here; I’m travelling with camel-drawn wagons that hold my swag and camping gear and food and water for ten days. The hassle involved in camel-transport has to be experienced to be believed, but it’s what my country cousins like to do, and because the camels plod along fuelled and watered only off the saltbush  and thorny scrub rather than requiring diesel fuel and expensive 4WDs.


    I’m invited along because these cameleers differ in one essential way from the early Australian explorers; they have a well-developed thirst for solar power to operate fridges, lights and torches, CPAP machines, and to charge mobile and satellite phones, cameras and GPS systems. And I’m the city cousin who knows how to build such gear and keep it working, as I have done.

    On these vast open plains, solar-powered satellite IoT devices will be the perfect solution, given only the prickly problems that need solutions at both ends of the chain. At the front-end, the problem is to provide affordable tank level monitoring at very low power levels and with zero maintenance for a decade or so. At the back-end, how do you effectively get data to folks who have such limited access to the Internet to check their data?

    So I entertain myself with these mental puzzles as I plod along for hour after hour while the country slowly unfolds around me. And as always, the solutions involve a multi-disciplinary approach that is the hallmark of so many IoT activities.

  8. If ever I saw a job description with a foot in both the engineering and marketing camps, that's it!

    I'd love to meet the person they hire; some pretty special qualities will be required with such a broad remit.

    I reckon Anat Efron from Thinxtra had those qualities (she was seriously impressive!), so I know there are people like that out there.

  9. The IoT and the Need for Heroes

    One of the hardest things to do when giving birth to new technologies – such as the IoT – is simply ‘to keep going’.

    In this series of essays on my own decades-long journey to get to the IoT starting blocks, I’ve talked about all the things that can and did go wrong in product development and roll-out.

    What I haven’t touched on is where I found the inner fortitude to keep surmounting a seemingly endless series of obstacles.

    This is not something normally discussed in the engineering journals.

    Yet the history of invention gives plenty of clues as to the vital role of mentorship and ‘heroes’ for those few engineers who actually managed to create working products when so many others fell by the wayside.

    Brian Thomin at AMDEL in the 1970’s and later Brian O’Neill at Monitor Sensors in the 1980’s were early role models for me. These senior engineers answered all my dumb questions and never once made fun of my naivete.

    I learnt much from these guys, but I learnt most of all from Jim Williams of Linear Technology.

    Jim was an engineer’s engineer, and endlessly patient in offering help to younger engineers. (Jim had no computer or email, but you could always ring him up at Linear Tech and ask for advice, if you had the balls. Few of us did…)

    But most of all Jim wrote; over 350 publications relating to analog circuit design, including 5 books, 21 application notes for National Semiconductor, 62 application notes for Linear Technology, and over 125 articles for EDN Magazine.

    He reached a lot of engineers, even those of us down at the bottom of the world in Australia.

    Jim never talked down to his audience or tried to bamboozle them with pages of mathematics. He used humour and simple explanations to teach electronic fundamentals through the real language of working circuits.

    I never met the man (he died of a stroke aged 63 in 2011) but I read everything he wrote - sometimes four or five times - as I tried to figure out how deceptively simple circuits worked.

    Much of the heavy lifting of the IoT will be done by young engineers skilled in coding and software.

    Yet the sheer breadth of engineering experience needed to orchestrate IoT products is likely to be found only in senior engineers who have come the long hard route to a deep understandings of how to make things really work.

    And I’ll bet many of them had been lucky enough during their careers to find older engineering heroes whom they could admire and emulate.

    Thanks Jim!

  10. Hi Tim

    I'm also wary of long manifestos that seem to have as their object the 'look at me!' objective.

    The IoT will take off only when there are customers.

    The best market research is a signed purchase order.

    As for manifestos, these are easy to spot: your brain goes to sleep mid-sentence. That happened just a half-a-paragraph into this press release.

    So I think you are on the right track here, by looking for the triggers that answer the questions of 'how indeed will IoT projects fire up?'

    That is, who's footing the bill? (because IoT product development is a darned expensive business)


    Andrew at MEA


  11. Thanks for the update, Tim

    I wonder when this CAT-M1 NB-IoT network will be switched on?

    Whew! Lucky we backed the Sierra Wireless horse for our Plexus product development; looks like SW is in there with Telstra getting in some practice runs in Tasmania (to shamelessly mix my sporting events metaphors!)


  12. The IoT and Fundamental Things

    Barrie Gilbert – inventor of the Gilbert cell found at the heart of all these potentially billions of IoT radios – was one of the most prolific electronics engineers of all time.

    For all those young engineers just starting out, and wondering how best to gain a toe-hold in this brave new world of the IoT, look no further than Barrie’s simple but powerful advice describing the source of his own inspiration, in an eight-page story entitled ‘Where Do Little Circuits Come From?’*

    “Discovering or inventing (- is there a difference?) new uses for a handful of transistors seems to be difficult for many young engineers entering the field of electronics today. Perhaps this is because they are taught that in confronting this Brave New Digital World they would be ill-advised to waste their precious college hours on such bygone and primitive notions as Kirchhoff’s and Ohm’s laws, or be concerned about such rudimentary concepts as the conservation of energy and charge, or bother to become adept in Laplace and Fourier analysis. The enlightened contemporary view is that everything of importance will be done in DSP sooner or later. Sadly, there is evidence to suggest that this message is increasingly being accepted. It is precisely the lack of these foundation skills, or even an awareness of the history of electronics, which makes it so hard for many new graduates to cope with component-level design, analog or digital.”

    “I get the feeling that the development of new circuit topologies is viewed by the newcomer to circuit design as something akin to magic. I’m thinking of those happy little tunes that weave just three or four active elements together in some memorable relationship, the themes, rich in harmonic possibilities, from which countless variations unfold. In these deceptively innocent and simple systems, cause and effect are inextricably bound: we are at the quantum level of electronic structure”      [Barrie Gilbert, 1991]

    Very little of the hype surrounding the IoT touches on just where the inspiration will come from for the billions of new sensors that will connect our gadgets to the physical world. There is a sense that these are all already out there on silicon, and that nothing remains to be done except to wait for their prices to fall to next-to-nothing to allow the IoT to take off.

    In my own long journey through the world of environmental measurements to the IoT gadgetry that we build at MEA today, I’ve done my day job by managing product development and mentoring the young engineers under me, giving them all the interesting and challenging work while I’ve covered the paperwork.


    But in my own time – in cafes, planes and at the kitchen table – I’ve worked for decades on the fundamental principles of new IoT sensors based on a single op-amp, working with the very DNA of electronics to fashion new sensors. In a series if papers in the international IEEE Sensors journal, I’ve shown how these simplest of circuits can be made to measure stratification in urban reservoirs, the hydraulic conductivity  and salinity of soils, and the sap flow and water stress within living plants. These days – down the back garden in my home lab – I’m happily entertaining myself finding new ways to create the ultra-pure sine waves needed to make electrical conductivity measurements in old-fashioned Wenner Arrays used to make the simplest of soil resistivity measurements.

    So my advice to young engineers wanting to ride the IoT wave is exactly that which Barrie Gilbert would give you; study the fundamentals and the history of your craft.


    *‘Analog Circuit Design – Art, Science and Personalities’, Jim Williams, editor. EDN Series for Design Engineers, 1991

  13. Hi Jason

    Thanks for your thoughtful advice on this matter.

    We have discussed this internally and believe that you have shown us a way forward.

    MEA will approach the various parties who own our weather station networks and discuss Creative Commons licences with them, while talking about how the data can be more broadly used for the good of all.

    No doubt there will be much discussion about 'fairness' and 'impartial' dealings between commercial competitors and so forth. We're OK with that.

    This is very fundamental issue for manufacturers of IoT technology, so your interest is appreciated.

    Best Regards

    Andrew at MEA

    • Like 1
  14. The Dilemma of Data Ownership

    For more than three decades, MEA has built weather stations for wind, solar and agriculture applications within Australia.

    In the past decade, there has been a shift from private to public ownership, with many hundreds of MEA weather stations being deployed within networks across whole agricultural regions in southern Australia. These stations give farmers up-to-date access to local weather data via websites hosted by various Government or statuary authorities interested in water use efficiency in irrigated areas and among many other applications outside of agriculture.

    Here’s an example from Western Australia: - DAFWA weather station network

    Other small private weather networks have begun to be rolled out in the last few years by cashed up agricultural companies whose mission is not to build weather stations, but to sell ‘decision support’ to farmers. For this to work, they have to fund the installation and maintenance of these imported automatic weather stations, then sell their agronomic services and seed and fertiliser products to recoup that cost.

    This is a very different model to that of a manufacturer such as MEA. We simply sell the hardware and maintenance services.

    While MEA may host public websites and data processing for such weather station networks, we have no ethical right to siphon off the data and feed it to others, such as farmers using Plexus on-farm IoT soil moisture systems.

    So we find ourselves on the horns of a dilemma; we know that farmers could extract extra benefit from all this wide-area climate data that our own systems are generating.

    But we don’t own the data because our business model has always been to simply sell and support the hardware.

    Yet another dilemma to be resolved in a world where data has more valuable than engineered products…


    • Like 1
  15. 'Plant-based' IoT

    One of the most interesting ideas that I've seen for powering tiny IoT sensor-radio nodes in forests and crops involves 'plant-based' energy sources.

    Essentially, a potential difference exists between the plant tissue in the stem and the soil, because of the different pH values within the plant compared to the soil. This potential is generated across the root membranes. [I might add here 'sometimes'. The actually potential difference depends upon the match between the soil pH - which varies with locality - and the plant's own internal chemistry. Sometimes the pH values of soil and plant simply match up due to coincidence, and there is no available energy]

    When I read about some University chaps messing about with this technology, I got pretty excited, because the very latest energy-harvesting technology can scrape up a working voltage of about 3V from just a few hundred millivolts of thermoelectric energy.

    So I made some measurements between grapevines and soil and connected up a low-voltage boost converter, only to have Mother Nature remind me that you can't beat the physics, no matter how exciting the potential application.

    TEGs (thermo-electric generators) have a very low source impedance, measured in ohms, while the plant pH generator has a source impedance that I measured to be in kilohms. The TEG-boost technology just flattened the potential without extracting any energy. Some other technology will be needed to make this work.

    Shame about that - it's the most perfect 'renewable energy', while the plant still lives. It would mean the IoT could run on a living 'bio-fuel'.

    Now if only I had time to work on such interesting electronic circuits...


  16. Hi Reuben

    3G/4G network connections work fine for ground-based applications because Telstra/Optus and their ilk go to the trouble of placing their base stations (cell towers) up high, at high density and with over-lapping spacing. So you are essentially transmitting through an 'air path', which is optimal for radio transmission distance.

    They are also allowed to use much higher transmission powers than ordinary mortals, so that helps.

    The difficulty (as Geoff Sizer has pointed out elsewhere) is that 3G/4G energy budgets are seriously heavy duty, making primary battery operation problematic. The modems themselves rarely cost less that $40 apiece.

    These conundrums will be answered somewhere around 2018 when the Telcos switch on Narrow-Band IoT services over these same cellular networks (its just a software tweak for them, apparently). You get all the advantages above but with an energy, monetary and data budget better suited to the IoT.

    The alternative is to go with a meshed network of your own, with the routers up high and the 'end-point' devices at ground level. If you can make it work!

    There is a tendency afoot to take radio connectivity too lightly, as though it is some sort of self-healing magic that will solve all your problems.

    Don't believe it! Check all the connections, no matter what effort is required. We still do this with every Plexus installation.

    If you just hope it will all work out, you will find your client breathing down your neck after you deliver the system, insisting that you fix all the 'off-air' nodes, and at your expense.

    Good luck


  17. Hi Reuben

    When Sigfox (Anat) visited us in Adelaide last year she had with her a little Sigfox-enabled temperature measurement device that she used to demonstrate connectivity to the Thinkstra base station on top of a tall building in the CBD. Our factory is in the foothills, so the transmission path was in clear air across the Adelaide plains to a high point in the city.

    So my message is that

    a) Sigfox can probably help you with a transmitter that will allow you to scout the terrain and assure yourself that you will get the needed coverage. (We always did this before installing our Sigfox-like radio systems more than a decade ago. No transmission path, no sale)

    b) this only works if the Sigfox base stations are already in place, and

    c) you can't beat the physics; look up the meaning of the "Fresnell Zone"



    • Like 1
  18. Afterword

    What will be the fate of the early innovators in the IoT arena?

    Are they a doomed species, to be pushed into oblivion when the big money turns its attention to grabbing market share and blowing away or gobbling up all the small fry who have been trying to create differentiated toe-holds as markets mature?

    This is a question that bites pretty close to home for MEA.

    MEA is a 5-year old start-up with 33 years of experience.

    Any company with some sort of longevity is by definition one that has re-invented itself again and again over the course of its history - markets and technologies pop in and out of existence like quantum particles, and good companies ride the growth of these waves and slip across to new waves as they collapse.

    What is clear – after a five-year effort to invent and stabilise Plexus Mark I as our latest on-farm IoT offering – is the need to do it all over again, simply because technology has moved rapidly onwards and we can drive down costs and increase performance by updating our technology.


    So here we are – down at the bottom of the world – and leading many other international agricultural IoT players by a full generation. That’s an edge we will lose if we don’t keep moving and advancing.

    To be a market leader, we will have to export Plexus beyond Australia’s shores. That will require a whole new generation of ‘smarts’ to make the technology simple and seamless. The MEA Law of Exponential Aggravation (known to a previous generation as ‘the Tyranny of Distance’) will see to that.

    So, stepping up to become an exporter will be a really tough call for our engineering, marketing, management and service folk. We’ll need all of our combined years of experience to carry that off… and a bunch of new investment.

    Can we do it?

    I’ve no answer to that; the future in a small business always seems perilous, but we’ve survived and gained a solid reputation.

    What I haven’t mentioned in this series of IoT essays is the over-arching need for sound company management.

    And that’s not me; I’m a starter, not a runner.

    Inventive folk such as myself are highly trained to jump sideways, circumventing problems with new off-the-wall solutions. This makes for a chaotic management style; all my moves appear to originate in left-field.

    MEA is a partnership, and my business partner is the solid methodical detail guy who runs our management and marketing and financial and strategic departments, while I have been allowed to focus on the engineering I love.

    Yet he and I will both reach retirement age in the next few years, and there’s a real need to re-invigorate the company with young people and fresh ideas. We can find those here on our doorstep.

    And my own fate?

    I’ve pretty-much lost interest in owning a company, but I find my hunger to develop new products burns more strongly with each passing year. I’d like to retire gracefully to the lab, to teach and to mentor young engineers, to learn new skills, and to create all those environmental sensors that fill my notebooks.

    I can be the Colonel Sanders on the Kentucky Fried Chicken bucket, but I don’t need to fry all the chicken myself.

    • Like 1

    The IoT and the 5% Rule

    I love old adages and, as I get older, I love saddling younger engineers with them, whether they want to hear them or not.

    Yet during 2016 – now deceased – I was painfully reminded of my own favourite: ‘The last 5% of any job consumes 95% of the effort’.

    2016 was a year burdened by the most extraordinary efforts, with what seemed to be almost no forward progress. We were busy cleaning up the last 5% of the Plexus project.

    The list of tweaks is long and tiresome and I’ve no wish to re-live them here. Suffice it to say that those efforts raised the bulwarks against future in-field failures through product recall, product revamp, hardware and firmware upgrades, re-tooling, re-jigging, re-thinking, re-engineering, re-documenting and re-vamping. It was enough to make a man think about re-tiring!

    So we got through all that and the Plexus on-farm IoT system is solid enough for us to keep marketing it until we obsolete it ourselves with new product offerings.

    Down in the MEA basement – where all the MEA technical folk live – I once again call in the scouts whose mission is to roam ahead of our main force and understand where technology is going. In the rapidly-changing world of the IoT, this is a critical role and best done by the young bucks. It will be their world.

    Mine is the larger role of digesting all this techno-babble and regurgitating it in its simplest and mildest form within MEA’s product development team. This is a very small group of three representing management, marketing and engineering. Our task is to answer the question ‘where to from here?’

    One of the other painful lessons of the past few years has been to learn that an engineering perspective has only a limited bearing upon these larger questions. Marketing folk present the customers’ and competitors’ perspectives and management must answer the question of how new product development will be funded, its likely profitability and our chances of affording it.


    In a small self-funded company like MEA, there is very little room for misjudgment here, so there’s plenty of round-table discussion going on, sometimes heated and sometimes for months on end. Talking is cheap, so we do all that up-front before we start shelling out our hard-earned dollars in actually creating hardware and software.

    I’ve been in a slump these past months, worn down by the inexorable pressures to finish the existing product development while totally fogged in over new product directions. However, I’ve lived with this sense of uncertainty often enough throughout a forty-year engineering career, so it no longer alarms me. It’s a precursor to that ‘Eureka’ moment brewing in my subconscious, where all these disparate pieces of market and technological insight that have been swirling about looking for a place to land, land.

    The world of the IoT has changed markedly since I was brooding away in a similar vein back in 2011, prior to giving birth to Plexus Mk I.

    Now I’m spoiled for choice with product offerings, multiple radio protocols, mesh- and star-networks on the same platform, 2.4 GHz and sub-GHz radios all on the same silicon, free software development environments and a huge reservoir of code and technology provided free by the big players, just to get us started.

    These are exciting times to be an IoT engineer, if such a beast actually exists.

    What the Plexus project has taught us is how to be a manufacturer.

    And to be one of those, you truly do have to finish off the last 5% of every product development, because you won’t get to play in Round 2 if you don’t.

  20. The IoT and those tiny radios

    Out the back of MEA stands our test yard, a fenced compound accessible from the factory floor via a ramp, and fully accessible to sun, rain and storm.

    For the past thirty years, it’s been this MEA test yard that has kept our reputation alive for being a reliable supplier of measurement systems for the bush. It’s our first line of defence against the horrors of system ‘out-of-box’ failures.

    Stated simply “the degree of aggravation for the service department rises exponentially with the distance from MEA’s back door.”

    Through this test yard have passed countless MEA weather stations, wind and solar monitoring systems, irrigation, hydrographic, stratification, soil moisture and custom measurement systems of all sorts. All destined for far-away places where they would be erected and put directly into action. The two weeks that systems spent in the test yard was our guarantee that we hadn’t missed any basic wiring cock-ups, unseen sensor logging or communications faults or system programming errors. Likely field faults were therefore reduced to those that occurred in transit or due to operator-error at the other end. We had test data to refer back to.

    Yet now the MEA test yard stands empty; Plexus and the IoT have changed all that.

    Test yards and custom-engineering go hand-in-hand, but how do you test IoT radios when you change lanes from custom measurement engineering to becoming an IoT manufacturer? Stuff is now rolling down a production line rather than receiving the tender devotions of a dedicated technician.

    This turns out to be a production measurement issue of high order.

    Our first pass at the problem was to assume that the ZigBee radio manufacturer (the mighty Texas Instruments) put all that comprehensive radio testing upfront, and that all we needed to do was to turn the handle on the sausage machine and Plexus radios would roll off into their shipping boxes.

    Pretty soon the rule of ‘exponential aggravation’ kicked in and we seemed to be having issues with weak radios under field conditions. Secondary lightning strike during storm activity was implicated in some cases, damaging the radio ‘rf front-ends’. But there seemed to be production issues involved too. This was likely to be the same old numbers game; the more systems that get shipped, the more likely you are to find the faults in that small number of units that aren’t fit for duty.


    Plexus radios were being 100% tested on the production line, but only over a through-air radio path of a few meters. Weak radios can make this hop, but not the 1 km expected of them over flat country under field conditions. In a mesh network, a weak radio can take down a segment of a whole network by failing to keep its end up.

    But how do you test radios short of the onerous old business of actually physically separating them by a kilometre?

    Turns out the old ways are best; conduct a full system test before shipment of a customer’s order, but with a twist. We do this inside MEA, and not outside in the test yard.

    It transpires that pushing 2.4 GHz radio traffic through double pre-cast concrete walls – with all their internal steel reinforcing mesh - is a suitably tough attenuation path to weed out under-performing radios.

  21. Behind the scenes of the IoT

    It’s the week between Christmas and New Year, and I’m alone inside MEA, manning the service desk.

    It hasn’t helped that a storm ripped through Adelaide on Boxing Day, shutting down power and phones to our factory in Magill and disabling all communications to the outside world, except for my solitary mobile phone.

    So I’ve sat here in the dim light coming through my office windows, and taken calls from across Australia from guys still in the field and grappling with installation and service issues. They’ve been directed to my number by a jolly message on the front page of the MEA website, deployed from outside the war zone by one of our own who lives where the power has remained on.

    It’s here at the front line that I find myself most grateful for all the diagnostic tools that we built into our Plexus-Green Brain on-farm IoT products. These enable me to dig deep into problems and kick-start systems remotely, or order up a swap-out system as a last resort.

    From a ‘big data’ standpoint all these network codes - battery voltages, fuse checks, power flags, measurement status bits, error and timing violations flags – are just so much redundant baggage riding pointlessly along with real data.

    But at the service desk, these tools – and those that allow us to look behind the scenes within the Green Brain databases - are an indispensable facet of determining where problems lie and what actions might dig us out of the current fix. All this while the installer is standing in the hot sun waiting for a call back and some lucid and succinct service advice.

    But that’s not even the really good stuff!

    At the highest level – seen by the customer – Green Brain itself combs through incoming data and ‘colour flags’ sites that show sensor faults or have failed to report in for over 12 hours. Green Brain can also pro-actively generate alarms by SMS and email, triggered by user settings.

    At the hardware level on ground, we’d set up our Plexus Hubs to respond to simple text message commands, which they deal with immediately. We can change network logging rates, query firmware versions, check times, kick off network resets, force unloads, generate reports, clear memory and alter network timing parameters – all remotely.

    Simplifying all this is a phone app, built by one of our senior technical officers in his spare time, just for the fun of it.

    Through all this, Plexus remains invariably polite, responding with a cheery ‘OK’ when called upon to make changes, large and small.

    Yet the future holds even more magic, known in the trade as ‘over-air-programming’ or OAP. This will allow bug-fixes and new features to be disseminated globally across Plexus networks while remaining invisible to the user.


    Need I say it?

    The data sheets for our current modems exalted their ability to conduct over-air-programming, but we could never make it work. Nor could the company’s own applications engineers; the possibility was eventually abandoned.

    Down in the MEA lab, we are working with the very latest generation of new OAP-enabled modems, while secretly harbouring fears that the gap between the promise and the reality of OAP will be filled with hot air.

  22. Sensors and Standards

    Sensors sit at the interface between the real world and the parallel IoT universe.

    And it is here – at this most messy of junctions – that Mother Nature wreaks maximum havoc upon young unblooded IoT enthusiasts.

    Here’s where raw sewage enters IoT systems and - in my view - no amount of artificial intelligence or post-processing of data can rectify the loss of that small piece of history.

    I like sensors. I like designing sensors. I like figuring out how to run them forever on the smell of an oily electron. I like deploying sensors and studying the data they produce while grappling with the meaning of what they’ve been reporting over time.

    So, it is with some sense of restraint that I refrain here from succumbing to that very personal enthusiasm for sensors themselves, and restrict myself to just talking about that other interface, between the sensor and the IoT transmitter.

    In my field of environmental measurements, ‘smart sensors’ talk to data loggers and Plexus field stations over an ageing three-wire serial interface known as SDI-12, short for ‘Serial Data Interface at 1200 baud’. This is an industry standard first released in 1988, following development by a coalition which included the U.S. Geological Survey's Hydrologic Instrumentation Facility and a group of private companies including Campbell Scientific. The SDI-12 Specification is maintained by a non-profit corporation called the SDI-12 Support Group.

    So there is an ‘SDI-12 Standard’ and, in principle, all sensors and all data acquisition systems adhere perfectly to that standard and work seamlessly and interchangeably.

    If only that were true…

    MEA qualifies sensors for operation on our Plexus on-farm IoT systems, and they must pass through a rigorous compliance testing procedure to get a ticket to ride. Any deviations from the SDI-12 standard are reported back to the sensor manufacturer during this process.

    But occasionally, other folk choose other on-farm sensors, and expect Plexus to make the measurements and carry the data load. Here’s where full compliance to the standard is critical for interoperability between different manufacturers.

    In this case study, mysterious data dropouts stacked up the barricades and ramped up the rhetoric between MEA, the sensor manufacturer and our common field agent. Who’s responsible?

    It is here – within these problem jobs – that good companies prove their mettle, despite the agony that goes along with having their brand tarnished in the marketplace by a system that’s simply not delivering critical irrigation data.

    This problem was hand-balled back to the engineers in MEA’s product development group, and we made trips to site and set up long-term tests in the lab back in Adelaide. Months passed as we waited for faults to show, and for theories and remedies to be tested.

    This is where the Standard is so useful, because he who conforms best holds the moral high ground.

    In this case, the Standard is actually embodied in a piece of hardware one purchases from the USA; it runs both SDI-12 sensors and recorders through exhaustive timing and response tests. All sins are revealed. A detailed report is generated.

    SDI-12 verifier.jpg

    MEA’s Plexus sensor interfaces passed with flying colours, but the sensor’s SDI-12 interface was in serious breach.

    But in this IoT business, who’s right and who’s wrong matters little. Fixing the fault and getting data to flow is all.

  • Create New...