A New Smart Metering Home Area Network Perspective

Back in the heyday of the North American smart meter roll-out of 2009, you couldn’t say “smartmeter” without bringing Zigbee and home area networks into the discussion.  Many meter companies as well as utilities had already decided to roll out smart meters before the 2009 American Recovery and Reinvestment Act (ARRA) .  The influx of funds were gasoline on a fire that was already beginning to burn.  Meter companies produced meters by the hundreds of thousands each month to be installed by a relatively short list of utility companies.

Around these smart meters was supposed to be the holy grail of customer interactive demand response, a new concept called the home area network or HAN.  The idea of a HAN is that the utility provides connectivity between the smart meter and other devices in the customer premises.  These devices could be indicating devices such as a display to show energy usage and cost, load control devicse that measure usage and control power to certain high demand loads in a house like electric HVAC and water heaters, and even other utility meters such as gas, water, and heat.  Not only did the utility companies lay the groundwork for interacting with their customers homes, they set up a data pipe for aggregating data from other meters so they could generate revenue for just handling data.

It all sounds pretty good, right?  A network that allows you to interact with the utility company with the goal of saving energy, living more green, and saving money.  Why then we ask do we not all have smart meters and use devices on the home area network (HAN) every day?  And what has happened since 2009 to stagnate the rollout of smart meters in the US?

Let’s look at how many large utility companies rolled out smart meters and where they are located.  The cost of a kilowatt-hour of electricity varies widely from less than 7 cents in Louisiana to more than 34 cents in Hawaii.  ( Source: Electric Power Annual. Energy Information Administration, Washington, DC. Nebraska Energy Office, Lincoln, NE. 2012 data. ) And the amount of electricity used in a typical home varies greatly depending on warm season cooling requirements and the price and availability of other fuels such as natural gas, oil, and propane for cool season heating.

A good deal of energy consumption is determined by human factors including historical reasons.  You find a large number of people who use propane as a heating fuel even though electric heat pumps are less expensive to operate in many areas.  There is the waiting period for older equipment to need replacement and also the lingering impression that gas heat makes for a more comfortable home than electric heat because of early heat pump performance or because of the high cost associated with older electric resistance heat.  I doubt that you could tell by walking into most homes built in the last 20 years whether the heat is from electricity or a fossil fuel source such as oil, coal, or gas.

The point is that customers are extremely complacent with habits and thinking when it comes to energy usage.  And there are not too many reasons to switch unless you live in certain areas.

Here in North Carolina, the electricity rate is in the lower third of the nation at around 10 cents per killowatt-hour.  And the utility companies offer customers the option of a time of use rate that rewards the consumer for using electricity at off peak times, mostly in the early mornings, early evenings, and night.  If you have a dishwasher with a delayed start feature, you probably already use it and this makes for an ideal use case from the consumer’s perspective.  And for 2 income households with no-one home during the day, air conditioning settings can be adjusted so that the cooling energy demands are shifted away from the mid-day high demand period and into the times when the residents are home.

The problem I see with voluntary demand response is that people are complacent and afraid to sign up for it or change to it.  There is a fear that the utility bill might actually go up instead of down if you don’t fit the ideal situation for demand response.  Some possible solutions for people who need to keep their homes cool during the high demand periods of the day could include alternative sources such as wind and solar or local energy storage.  But these are expensive options and could actually lead to further costs and inefficiencies in some cases.

In an upcoming article, I will make a case for devices that allow customers to kick the tires on the time of use rate plans and decide if they make sense in their particular situation.  I will revisit that topic soon.

We started off this discussion around the home area network and Zigbee.  So why have I digressed?  Let’s connect again to that HAN and think about where we are today with regards to the 2009 smartmeter rollout:

  • Ask the question as to whether Zigbee is still compelling and whether it was ever compelling to the consumer?

Zigbee was developed as a personal area networking standard and was adopted by companies in the lighting and home automation markets.  Neither of these markets was mature in 2009.  So smart meters was the killer app that was going to make Zigbee successful.

Zigbee was touted as interoperable, robust, and secure.  A new Zigbee profile for smart energy was created and a layer of security and authentication was placed around it because certificate based authentication was a desirable feature that allowed the utility companies to exactly control what devices could join the home area network.  As a Zigbee alliance member, a proven solution by Certicom was the one chosen.

The promise of interoperability and trouble-free networks marketed heavily towards the utility companies created quite a pull within the industry.  All of the sudden, meter companies were being driven by their end customers to have Zigbee Smart Energy in their meters.  In some cases, the utility companies were even influencing the choice of Zigbee silicon and protocol stack providers.

Fast forward 5 years to 2014.  We w0n’t dwell on the arduous and lengthy process through which the Zigbee Alliance drove itself for Smart Energy 2.0.  The idea was to “fix” what was wrong with Zigbee Smart Energy 1.x.  But it was more like a complete re-write. Transitioning from the Zigbee 2006 underlying stack to an IPv6 stack that depended on unfinished work from the Internet Engineering Task Force (IETF) to implement a robust and secure solution.  In retrospect, was the move to Smart Energy 2.0 a move to preserve the Zigbee installed base by using a supposedly interoperable and physical layer agnostic IPv6 stack?  I think that some members of the Zigbee alliance saw well beyond the IEEE 802.15.4 RF physical layer that was the only one that the original standard allowed.  So when the Home Plug alliance, with their broadband power-line communications, merged with the Zigbee Alliance they certainly had smart energy home area networks in mind.

Home plug was another one of those standards that had not quite caught on like Zigbee.  So looking for a killer app like smart energy was a natural thing to do.  And with TCP/IP based protocols that could be routed over different network physical topologies, the vision for a new era of smart energy networks was born.  An emerging extension to the IEEE 802.15.4 standard called IEEE 802.15.4g was emerging and would reach final status some years later and extend the network to include a neighborhood area network or NAN.  IEEE 802.15.4g encompasses some of the original IEEE 802.15.4 standard, namely the 868 and 902 MHz ISM band channels originally set aside for a Sub-1GHz Zigbee band and the O-QPSK modulation technique from the original IEEE 802.15.4.  Lots of extensions includeing more modulation schemes (FSK, O-QPSK, and OFDM) were added in the interest of interoperability and robustness. And frequency bands were added to include those already in use in smart energy networks around the world including Japan (920 MHz), China (700 MHz), Europe (169 MHz), and the Worldwide 2.4 GHz range.

So we ask, what is the future for Zigbee Smart Energy and will the Smart Energy 2.0 profile be deployed widely?  One of my previous articles makes reference to the smart metering networks being the original Internet of Things.  This was true as there were not widespread networks of low cost embedded devices with two way communications until the deployments of 2009 in the US.  Coincidentally, the utilities were the original “Big Data” companies managing huge amounts of collected metering data and translating it into bills to their customers.   Today, there are a lot more networks of things than just utility meters.  And the software and servers processing enormous amounts of data have expanded.  The consumers are becoming familiar with big data, cloud computing, and the Internet of Things through everyday products they can install in their homes such as connected lighting.

Zigbee is playing a part in connected lighting for three reasons:

  1. Lower Cost than WiFi
  2. Low Datarate Radios are able to function correctly at elevated temperatures inside lights.
  3. A Zigbee mesh network can reliably cover a larger area than a WiFi network

Things are changing though.  A small number of capable companies like Atmel have developed or acquired WiFi technology that is low cost and robust.  Expect to see WiFi chips that cost well under $4 and can operate at 105 or 125°C.  The problem with Zigbee (as well as Home Plug and other network technologies that are non-ubiquitous) is that you can’t just bring a Zigbee device home and turn it on.  You also have to purchase a gateway that connects the Zigbee device to your home Internet connection.  Most often, this gateway connects Zigbee and WiFi.  While there are system level issues to be dealt with having two radios operating within the same unit on the same frequency band, these have been solved.  But the fact remains that the gateway adds cost and complexity to the setup.  Consumers would rather just connect their things directly to their home WiFi access point.

Utility companies did not have this challenge back in 2009 because they chose to run a closed network.  If you wanted a smart thermostat or in-home display that communicated with your electric meter, chances are you had to purchase it from your utility company and it came pre-loaded with the certificate needed to participate in the home area network.  As you can imagine, customers have not been given this closed network idea a warm reception.  With the recent proliferation of smart phones and tablet computers, customers want to manage their home energy usage on those devices and not some utility provided in-home display with limited functionality.  But the problem is that none of the smart phones or tablets have a Zigbee radio and the utility company does not allow any other devices to access the home area network.  Therefore we are left with a gap between what the utility companies rolled out in 2009 and what their customers want to have.  My prediction is that this has led to a lack of sales of in-h0me devices and adoption of the interactive demand response products promoted by the utility companies.  In many cases, the utility companies have a nice 2-way communications path to the smart meter where a dormant Zigbee home area network radio sits.

Looking towards the future, the technology exists to have WiFi and even Bluetooth radios in each and every electricity meter.  These new WiFi controllers are low cost and robust.  And they even include TCP/IP stacks for easy integration with existing meter microcontrollers that may not have the spare memory and processing power to run a network stack.  Much like the SoC devices dominated the Zigbee deployment because they could function as a Zigbee network co-processor, these new WiFi devices can offload most or all of the network processing and retrofit existing meters and designs easily.

Questions still remain about how the utility companies will roll out WiFi and completely interoperable meters.  After all, the system is no longer closed and any node on the network can talk to the electricity meter.  There will undoubtedly be some kind of split between what is available to anyone who asks for it, and protected configuration and billing data.  The TCP/IP stack must include TLS security and root keys will need low cost high security protection such as what the Atmel ATECC108 asymmetric crypto authentication device provides.  These nodes will also require protection against unauthorized firmware upgrades and physical attacks.  Authenticated access to utility company only information will also have to be planned for.  It is likely that the meter data collection will still be done via the new IEEE 802.15.4g NAN standard, but a TCP/IP based stack will also be run.  A merger of technologies to put IEEE 802.15.4g and WiFi (IEEE 802.11a/b/g/n/ah) on the same device will further reduce costs.  The electricity meter will start to look like a computer with one network stack running over multiple network interfaces but with separate access control.

Sometimes I remind myself that when I was in engineering college at NC State University back in the late 1980’s we got new computer systems for all computer science and engineering labs.  These systems ran a new 32-bit MIPS processor at a lightning fast 12.5 MHz.  And on those systems we could run multiple processes in different windows and we could connect to computer systems across the globe through the 10 Mbps Ethernet connections to the campus network which was routed into the Internet as we knew it back then in the pre world-wide-web days.  An Atmel SAMD20 Xplained Pro evaluation board runs at 4x the clock speed of those DEC workstations back in the 1980’s.  And I had ran Ethernet drivers and a UDP stack on 10 MHz Intel 8-bit processors at my first real firmware job in the 1990’s at Exide Electronics with only 256KB of program memory and 32KB Data SRAM.  So a small MCU based on the ARM Cortex-M0+ core like the Atmel ATSAMD20 or ATSAMD21 makes an ideal host processor for an electricity meter and can easily support the Atmel WINC1500 WiFi Internet Link Controller for connectivity.  For even better integration of meter functions, the Atmel ATSAM4CM offers dual ARM Cortex-M4 cores with memory protection unit and an integrated analog front end with either 4 or 7 channels for measuring and computing energy consumption or production.  Add the Atmel AT86RF215 dual-band radio for Zigbee/802.15.4 and the 802.15.4g NAN along with the WILC1500 WiFi controller and now you have the smartest meter money can buy at a BOM cost much less than typical smart meter solutions did in 2009.

With WiFi and new connectivity choices being driven by the Internet of Things and consumer choice, your next electricity meter may well come equipped very differently. There are decisions to be made and problems to be worked through.  Good engineers will make it happen.  So when your NEST Thermostat running Google’s Thread stack can work seamlessly with your electric meter and also connect to your smart phone, you can thank the suppliers who have brought this new generation of smart solutions for energy measurement, data processing, connectivity, and security like Atmel.  Contact me or your local Atmel Representative or sales office for more details. ( http://www.atmel.com/about/contact/distributors/default.aspx?contactType=Sales%20Representative )  We would love to help!

Posted in Uncategorized | Leave a comment

Look Mom I can shrink Arduino even more! And it cost only EUR 4.00!

Nice. Small is great! Could use a few of these for my daughter’s Halloween skating props.



On October 11 I blogged for our OLIMEXINO-NANO which is already in production and will be ready for sale on November 1st 🙂

With dimensions only 30×30 mm NANO is really cool, but we wanted to see if we can go smaller 😉

As you know on November 2 and 3rd there is OpenFest and we are partnering with the organizers and will make hardware hackathon where everyone who want to learn soldering to may build his own Arduino which to take with him. For this purpose we had to make some small but easy to solder board mostly with PTH components which are easy to solder for beginners.

This is how we made OLIMEXINO-85 you see above on the picture it’s only 30×20 mm and it’s made to make easy breadboarding:


For OpenFest we are preparing 100 kits of OLIMEXINO-85 which will contain all components, so everyone who want…

View original post 238 more words

Posted in Uncategorized | Leave a comment

Retro AVR Demo from 1997…On STK600

Running the 90S1200 demo on a modern STK600 to reverse engineer and validate the pin mappings. The original demo was on a little coin cell powered board and was carried by every Atmel salesperson in 1997 when the AVR was introduced. The 4 LEDs represented the relative speed of the AVR versus the popular competitors, some of which are still around today. At that time, very few people had seen the level of integration and performance in the little 20 pin AT90S1200. It had its own RC oscillator, EEPROM, In-System Programmable Flash, and I/O that could drive 20 mA. And it could sleep with under 1 Microamp of current draw and wake up in a few microseconds from an external interrupt (as implemented in the demo)

Next step will be re-coding the demo on a new Tiny4313 AVR using Atmel Studio 6.1 and C so that I can use the little demo board for a new project.

Video | Posted on by | Tagged , , , , | Leave a comment

Using a Digital Temperature Sensor for Long Leaded Thermistor Replacement

Plain old NTC (negative temperature coefficient) thermistors have been around for a very long time.  Combined with a suitable resistor network and calibration, they can provide accurate temperature measurement and indication of relative temperature changes.   Today, there are small chip component NTCs that mount on a PC board and provide a way to measure the temperature on the board.  That is good when the PC board is what you are interested in measuring such as a small board encapsulated inside a smart battery pack or other enclosure.  But what happens when you want to measure ambient air temperature or some other fluid separate from the PC board?

Leaded thermistors have found applications where flowing air above the PC board needs to be measured.  Thermistor suppliers offer devices with long leads that allow the components to be thru-hole mounted with the raised body located inches above the board for measuring flowing air temperature.  This is especially desirable in enclosures with forced fan cooling such as network and telecom equipment and computers.  By placing a thermistor off the PC board in the stream of incoming and/or outgoing air, there is the ability to determine quickly whether the equipment is overheating and the effectiveness of the forced air cooling.  Failures such as a broken fan, restricted filter, or blocked inlet or outlet can be detected and action taken before any equipment damage occurs.

What if we wanted to perform this same function using a digital temperature sensor with I2C serial interface like the Atmel AT30TSE750?  The advantages of a digital temperature sensor are several and include:

  1. No Calibration Required
  2. Interface to a Processor Without A/D
  3. Lower Power Consumption
  4. Immune to Noisy Analog Measurement

So a digital temperature sensor sounds like a great idea, right?  The next problem is that the  digital sensor is in a small IC package that mounts on the surface of the PC Board.  Atmel offers the AT30TS family of devices in SOIC and WSON 8 pad packages.  These are great if you want to know the PC board temperature, but limiting if you must know the temperature of the air flowing above the board.

My thoughts on a proposed solution to this problem is a low thermal mass PC board that can mount the sensor any desired distance from the main PC board.  Using thin (0.032″ or less) FR4 material and having no embedded copper plains running the length of the board means minimal thermal coupling between the main board and the thermally isolated sensor.  Further steps can be taken to minimize thermal coupling and impedance to airflow by voiding areas of the FR4 and also by making the board as physically small as possible.Image


A PC Board shaped somewhat like the picture above with a right angle header soldered to the holes in the bottom can become a replacement for a long leaded thermistor in almost any application that includes a processor.  Making an intelligent temperature sensing node with an Atmel AT30TS or AT30TSE temperature sensor and an Atmel MCU that can measure ambient temperature is easily realized using this creative PC board approach.  And since the PC board is long and narrow, it is possible to yield a high number of boards per panel, lowering the per board cost and making this a feasible solution.  Eliminating the factory calibration required with traditional thermistors and the risk of inaccuracy due to uncalibrated sensors makes the Atmel AT30TS sensor an attractive option for thermal management in high availability systems.

Posted in Uncategorized | Tagged , , , , , | 1 Comment