Smart Meters

By Jim Sorensen

Jim Sorensen shares his insights on smart meters used for utility power measurement.

Sometime ago I promised to look into smart meters for utility power measurement and I wanted to share what I learned.  This sermon, while not inspirational, is at least marginally factual and it points to a couple of open doors both with respect to technology and politics, my two favorite subjects!

I had originally planned to impose a somewhat more technical study on you including derived results from whatever I could glean from power utilities and meter manufacturers on the assumption that not much work had been done in the RF radiation area or in the area of possible exposure to humans and human medical devices by these instruments.  Both of these were incorrect assumptions.  Smart meters are in reality rather unexciting, simple technology that uses, rather re-uses, established communications protocols and hardware in a relatively complex “network” for the purpose of collecting data for electric bills, in this iteration.  Future iterations might well include such things as the meters ability to “talk” to your home computer with recommendations on utility use and even with the possibility that an addressable meter might be able to literally shut off power to individual appliances or your home for things like non-payment of bills or for chronic over-use of power.  It’s interesting to note that some sort of socio-political decision as to what constitutes “over use” would be is obvious.

To be fair, there is already in place an FPL program that controls power to things like Air Conditioning units and water heaters for which customers are paid a small fee.

Getting right at Prof. Neil Muncy’s field problem, typical Smart meters produce intentional and spurious signals in both the unlicensed 900 MHz and the 2.0 GHz frequency bands.  Both channels use power levels of less than one (1.0w) watt and both produce intentional signals that can be statistically safe for human exposure but significant where other, more sensitive, technologies might be concerned if they are near the meters…depending obviously upon what those things are and what you’re using to measure things.  If the meters are properly designed and accepted by the FCC, both produce unintentional signals which are about 60 dB below the intentional carrier, minimum.  Both are spread spectrum or frequency hopping signals and might interfere with other links you might set up for measuring places where smart meters are in use.  The 2.0 GHz output is part of the Blue-Tooth spectrum.  In the case of my home office, I use a blue-tooth network for scanners and printers so I’ll need to pay attention.  FYI:  My field gear is all wired.

When engaged in field measurements, being aware of the ground one is on and being equally prepared is half the battle.

Smart meters have been in the news as some sort of salvation of power costs and with reference to what’s seen as a declining capability by the power industry to keep up with demand.  The current administration has discussed being able to monitor individual usage and to control it somehow.  Smart metering with the ability to selectively shut off power is the obvious goal.

I had a chance to talk “smart meters” with Silver Springs Networks and I-Tron, companies that build both the software and hardware for just about all of the Smart Meters in use in the US and Canada both of whom sent copious data and told me:

1)  Yes, the smart meters make RF noise both intentionally and unintentionally in both the 900 MHz and the 2.0 GHz unlicensed bands.  This is more of an FCC definition…radio stations are “intentional” radiators and computers or coffee pots with micro-p’s aren’t, for example.  The typical smart meter uses an on-board card and that on-board card does things like keep track of your bill.  In one iteration, this is “tickle-and-spray” transponder technology in which the meter reader drives by and his gear outputs a small signal which tickles your card and gets it to spray whatever data it’s collected over the time since last tickled (TSLT?) to the truck for “meter reading” purposes.  There is a slightly different technology in which your meter outputs on a timed basis a burst of consumption data to a centrally located neighborhood receiver that collects the data from a group of meters and then ships it back to the main office either by DSL line or an RF Link.  If DSL there is no related radiation problem and if by RF link, my guess is that the link would be low power and highly directional, well above (height-wise) the surrounding terrain.

This is significant from an RFR perspective in that the duty-cycle, or the time switched on, of a radiator is significant.  In the case of the “tickle-spray” technology, it’s only on when tickled and then only for a short period of time…presumed to be long enough for the data to get through.  Given this, the corrected exposure time for humans within the critical distance is on the order of a year for a noticeable dose to occur.  As you can see, a small wattage transmitter, an isotropically inefficient antenna, and a high operating frequency coupled with a very short duty cycle provides significant “safety” from the non-ionizing radiation transmitted.  I guess I should note that the word “safety” should not be misunderstood.  These are all cases of compliance with FCC, EPA, and OSHA regulations and have nothing whatsoever to do with any possibility of real human damage.  The standards set as far back as 1988 for these kinds of non-ionizing radiation are intentionally set to be on the order of 6% (informed workers) and 1% (uninformed public) of the actual calculated maximum “safe” dose of this kind of radiation and the exposure time windows scaled similarly.  There is no way for anyone to know how any given individual will react to any of this kind of exposure but to date there have been no reported cases of permanent damage to a human from exposure to this kind of “heating” radiation and there seems to be little statistical evidence of these data being inaccurate.

2)  One pretty obvious question is why not use the power lines themselves to haul the data rather than an RF system?  I have not tested them but am told by all concerned that power-line connected networks are too slow for the number of nodes and for the quantity of data collected.   I’m told that there is a study underway to decide if it’s cheaper to build this kind of network using intense data compression as opposed to a “drive-by” system.

3)  Both the FCC and the State of California have done tests and the results show that the on-channel and spurious emissions are well within FCC guidelines for non-ionizing RF radiation.  The submitted FCC approval from I-Tron shows these levels to be very small fractions of the allowable signal strength, which is what I would predict from reversing out the technology probably in use and deriving levels from that model.  This was a concern voiced by several people I have chatted this with, several of whom use pacemakers and one who has an on-board defibrillator buried in his gut-works somewhere.  His concern is that he’ll find himself in places where higher levels of RF might be present with no warning.  Who’d think on the other side of the wall from an electrical meter?  Who’d necessarily know?   I suppose that a lobby sign of “Smart Meter in Use” might do the trick but there seems to be no real risk…at least with the present iterations of the technology.

Sadly, there is no firm signal reference for RF interference diddling the technology used in typical personal medical electronic devices so it’s difficult to conclude what’s safe and what’s not.  So far all the meter testing done seems to indicate that signal level and interference levels are well below the FCC guidelines for hazards but that’s for “heating hazard” signals impacting the entire human body with a mass equal to a human sized “water bag” not for micro-technology in active circuits.

5)  According to my source, the intentional emissions of the units fall into the general guideline of the FCC for such devices operating in the frequency bands outlined.  Probably the easiest way to follow up on this is to look in the FCC website for OST bulletins discussing non-ionizing RFR compliance. Rather than parse it out here, I’ll simply tell you that there are bulletins, rule-making, and white papers available which discuss the acceptable levels of non-ionizing RF radiation allowed and both the FCC and California tests show that a typical smart meter outputs both intentional and unintentional signals that are small fractions of a percent of the allowable limits at distances that are much closer than you and I would typically stand from them.

In the case of the California tests, a meter array with something like 20 meters was set up and measured which again showed that there was no consistently coherent and additive effect.  While signals can indeed “add” through superpositioning of the waveforms, it’s not likely that given the time differentials across the array and the time differences within the various transmitters in the array that any sort of significant additive effect would be found.  It’s important to note that superpositioned signals can appear to be minimal depending upon how the measurement is taken and depending upon what’s used for making the measurement.  Diode detectors, for example, have been shown to lack accuracy in the presence of identical signals that are both phase and magnitude coherent at the frequencies under consideration.  In the case of the California measurements, it seems to be have been determined that while diode detectors were used, the probes were hand-held so the distances and movements of the pick-up probes was inconsistent and the probability of identical spacing and time span for measuring was almost impossibly random.  Couple in the fact that the actual absolute frequency of the generators is equally random within the assigned frequency span and the frequency transitional shifts even more so, again, the possibility of signal alignment to produce a “directional gain” in the antenna group is statistically almost impossible to achieve much less to predict with any accuracy.

When you couple in the duty-cycle of the transmissions, you end up with exposure-time curves in days and weeks rather than in minutes and seconds.  Since the effect of this kind of non-ionizing radiation is essentially tissue heating and not ionizing destruction of tissue and cells, the cooling which follows exposure tends to quickly normalize the changes induced.

6)  For anyone interested most smart meters use an inverted-F antenna (on the PC board) mostly to take the box the meter sits in into account.  A typical inverted-F antenna, in off the shelf packaging, has a measured gain of about 3.3 dB and shows a cardioid radiation pattern. The fact that most of these makers are using the low 900 MHZ band tells me that they are using a purchased quarter-watt watt transmitter module using the antenna gain to result in an output around a watt rather than something they designed and built.  The whole affair is probably a nice re-packing of existing technology designed to collect data and spray it around for the meter reader to snag.

These answers are not sufficient to form any kind of legitimately informed opinion but they do point to what most of us had guessed which is that the smart meter program is tacked on technology tested within very narrow windows rather than something new.  That said it also appears that consideration of the possibility of harm to humans either through direct “heating” (non-ionizing) exposure or from impacting implanted or adjacent medical appliances was considered and has been managed at least within the guideline set by the various agencies governing on both the State and Federal levels.

The one area I did not get information about…not for lack of asking but for lack of a response as yet…is the area of whether a public utility company can ask for the State to grant it immunity from any harm in a medical sense it might do through interference with a pace-maker or defibrillator.  I have discussed this with attorneys for the PUC in Florida who told me that they are discussing the question with the Florida Attorney General and will reply in due course.  Perhaps the famous “McDonald’s Signs” of “Microwave Oven in Use” will give us guidance, at least in the sense of a commercial enterprise.  While not a utility (one only can wonder why) McDonald’s is ubiquitous where fast food is concerned and most Americans are in a McDonalds at least once a month…or so they tell us.  I could not find a statistic as to the number of deaths caused by exposure to the incidental radiation from McDonald’s ovens stopping pacemakers but I imagine that it happened at least once, thus the signs.

Clearly smart meters are not microwave ovens and while at least some of the radiation from a smart meter is in very similar frequency bands the signal levels between a 1000 watt oven and a less than 1 watt smart meter should be obvious.  Even so, and understanding that a typical residential house is not a McDonald’s restaurant so it’s hard to correlate the two, but the issue of unintentional non-ionizing RF radiation causing a disruption to the normal operation of a personal medical device is very real and must be considered.

I regret that I have no data with which to work other than that published by both the FCC and California and while I don’t doubt that data there remain questions in my mind but since the present state of the utility metering art seems to be limited to collection of customer consumption data and spraying it on request to the meter readers, whether or not in metering trucks, I can’t see a lot of danger from having this particular camels nose under the tent.  It’s knowing that it’s there that’s important at this point.

I’m adding “smart meters” as a question about the premises I’m measuring to my field inquiry questions list.  It might prevent having to go back for free to collect non-anomalous data.  If you use technology in either the 900 MHz or the 2 GHz “unlicensed” bands you might need to carefully select frequencies in order to get your gear to work or to get nice clean data.  You might perform a scan rather than risking accidental “crashing” of similarly tuned transmitters.  Amusingly, frequency hopping absolutely assures crashing but opens the statistical time-line on exactly when they’ll crash remarkably!