Remote Meter Reader Data Transmitter Power Supply

A number of companies are developing products that will allow utility providers (water, gas, and electric) to read the customers' meters remotely. Data is transmitted from the meter to a nearby data collection unit (DCU) via an RF link. The gas and water meters are typically powered by a small lithium battery. The battery is supposed to last for 20 years. This is difficult to achieve with the typical power-supply topology.

That RF data link transmits usage data at intervals, which could be once every 20 minutes, or once every two hours, depending on what the utility needs to know. The current draw for one of these RF modules powered from 3.6V might be a few hundred mA — but of course it is just a short burst (very short duty cycle), so overall power draw is low enough that the cell should last many years. But it probably won't.

These small lithium cells, like all cells and batteries, have A-hr (ampere-hour) or mA-hr ratings that indicate that (if they were perfect devices) you could draw a certain current for a specific number of hours, at which point, the cell would be completely depleted. A 1A-hr cell could supply 1A for an hour; or 500mA for 2 hours; or 5A for 12 minutes.

With the very small cells, besides the mA-hr rating, you need to be aware of their peak current draw — and not exceed the manufacturer's specs. If you do, you will shorten the cell's life appreciably.

Here is the way to work around this. Use that small cell to power a high-efficiency boost switcher. You can reasonably expect to boost from 3V to 15V with efficiency above 90 percent. Charge a very large capacitor through a resistor to 15V. That should keep the input current to the boost switcher quite low. Use that charged capacitor as the DC source to power a high efficiency buck regulator. Set up the buck switcher to provide 3.6V. You'll get 3.6V for as long as you need while the big capacitor runs down. And the lithium will never be taxed with high current draw.

To improve overall performance, we can add a very low power microcontroller (MCU) with a built in real time clock (RTC). The RTC can wake up the MCU at the appropriate time. The MCU can enable the boost switcher for the right amount of time. Then it can enable the buck switcher and the transmitter module and send the data. And then everyone goes back to sleep.

The whole thing should look like this:

It should be possible to integrate everything except the power cell, the two inductors, the big capacitor, and the RF module onto an IC.

Have you ever designed and built a power supply like this one?

14 comments on “Remote Meter Reader Data Transmitter Power Supply

  1. Scott Elder
    July 19, 2013

    Hey Brad – What about using water pressure as a carrier for communicating information rather than RF?  The water company can modulate the water pressure remotely (at a random hopping frequency carrier) and the local meters that are addressed respond back with local pressure modulation (faucet valve on/off).  

    Its not like the data rate needs to be Mbps.  sub 1bit per second would be fine.  1000 homes x 128 bits of data /1 month.  That's like communicating to Voyager I.



  2. Brad_Albing
    July 19, 2013

    Hmm… never thought of that. Wonder if anyone ever looked into it.

  3. Netcrawl
    July 20, 2013

    @thanks Brad that was great! I remember a Popular Electronic magazine issue, which featured a project implemented with discrete transistors for doing this kind of communication by regarding water as a sheet resistance, its totally an amazing stuff to see, and quite interesting.   


  4. Davidled
    July 20, 2013

    Microchip developed the demo board for smart grid technology with lower system cost and long battery life.  Microchip presents initial outline for gas, water and heat flow meters in the application level.

    The web link is below:

  5. Brad_Albing
    July 21, 2013

    @Netcrawl – Hmm… regarding water as a sheet resistance – that sounds like a variation on transmitting signals (baseband, not RF) thru the ground. Looks like another blog topic. Stay tuned for that….

  6. Brad_Albing
    July 21, 2013

    @DaeJ – interesting Smart Grid part. thanks.

  7. eafpres
    July 22, 2013

    Hi Brad–out here a lot of electric meters got cellular modems.  There is plenty of power to run them in that case, of course.  In the water case, a lot of homes have the meter inside, and an electric powered readout outside.  Again, plenty of power to run them.  Interestingly, Denver Water used to have a large number of un-metered customers.  I lived in a house in the 70's that had no water meter, and was charged a flat rate.  Nowe, they have moved ahead with meters and AMR.  On the site they claim that with wireless transmitters on the meters, a few trucks driving around can read more meters per day than a previous staff of 33 persons.  From the utility point of view, the initial savings are the reductions in labor costs (i.e. fewer jobs).

  8. Brad_Albing
    July 22, 2013

    @eafpres – Re the electric meters, yep, those are the easy ones. Altho', some have a battery just for backup. That way, they can report power failures which also comes in handy.

    And for the gas and water, as you noted, they either drive around and interrogate the meter from the street or use a transmitter + a DCU mounted on a nearby utility pole to grab data from a neighborhood and send it out via landline or cell-com link to the head-office.

  9. Dirceu
    July 22, 2013


       a good idea, particularly if you already have the boost and buck modules already available. But a potential problem, I think, is the time to charge that big capacitor – Might not be suitable for some applications. Also, what's the gain in efficiency when compared to using a single SEPIC based buck-boost converter, if the intention is to get the most of the battery (for ex., the a Li-ion ranging from 2.5 V to 4.2 V with a 3.6 V fixed output voltage)?

  10. Brad_Albing
    July 22, 2013

    Well, you actually don 't want a SEPIC. The intent is not to create a power supply that can make (in my example) a 3.6V supply from a battery voltage that might be above or below the output. Instead, the intent is to step up the battery/cell voltage to something several times bigger while drawing hardly any current from the cell – then charge a big capacitor. You've got plenty of time to charge the cap, so that's not an issue. That's actually the intent – slowly accumulate a high voltage (well, 15V in my example) that is capable of supplying plenty of current – altho', of course, just for a short period. But that's OK too – that's all you need to transmit the data. Then start over.

  11. Dirceu
    July 23, 2013


       thanks for the details. Maybe I did that comment because, in your drawing, I saw no load connection to the high voltage (15 V).

  12. Brad_Albing
    July 23, 2013

    @Dirceu – understandable. The only load is simply the step-down (buck) switching regulator.

  13. MLM-TS
    July 26, 2013

    I think you forgot that half the energy required to charge the large cap will be dissipated in the series resistor during charging.  Write the equations, do the integration – (1/2)*C*V^2 on the cap, (1/2)*C*V^2 lost in the resistor.

  14. Brad_Albing
    July 26, 2013

    In my simplified version, quite so. But a version could be done that made use of a contant current mode in the boost switcher – if it were sufficiently sophisticated. That'll be a topic for another time.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.