5-10 years battery life of GPRS Water Meter, How?

[Ichan]

A GSM/GPRS Module in idle state will steal about 1mA or more, a hall effect water flow sensor also will steal about that range - just two of it will took at least 2mA.

Let say a D Size Lithium Thionyl Chloride battery used, 3.6V/19000mAH (yes, 19 Ampere-Hour) - at 2mA discharge rate the battery will last for 19000/2 = 9500 hours = 396 days = 13 months = just over a year.

Here they claim theirs: The internal lithium batteries will last for 5 to 10 years!

How they do it? ANY idea appreciated.

-ichan

[Old_but_Alive]

the flow of the water vane probably generates electricity.

maybe not, as it says it connects to an existing sensor.

TI and Linear tech seem to be heavily into generating electricity from vibration etc

[Top]

For 15 minute readings and daily uploads the batteries will last for 5 years.

The 5 years is based on counting one pulse every 15 minutes and powering up the GSM/GPRS module once a day.

[alexisnik]

Exactly, stand-by with RTC wakeup can be under 1-2 uA (RTCs can draw nA). They schedule the next wake-up, put the microprocessor to deep sleep (or even power-off). When the next measurement is due, they wake the processor and they log the value. With large Li-Ion batteries and a good sensor, this can occur every second with very low energy. The GPRS modem wakes up only several hours (or days) and they send all the gathered data.

[Parmin]

I doubt the GPRS module are placed on IDLE when not used, after all if they are not used to poll incoming signal so there are no point to keep the module up.
I think there should be a cut off switch to totally cut off the power to the module and turn it on only when required.

[Ichan]

For 15 minute readings and daily uploads the batteries will last for 5 years.

I missed that words.

So the GPRS module only POWERED once a day, not being put on IDLE nor OFF state - one question answered.

Anyway what GPRS module with the lowest power profile? I found LEON G100 is low enough, 90uA OFF, 0.99mA IDLE, 410mA GPRS - any other better?

How about the sensor? How can they read the sensor once every 15 minutes? All water flow sensor that i knew has PPS (pulse per-second) output, eg. 10 PPS = 2 L/minute. The mcu can be put into sleep and waked-up by this pulse, but the sensor itself (hall effect) have some quiescent current - it will took some mA all the time.

They words about compatible water meter is a big question mark.

I read some about energy harvesting things, had a though to use low light photovoltaic energy source and BQ25504 looks very interesting. One doubt with it is the output current capability (for GPRS module), all docs tells that the load should be connected to the VStore not VBat - and there is no data how much maximum current can be drawn from it. Anyone?

Edit: Whew, i just realize this is my 500th posts - should i celebrate it? 

-ichan

[solutions]

Heroes don't celebrate, they just humbly do good deeds. Congrats, Ichan. Got ya by exactly 2x with this posting, though 

In any case, TI has some really good app notes on water meter reading - I think with some talking about battery life:

http://www.ti.com/solution/water_meter

[Ichan]

It must be a God above Hero  , congratulation to you too Solutions.

I just changed my Avatar, not to celebrate anything but just as a milestone.

Thanks for the link, i will dig it more later - now focusing on the sensor.

-ichan

[alexisnik]

I doubt the GPRS module are placed on IDLE when not used, after all if they are not used to poll incoming signal so there are no point to keep the module up.
I think there should be a cut off switch to totally cut off the power to the module and turn it on only when required.

GPRS power is completely cut off during sleep of the module, except for when it is actually used to communicate. You can do it using the ENABLE pin on it's power supply or a MOSFET. This is the standard technique.

I missed that words.

So the GPRS module only POWERED once a day, not being put on IDLE nor OFF state - one question answered.

Anyway what GPRS module with the lowest power profile? I found LEON G100 is low enough, 90uA OFF, 0.99mA IDLE, 410mA GPRS - any other better?

How about the sensor? How can they read the sensor once every 15 minutes? All water flow sensor that i knew has PPS (pulse per-second) output, eg. 10 PPS = 2 L/minute. The mcu can be put into sleep and waked-up by this pulse, but the sensor itself (hall effect) have some quiescent current - it will took some mA all the time.

They words about compatible water meter is a big question mark.

I read some about energy harvesting things, had a though to use low light photovoltaic energy source and BQ25504 looks very interesting. One doubt with it is the output current capability (for GPRS module), all docs tells that the load should be connected to the VStore not VBat - and there is no data how much maximum current can be drawn from it. Anyone?

Edit: Whew, i just realize this is my 500th posts - should i celebrate it? 

-ichan

Sensors can generally draw much lower than mA when they are in low-power mode. I don't know about water flow sensors, but for example new accelerometers, when they are in low power mode, they can generate a Free Fall interrupt (or activity interrupt), only by using <10 uA. (Yes, micro-amps).

About the GPRS modem, the data you wrote are the average consumption. For example, GPRS modems during network registration, have a peak consumption of about 1.7 to 2.5 AMPS! Due to this, you have to connect them to large capacitors (or even better super caps), and you need to have burst capabilities at your power supply.

Regarding the BQ25504, I think that you should add a LiPo charger IC at its output, since LiPo batteries have very low discharge. Power transferring in energy harvesting applications is tricky in general. Regarding VSTOR, I believe it is intended to be used with very small loads. For a solar panel solution, make sure you use the MPPC function on the chip.

- Alex

[Ichan]

Yep, the gprs module should be turned off by cutting it's power supply - if turned off by enable pin then it will still draw several hundreds uA or even some mA.

I am still struggling with the sensor. Mostly low cost (important) water flow sensor is paddle / turbine mechanism with some magnet on the rotor - this magnet rotation is sensed by hall effect sensor on the outside of the sensor unit, sample of this kind of sensor attached.

The problem is (standard) hall effect sensor ic draws some mA in use, and in this case it can not be turned off as it has to count the water flow all the time. I had a though to replace the standard hall effect ic with the micropower type - but then found they are too slow for this application, in about 100ms period range - ie. about 10 pulse per-second. I need it to measure up to 1000 pulse per-second.

The compatible water meter and 15 minutes measurement period on the reference unit is very intriguing, i guess this become similar with "remote reading" discussed HERE. Another attachment is mechanism of common water meter, extracted from Popular Science Magazine July 1950 - interesting old mechanism.

The choice between primary or secondary battery usage still under consideration, this is between Li-SO4 vs Li-Ion/LiPo - the sensor issue need to be solved first.

How large and what kind of cap near the gprs module need to be? Any clue on how to calculate?

-ichan

[Top]

They don't claim to be using a hall-effect sensor. They are interfacing to an existing meter; the pulse output from the meter is a mechanical contact like a relay contact with one pulse per litre (or per 10, 100 or 1000 litres depending on model). I presume what they mean by '15 minute readings' is that the consumer is on average using a litre every 15 minutes so your micro will only be powered up for several mS every hour and your GPRS for several seconds a day; even a small battery would last a very long time.

See: http://www.bmeters.com/Pdf/Products/GSD5-R.pdf

[Ichan]

Aha, so the key word should be "water meter with pulse output", thanks for the info.

Looks like it use reed switch for the pulse contact, is that reliable for long time? I need to find it out.

The next problem seen is if the local water company i deal with willing to change their existing water meter (which has no pulse output).

Top, i knew you are experienced with gprs module. A question, how to use a gprs module with limited current capacity power supply (let say 500mA max)?

-ichan

[alexisnik]

A question, how to use a gprs module with limited current capacity power supply (let say 500mA max)?

-ichan

Hi, as I wrote in my previous post, the best way is to use supercapacitors...

[Ichan]

What supercap? The one usually used for backup (rtc, ram, etc) has too high internal resistance.

What value? Larger is better, sure - but not the answer i seek for.

-ichan

[Top]

This application note has information about sizing the reservoir capacitor for a GPRS system.

http://www.fairchildsemi.com/an/AN/AN-8019.pdf

[Ichan]

Hi Top, i will buy you a drink if you are near me - thank you.

Now i tend to use primary battery, non-rechargeable LiSOCl2 - high energy density, very low self discharge, but low discharge current.

That is the reason. The app note is about usb modem, just realize it is similar situation 500mA max - so i dredge up my old stuff, here it is: Falcon Samba75 USB Modem.

The module is Siemens MC75 (now Cinterion), on one side there are four 470uF tantalum capacitor, and on the other side there is Vishay FX5545G201 buck converter and six 22uF tantalum cap near to it. Again, the answer is obtained, only need to find a low power buck/boost converter to provide stable supply for the gprs module which can be shut down.

Note where it sits on the photo - it is 2x Maxwell 350 Farad 2.7V Ultracap, to play with the BQ some other time  .

Thank's everyone, for everything.

-ichan

[alexisnik]

What supercap? The one usually used for backup (rtc, ram, etc) has too high internal resistance.

What value? Larger is better, sure - but not the answer i seek for.

-ichan

I have designed a GPRS system for ultra low power using supercaps. Check out this:

https://www.google.gr/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1&cad=rja&ved=0CDUQFjAA&url=http%3A%2F%2Fwww.cap-xx.com%2Fresources%2Fdocs%2Fppr_2003_GPRS-perf.pdf&ei=i-v2UNSFPKaZ0QXW24HoDw&usg=AFQjCNG_-bWFft932nj5sfkUotzatO8wWQ&bvm=bv.41018144,d.d2k

and this:

http://www.element14.com/community/docs/DOC-24910/l/why-you-should-use-ultra-low-esr-super-capacitors-in-gsmgprs-applications

(google search for supercapacitors gprs)

I had used ~0.5 Farad @ 4.7V, because Cap-XX had a minimum quantity in their order... The calculated value for 2.5A pulses was about 0.8 Farad so I used 2. The system's total power consumption improved greatly compared to standard capacitors.

[Ichan]

Hi Alexisnik,

I do considering the use of ultracap, but seems the low esr one is too costly for production quantity - other cap taken into consideration is solid polymer capacitor.

Thanks for the link, there is an interesting comment at the bottom of the second link, quoted here:

Skippydownunder Feb 20, 2012 11:34 PM
Thanks for your post above, I am having difficultly with a board which has an Telit module gc864, it keeps intermintently dropping using a single Lithium Thionyl Chloride saft battery lsh20. It seems it cannot provide the 2A peeks. I have an LT1308ACS8#PBF.  buck boost to increase the voltage from 3.6 to about 4 volts and have tried the caps after the buck boost and before the gsm module but the problem persists. I am wondering if you have any reference designs using a Lithium Thionyl Chloride saft battery which you could share.
Thanks and Regards Mark

This case is very similar with mine, i have to be careful with the design - have to do some experiment before.

-ichan

[Faros]

@Ichan

I have done some battery based Micro Power Designs myself and I have found that saving power while saving cost of component/manufacturing is a challenge that consume time/effort surpassing the principal application (in case of several K's Qty where every cent is multiplied by a 3 digit factor).
I use Microchip MCU's and the MCP series power management components just because I am geared for it and also for availability reasons. their XLP series is great when it comes to power consumption.
Recommended readings:  
Low-Power Design Guide
http://ww1.microchip.com/downloads/en/AppNotes/01416a.pdf
Battery Power Applications Design Guide - Microchip
ww1.microchip.com/downloads/en/DeviceDoc/39610d.pdf

Beside considering power saving techniques provided above; still more saving can be achieved, for example I have found that using 3V fast (2-6 msec) latch relays useful in continual 24/7 operated circuits, the D size energizer (or any other brand) E95 batteries (Zinc-Manganese Dioxide) are capable of providing 19'000 mA/hour at 25mA discharge rate with the advantage of 7 years shelf life(@25degrees), availability and above all cost. equivalent lithium is half sized and 8 times the cost. Microchip has MCP's that are dedicated to such batteries and will suck the battery dry down to 0.6V.
also, using P-channel logic level MOSFET with 2.7V Vg and Rdson < 3 Ohm are wreath considering, those can switch your peripheral devices efficiently while keeping cost, size and complexity acceptable provided that you deal with transient / latch-up .. etc in your design. I have been happily using them and they work like magic.

Once we agree on those points we can move to the sensor ... -:)  
  

[Ichan]

Hi Faros,

I can see experiences on your words, fully agree with saving cost on low power design is challenging plus extra effort for gprs things.

I choose PIC18F47J53 for this as the device has luxurious spec to put LCD of any type and wired comm channel (USB here), while i myself think those are not required for a remote metering device.

The latching relay thing is new to me, but on this case pfet is still the choice as the high power duration is very short - i usually use Alpha Omega product, good value for money IMO.

I am still in dilemma with the battery, mostly because the gprs thing:
1. Rechargeable Li-Ion is the highest cost (battery + charger + protection), i will try to avoid this.
2. Alkaline is very cheap (1/10 of LiSOCl2) but 1.5V and steep discharge voltage, will need boost converter maybe two - one for mcu + peripheral and the other for gprs.
3. LiSOCl2 is 3.6V and flat discharge voltage but the battery is expensive. I am thinking not to use ANY voltage regulator by using this battery.

Any opinion about #3? Do we need voltage regulator on battery operated low power device?

Other thing, anyone know what gprs module with SSL support? Telit?

-ichan

[solutions]

How about a small NiCAD stack that you trickle charge off the "main" battery just before you light the candle on your GSM/GPRS module (I don't understand why you are using this radio - there are lower power radios specific to meter reading)?

[alexisnik]

How about a small NiCAD stack that you trickle charge off the "main" battery just before you light the candle on your GSM/GPRS module (I don't understand why you are using this radio - there are lower power radios specific to meter reading)?

Solutions, do you have specific examples of such radios? If you are suggesting low-power wireless, I assume that there will be no connectivity near the meter?



Ichan,

keep in mind that normally you have to give AT LEAST 3.6V to the GPRS modem, to ensure good operation... We had used a TPS62110 set to 3.7V. That is because you have voltage drop when the modem draws 2A of current. Small supercaps are not really expensive (depending on what  you mean by expensive). The one we used was a bit more expensive:

http://www.mouser.com/ds/2/87/Bus_Elx_DS_4308_PM_Series-7787.pdf

but as far as I remember, it was because it could deliver high current and have a self-discharge of 10uA.

[Ichan]

We are here have a rule for component price:

Code:

if((passive && !essential) > $1) YourSin++;

And I already have a bunch of it, NiCd and SuperCap will just add it badly. TPS62110 is a buck converter (?).

GPRS is because of its coverage, and 100% water supplied houses had the service available.

And now we just have excessive supplies of water (PHOTO: AFP)






Help her, will ya?

-ichan

[solutions]

The GPRS makes sense, though as was mentioned, why not charge the battery/cap off the water flow? The amount of energy you need to pick off is very negligible, you can read the flow off that generator, and you need to install the meter into the line, anyway.

We had more water than what you've pictured when I was in Thailand last year - that water has some NASTY stuff in it, so wading in it carries a lot of health risks.  Half the deaths in Thailand were electrocutions, things like people merely picking up or unplugging the TV - the "electricians" don't seem to care about how hot and neutral are wired (I got zapped picking up a TV while standing barefoot on a DRY (fortunately) concrete floor a couple of years before), and ground is something you walk on. Now all you need is one of your fairly-regular  7.0's, or a volcano going off, with the flood there already....



^ send me your sister's phone number, BTW...she's a cutie  

[Faros]

Dear Ichan,

About temporary (short period) high current drain in circuit that is designed for 24/7 monitoring.
Super capacitor:
There are two types of super capacitor, (1)  up to 3F with an ESR in a figure around 75 Ohms and those are intended for temporary voltage hold of MCU’s, RAM’s and sometimes micro power circuit. The subjected loads should draw not more than 1mA (typically 200 UA); at this current rate the ESR could drop a tolerable 100 mV. Although 75 Ohm ESR sounds huge, it has the advantage of not loading the circuit power supply when 1st turned on.  A flat (discharged) capacitor with ESR of 75 Ohm will draw an acceptable 67 mA @ 5V when 1st turned on and going down to zero when completely charged. On the other hand (2) there are up to 350 F capacitor intended to provide enough energy when needed, those have an ESR of 0.0x (my 35 F has an ESR of 0.02 Ohm), however they require appropriate power management since the 0.02 Ohm will theoretically draw 250 A when 1st turned on !!! …

So for type (2), a constant current charging is mandatory to avoid the super Cap short circuiting your power supply when 1st turned on, CCR will keep the voltage/time relation linear and you should then tickle the charge against leakage, I would prefer that such action is to be proficient using your MCU. One more thing, the type (2) super Cap will lose 30% of its capacity and has its ESR tripled by usage, so always consider this drawback in your design.

DC-DC converter is a bit tricky when it comes to power saving, a > 90% efficiency could come down to 55% in your range of interest. Go to the curves section for proper assumption of your circuit performance.

About the cost of super capacitor, here is an official example form a Chinese supplier (manufacturer)

SM-5R5-405UL (5.5V 4F -10%~+20%): 1.1USD/pc, MOQ=200, FOB Shanghai, T/T, 15 days delivery time. You can find them at www.greentechee.com .

Let us discus the alternative solutions furthermore and then choose the most convenient among them.

I will assume that your minimum target quantity > 900 pieces (I always manufacture 10% extra for 1K Qty and 5% for more) and your target field service time is at least 3 years@50 degrees without altering (changing batteries). I will also assume that The GSM will send monthly consumption values and a malfunctioning massages (if any) as an SMS.  The SMS will contain Monitor S/N, date/time stamp for current reading, date/time stamp for the reading before, and current consumption. also need to know min2max L/min and pressure as well as tube diameter.

Please feed me back if my presumptions are correct.

Regards,