Author Topic: Low Power and RFM69HW calculation question  (Read 4567 times)

overlord

  • Newbie
  • *
  • Posts: 26
  • Country: de
Low Power and RFM69HW calculation question
« on: October 15, 2016, 07:08:48 PM »
Hello,

first I would to thank Felix and the other forum guys for the support of my previous questions.

I have created a sample with BME280 which consumes in Deep Sleep Mode around 6-7 uA with RFM69W (stock Moteino including regulator).
I also created a variant with Moteino R4 and RFM69HW which consumes ~10 uA in Deep Sleep Mode using the same sketch, which is something which does not worry me much.
What worries me is the following:
My sketch send every 10 minutes a very short message keeping the radio on for 20 ms. And every 20 minutes two messages for 40 ms.
I have a 9 Volts Alkaline Battery with 600 mAh capacity connected. I assume 150 mA power consumption for RFM69HW + MCU for the time slots above.
According to the calculator on that site http://oregonembedded.com/batterycalc.htm my Battery will last 3.3 years with 30ms in average 150 mA 6 times a hour rest of the time 10uA in deep sleep
I can't believe that since my battery level I'm reporting seems to be decreasing more faster, maybe because the RFM69W draws too much power from the battery in a very short time, so maybe I need to add a buffer cap with low leakage current. Maybe I overload the battery in some way which reduces the lifetime too fast?
Is there a way to reduce the RFM69HW to a power consumption level similar to RFM69W in the RFM69 library? I found this article dealing with the inner core of the PA1 PA2 and Boost registers:
https://www.andrehessling.de/2015/02/07/figuring-out-the-power-level-settings-of-hoperfs-rfm69-hwhcw-modules/
What would be the recommended value for RFM69::setPowerLevel ?
Or do I need to change RFM69::setHighPower and RFM69::setHighPowerRegs for that purpose?

Best Regards,
René



WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: Low Power and RFM69HW calculation question
« Reply #1 on: October 15, 2016, 09:47:47 PM »
Hello,

first I would to thank Felix and the other forum guys for the support of my previous questions.

I have created a sample with BME280 which consumes in Deep Sleep Mode around 6-7 uA with RFM69W (stock Moteino including regulator).
I also created a variant with Moteino R4 and RFM69HW which consumes ~10 uA in Deep Sleep Mode using the same sketch, which is something which does not worry me much.
What worries me is the following:
My sketch send every 10 minutes a very short message keeping the radio on for 20 ms. And every 20 minutes two messages for 40 ms.
I have a 9 Volts Alkaline Battery with 600 mAh capacity connected. I assume 150 mA power consumption for RFM69HW + MCU for the time slots above.
According to the calculator on that site http://oregonembedded.com/batterycalc.htm my Battery will last 3.3 years with 30ms in average 150 mA 6 times a hour rest of the time 10uA in deep sleep
I can't believe that since my battery level I'm reporting seems to be decreasing more faster, maybe because the RFM69W draws too much power from the battery in a very short time, so maybe I need to add a buffer cap with low leakage current. Maybe I overload the battery in some way which reduces the lifetime too fast?
Is there a way to reduce the RFM69HW to a power consumption level similar to RFM69W in the RFM69 library? I found this article dealing with the inner core of the PA1 PA2 and Boost registers:
https://www.andrehessling.de/2015/02/07/figuring-out-the-power-level-settings-of-hoperfs-rfm69-hwhcw-modules/
What would be the recommended value for RFM69::setPowerLevel ?
Or do I need to change RFM69::setHighPower and RFM69::setHighPowerRegs for that purpose?

Best Regards,
René

Without doing something a bit more fancy, you'll be better off with two AA batteries running with no LDO than a 9 volt and an ordinary LDO.  The reason: the LDO will simply turn the 9 volt's excess voltage into heat, and a 9 volt battery typically has fewer mah than a good AA cell.

There do exist power converters (often buck regulators) that can less wastefully convert 9 volts down to your target voltage, but many of those have a higher quiescent current.  So, at a simple level, there's that as a trade-off, although there are ways to work around that for your use case (such as, for example, only briefly turning on the converter just long enough to charge a capacitor which you then coast from).  Some ultra high efficiency voltage converter IC's will help manage that for you automatically, and you might try searching them out on digikey or the like if for other reasons you want to stay with 9 volts.  If you go that route, one of the trade-offs becomes cost, because they'll generally  be more expensive than LDO's.  Also, there may be ripple to contend with, which may or may not matter to you, the magnitude of which might be another parameter in your search.

Perhaps others here can provide additional pointers also to get the most from your 9 volt battery.
« Last Edit: October 15, 2016, 10:23:55 PM by WhiteHare »

raggedyanne

  • Jr. Member
  • **
  • Posts: 80
  • Country: au
Re: Low Power and RFM69HW calculation question
« Reply #2 on: October 16, 2016, 07:58:41 AM »
I think whitehare has covered most of it. I use to buy batteries on ebay quite cheap from China alas 99% of their batteries are fake or their label is untrue. I bought a 9v 800mah which turned out to be 80mah , one that never charged because it was filled with soil or ones that expired years beyond their expiration (yes batteries have an expiration)  :'(

Buy a branded battery from a shop in sealed packaging that has not expired.

joelucid

  • Hero Member
  • *****
  • Posts: 867
Re: Low Power and RFM69HW calculation question
« Reply #3 on: October 16, 2016, 01:55:58 PM »
With these settings a 9v battery should last a couple of years. The discharge curve of 9v alkaline is pretty steep initially down to 9v or so and then flattens out - maybe that's what you're seeing?

Rfm69hw wise I'm finding that I very rarely need more than -2 dBm. So if you setPowerLevel( 0 ) which translates to 4 dBm or 4x as much power you're likely covered unless you have gateway issues or a very challenging rf setup.

Of course it's true that 2x aaa is a much better approach than 9v - but the 9v batteries do have an appealing form factor.

Joe

overlord

  • Newbie
  • *
  • Posts: 26
  • Country: de
Re: Low Power and RFM69HW calculation question
« Reply #4 on: October 17, 2016, 02:40:44 PM »
Thanks for all replies. I found out that battery voltage changes slightly when temperature in room is changing. Thats why I was afraid of something draining the battery too fast.
But this is not the case. The only issue is that Battery life gets shorter if peak power is to high, but I don't care I think 2-3 years is fine.
http://www.powerstream.com/9V-Alkaline-tests.htm


Felix

  • Administrator
  • Hero Member
  • *****
  • Posts: 6401
  • Country: us
    • LowPowerLab
Re: Low Power and RFM69HW calculation question
« Reply #5 on: October 17, 2016, 04:42:57 PM »
I have created a sample with BME280 which consumes in Deep Sleep Mode around 6-7 uA with RFM69W (stock Moteino including regulator).
I also created a variant with Moteino R4 and RFM69HW which consumes ~10 uA in Deep Sleep Mode using the same sketch, which is something which does not worry me much.
Actually in the the photo I see a DIY Moteino, not a LPL stock Moteino, something you've built yourself? Is this the variant you refer to in your first post being 10uA vs 6uA (stock)?

raggedyanne

  • Jr. Member
  • **
  • Posts: 80
  • Country: au
Re: Low Power and RFM69HW calculation question
« Reply #6 on: October 18, 2016, 01:58:54 AM »
The previous industrial duracell was the procell , the newest is the quantum, the one you have , like the moteino is a reproduction.
« Last Edit: October 18, 2016, 02:05:26 AM by raggedyanne »

joelucid

  • Hero Member
  • *****
  • Posts: 867
Re: Low Power and RFM69HW calculation question
« Reply #7 on: October 18, 2016, 02:06:20 AM »
Quote
The only issue is that Battery life gets shorter if peak power is to high, but I don't care I think 2-3 years is fine.

BTW while that is very true for constant discharge current many battery types are more forgiving in the pulsed discharge scenario you have. What happens is that the battery recovers substantially in the idle periods and is ready for another short high current pulse afterwards. LiMnO2 coin cells are notorious that way and I also believe it applies to alkaline batteries to some extent.

Joe

overlord

  • Newbie
  • *
  • Posts: 26
  • Country: de
Re: Low Power and RFM69HW calculation question
« Reply #8 on: October 18, 2016, 03:26:16 PM »
I have created a sample with BME280 which consumes in Deep Sleep Mode around 6-7 uA with RFM69W (stock Moteino including regulator).
I also created a variant with Moteino R4 and RFM69HW which consumes ~10 uA in Deep Sleep Mode using the same sketch, which is something which does not worry me much.
Actually in the the photo I see a DIY Moteino, not a LPL stock Moteino, something you've built yourself? Is this the variant you refer to in your first post being 10uA vs 6uA (stock)?
Hello Felix yes my DYI Moteino uses 9-10uA stock uses 6uA same sketch. I assume some slight dervivation of the regulator or I need to better cleanup the board after soldering.
There is a MCP1703A and MCP1703 variant I assume I got the MCP1703. Maybe also the RFM69HW uses more quiescient current.
 

perky

  • Hero Member
  • *****
  • Posts: 873
  • Country: gb
Re: Low Power and RFM69HW calculation question
« Reply #9 on: October 18, 2016, 03:58:59 PM »
Interesting, the MCP1703A now have voltage in to voltage out curves showing it might be used for clamping (i.e. output voltage follows input until regulation). It's still very unclear as to what it's output impedance would be as the voltage rises (although figure 2-35 appears to show it starts working at around 1.5V rather than the 2.7V quoted). Ground current still rises though until regulation so not as good as it could have been.
Mark.

Felix

  • Administrator
  • Hero Member
  • *****
  • Posts: 6401
  • Country: us
    • LowPowerLab
Re: Low Power and RFM69HW calculation question
« Reply #10 on: October 18, 2016, 04:32:43 PM »
Moteinos (so far) come with MCP1703.

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: Low Power and RFM69HW calculation question
« Reply #11 on: October 20, 2016, 02:07:38 AM »
I notice that TI has provided a design for a TH wireless node that TI says can run for 10 years on a single  CR2032 battery.  Without much effort I think you could adapt that design to use a Moteino for similar results:  http://www.ti.com/lit/ug/tidu995d/tidu995d.pdf
It uses a TPL51111 and also a high efficiency boost converter.

Felix

  • Administrator
  • Hero Member
  • *****
  • Posts: 6401
  • Country: us
    • LowPowerLab
Re: Low Power and RFM69HW calculation question
« Reply #12 on: October 20, 2016, 08:24:11 AM »
Interesting implementation, they put a booster on a CR2032, maybe to get all the juice from that coin cell, but not sure how that makes a ton of sense in terms of added design complexity since once the coin cell drops V dramatically it won't have lots left anyway ???

Granted we can learn a lot from this design and its different parts but I wonder if all the complexity justifies the outcome.

I would dub a TH node as one of the most "basic" of all nodes. Typically TH does not change so much that you'd want an alert/interrupt or immediate notice so fixed periodic wakeups are perfectly acceptable and not a downside. That also produces a smooth fixed interval log/graph when looking at the data.
We've seen some uber small and "simple" TH implementations here based on the venerable 328p, from our forum resident low-power champions with only a few components and the RFM69, some coin cell powered.

Looking at this TI we see a much more complex design, just to report TH, otherwise at very low power. My feeling is this is overengineered for such a simple task, (well of course we could argue this can do anything else that a coin cell supports, not just TH). It's about tradeoffs too, if you can get away with a TPL5110, atmega328, TH sensor, coin cell, RFM69 supporting passives and not much else, that simplicity wins in front of a complex overengineered design from a top brand like TI.

joelucid

  • Hero Member
  • *****
  • Posts: 867
Re: Low Power and RFM69HW calculation question
« Reply #13 on: October 20, 2016, 08:25:07 AM »
Quote
I notice that TI has provided a design for a TH wireless node

Wow, I certainly hope that their TH sensor really is that sensitive to voltage swings. Otherwise this would be a lot of unjustified complexity. Thanks for positing, the pass through mode booster is an interesting approach for motes that do need 3.3V.

Joe

TomWS

  • Hero Member
  • *****
  • Posts: 1925
Re: Low Power and RFM69HW calculation question
« Reply #14 on: October 20, 2016, 09:13:38 AM »
Looking at this TI we see a much more complex design, just to report TH, otherwise at very low power. My feeling is this is overengineered for such a simple task, (well of course we could argue this can do anything else that a coin cell supports, not just TH). It's about tradeoffs too, if you can get away with a TPL5110, atmega328, TH sensor, coin cell, RFM69 supporting passives and not much else, that simplicity wins in front of a complex overengineered design from a top brand like TI.
Bear in mind that TI doesn't publish example circuits for the benefit of mankind.  They publish circuits that use their parts so that they can sell their parts.  The complexity may be unwarranted for a TH mote, but it could be useful in more complex systems. 

Blindly  copying a silicon mfr's circuit isn't necessarily going to provide you with the best implementation.  One should take these circuits as a basis for learning about and understanding the components so you can design your own optimum solution.

Tom
« Last Edit: October 20, 2016, 09:16:38 AM by TomWS »