Author Topic: A solar supercap powered Moteino (15Farad charged by BQ25504)  (Read 71167 times)

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
A solar supercap powered Moteino (15Farad charged by BQ25504)
« on: February 07, 2017, 05:31:03 PM »
I built a little solar charger based on the BQ25504 that charges a 15F supercap.  I programmed the resistors to limit the voltage to no more than 3.6v.  Last night I ran an accelerated load test using an 8Mhz (internal resonator) Moteino with its voltage regulator removed and with the solar panel disconnected.  The Moteino woke up once a second to do 3 things:  1. turn on an LED for 1ms to simulate a sensor load, 2. check the voltage level while the LED is on (so that it was reading a loaded voltage), and 3. transmit a packet containing the voltage data using the RFM69HW . Bottom line: 14,111 packets transmitted before the voltage dropped below 2.31v,  after which the RFM69HW stopped transmitting.

Not bad for a first attempt.  :)

ChemE

  • Sr. Member
  • ****
  • Posts: 419
  • Country: us
Re: A solar supercap powered Moteino
« Reply #1 on: February 07, 2017, 07:06:17 PM »
Neat project, I'll be watching this.  I wonder why your radio stopped transmitting at 2.31V when others have successfully run the HW down to 1.8V.  Perhaps BOD cut your trial short?

EDIT: I'd be extremely interested to see how many packets this test bed could get sent using my automode radio code.  I would have to rework it a bit to appropriately deal with a HW but you can sleep the uC much faster since you are not awake while the packet is being sent.
« Last Edit: February 07, 2017, 07:12:03 PM by ChemE »

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: A solar supercap powered Moteino
« Reply #2 on: February 07, 2017, 08:07:13 PM »
Neat project, I'll be watching this.  I wonder why your radio stopped transmitting at 2.31V when others have successfully run the HW down to 1.8V.  Perhaps BOD cut your trial short?
I wonder the same thing.  I have BOD turned off at the fuse level, and I never turned it on in software, so BOD's not a factor.  The RFM69HW is configured to run at 300kbps but at power level zero (set using the method in Felix's library).  According to Felix, the RFM69HW requires a minimum of 2.4v to run because of the particular RF switch it contains.  So, I''m not sure how people are able to run it from 1.8v.


EDIT: I'd be extremely interested to see how many packets this test bed could get sent using my automode radio code.  I would have to rework it a bit to appropriately deal with a HW but you can sleep the uC much faster since you are not awake while the packet is being sent.

For that test, all you need is the supercap, which is: https://www.digikey.com/product-detail/en/vishay-bc-components/MAL219691203E3/4701PHBK-ND/5015885
You can simply hook it up to a power supply to charge it to 3.6v, disconnect it, and then connect it to a moteino just as you would a battery.

I tried a number of supercaps, and some superficial testing suggested this one seems to have a very low self discharge rate.

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: A solar supercap powered Moteino
« Reply #3 on: February 08, 2017, 08:17:21 AM »
Though I haven't yet quantified it, speaking loosely the 15F supercap charges up pretty quickly using the BQ25504 solar charger.  Unlike many  solar chargers, it doesn't need direct sunlight.  That's where the MPPT really helps. Normal ambient indoor sunlight is plenty good.  I'm still experimenting with different mini solar panels to see which one is the sweet spot of size and voltage.  However, what will be no less interesting is just how long the Moteino will run on a single charge with a more normal update pattern of, say, once every 5 minutes.  Based on this early test, together with what appears to be a very low self discharge rate of the supercap, I think the odds are reasonably good it may run for more than a month!  If so, then that alone would carry it through a lot of cloudy days.   At least where I live, I doubt we have ever had 30 cloudy/overcast days in a row within recorded history.  The supercap itself is rated for 100,000 charge cycles if operated within normal, indoor ambient temperatures.
« Last Edit: February 08, 2017, 08:54:29 AM by WhiteHare »

ChemE

  • Sr. Member
  • ****
  • Posts: 419
  • Country: us
Re: A solar supercap powered Moteino
« Reply #4 on: February 08, 2017, 09:26:59 AM »
This is a very interesting project given all that.  So 8,333 years before the supercap is fried; that will work.  Solar panels are good for 40ish years though we really don't yet know when they are dead either.  This all indicates that you can dial back quite a lot on the capacity of the cap maybe 1F would be enough to keep things chugging along given panels and the chip can still make current on cloudy days.  Still a very interesting option versus 2xAAA lithium batteries though more complex to implement.

Maybe something like this to keep a teeny form factor: https://www.digikey.com/product-detail/en/panasonic-electronic-components/EEC-RG0V155H/EEC-RG0V155H-ND/5131456
« Last Edit: February 08, 2017, 09:33:32 AM by ChemE »

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: A solar supercap powered Moteino
« Reply #5 on: February 08, 2017, 10:04:39 AM »

Maybe something like this to keep a teeny form factor: https://www.digikey.com/product-detail/en/panasonic-electronic-components/EEC-RG0V155H/EEC-RG0V155H-ND/5131456

Well, in terms of form factor, the 15F supercap  I'm already using is actually smaller than the one you're proposing.   ;)

What's hard to evaluate though is the self-discharge rate of various supercaps without actually getting one to try.  The datasheets don't give good info, and the few I've sampled have been all over the map.  Making it worse is that supercaps behave more like weak button cells than regular capacitors.  That is to say, they have voltage "recovery" after using them in bursts, and the unloaded voltage is not a good indication of what the usable voltage is.  So, although I'm sure it should be possible to measure the self discharge rate, it's a bit tricky, and I haven't figured out a good way to do it yet.  i.e. doing it properly will probably involve more than just measuring the starting and ending voltage after a given lapse of time--which is how self discharge tests are typically done--but will instead need to consider how much of that is actually usable energy.

Any suggestions about how to setup a good test would be more than welcome.
« Last Edit: February 08, 2017, 10:15:10 AM by WhiteHare »

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: A solar supercap powered Moteino
« Reply #6 on: February 08, 2017, 11:03:43 AM »
So, why not just use a primary battery and forget about solar altogether?  Batteries are easier and cheaper, but for low latency remote control, you're either going to need big batteries or be prepared to change them a lot.  With solar, I'm hoping it will be both smaller and set-and-forget. 

For instance, if you want to turn on a light or a motor or something at the push of a button, your receiver will need to do a lot of receiving, and at 16ma that adds up.  Even with listen mode, it adds up if you want quick response: say, within 100ms latency, or maybe 1 second tops.  Button cells alone won't last very long in that application before they need changing.

So, if there appears to be some overkill in the design, it's because I'd like to also have it cover this more challenging use-case of low latency remote control plus maybe do more, like light the light or spin the motor or make sounds or what have you.  I think the charger itself is already about as good as it's going to get, and so the variables I'm playing with now are just panel size, panel voltage, and capacitor capacity and self-discharge rates.  Plus, any improvements on the moteino side reduce those requirements.  If I do this, I really do want it to be robust and highly reliable, and so I'm leaning towards overkill rather than run the risk of underkill.  Otherwise, it's back to batteries....
« Last Edit: February 08, 2017, 11:30:39 AM by WhiteHare »

ChemE

  • Sr. Member
  • ****
  • Posts: 419
  • Country: us
Re: A solar supercap powered Moteino
« Reply #7 on: February 08, 2017, 11:17:24 AM »
Well, in terms of form factor, the 15F supercap  I'm already using is actually smaller than the one you're proposing.   ;)

Yeah I failed to notice that, and I also failed to notice that the part I linked to has to be bought in multiples of 500!  In looking at the datasheets, it also looks like the smaller caps can't handle the kinds of current that we need anyway so 15F does look to be a nice size.  Your use case of a remote controlled motor isn't one I've considered so I understand the need for extra beef.  I'll be curious to see where you take this.

joelucid

  • Hero Member
  • *****
  • Posts: 868
Re: A solar supercap powered Moteino
« Reply #8 on: February 08, 2017, 11:21:36 AM »
Cool project!

Quote
Neat project, I'll be watching this.  I wonder why your radio stopped transmitting at 2.31V when others have successfully run the HW down to 1.8V.  Perhaps BOD cut your trial short?

I bet this is because the rfm69hw draws much more in Tx than the LED. Vcc should be measured during tx for this.

Felix

  • Administrator
  • Hero Member
  • *****
  • Posts: 6866
  • Country: us
    • LowPowerLab
Re: A solar supercap powered Moteino
« Reply #9 on: February 08, 2017, 01:14:43 PM »
Cool project!

I bet this is because the rfm69hw draws much more in Tx than the LED. Vcc should be measured during tx for this.

Maybe the problem is dips below 1.8v which trigger the perpetual reset condition (and I don't know the current draw of that without digging the DS), although if it could recover from that back to the 2.3v I think the 328 should come back to life. But again I havent looked at the DS so I'm talking hypothetically.

On a separate note - the RFM69HWs have started to come with a new RF switch, which I could not identify (marked with "100" and the morse code for dating). Not sure about the voltage limits of this without actual testing, but it could reset the assumptions about the HW working down to 1.8v.

WhiteHare maybe you can check this mote to see if it has this new switch or still the old G4C switch?
Also - if you share the schematic at least, this will be a sticky!
« Last Edit: February 08, 2017, 01:17:13 PM by Felix »

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: A solar supercap powered Moteino
« Reply #10 on: February 08, 2017, 01:42:27 PM »
To settle the matter I just now plugged the very same Moteino into a 1.8v bench power supply, and... it works!  So, I think Joe's explanation is probably right.  When I find some time I'll switchover to checking the voltage during the RFM69HW transmit and then sending that voltage measurement instead in the follow-on packet.

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: A solar supercap powered Moteino
« Reply #11 on: February 09, 2017, 06:52:15 AM »
Joe was right.  It's confirmed.   :)  I changed the code to read the voltage immediately after the Tx (also now a single shot voltage measurement--no longer waiting for two of the same voltages in a row).  Now on the new accelerated load test, the voltage that got reported on the very last transmission before failure was 1.63v.  Technically speaking, that was the voltage measurement that was taken immediately after the second to last transmission, so the voltage reading taken just after the very last transmission would have been even lower.

Also, as before, I do turn on the ADC and immediately start an ADC conversion (to be thrown away) during the wakeup ISR.  Tx takes about 1ms, so by the time the single shot voltage measurement is taken immediately afterward, the ADC has already been active long enough to get a decent voltage measurement.
« Last Edit: February 09, 2017, 08:19:20 AM by WhiteHare »

perky

  • Hero Member
  • *****
  • Posts: 873
  • Country: gb
Re: A solar supercap powered Moteino
« Reply #12 on: February 09, 2017, 08:16:54 AM »
So where is that extra 2.31V - 1.65V voltage drop coming from? I thought supercaps had relatively low ESR values (higher than normal caps, but lower than batteries). Is there any other resistance in line like a ferrite? Or has the supercap you've chosen an older generation type with high ESR?

Mark.

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: A solar supercap powered Moteino
« Reply #13 on: February 09, 2017, 08:22:49 AM »
The rated ESR of the supercap is 1.8 ohms.  The nominal Tx current drawn is between 30-40ma.  So, that alone doesn't account for it.  I got an impression that the voltage drop seems to get progressively bigger as the voltage declines.  Perhaps that's a clue.
« Last Edit: February 09, 2017, 08:31:39 AM by WhiteHare »

WhiteHare

  • Hero Member
  • *****
  • Posts: 1300
  • Country: us
Re: A solar supercap powered Moteino
« Reply #14 on: February 09, 2017, 09:46:32 AM »
Presently I'm using a 60mmx60mm 4v solar panel, but I would guess I could use one that's half that size, or even smaller.  The way I'm looking at it, the panel only needs to be large enough to fully charge the supercap from 1.8v up to 3.6v on a sunny day but just from ambient indoor light.  Then, there'd be enough stored power to last 30+ days, by which time there will have been another sunny day, and the cycle repeats.  For a simple sensor node, that should cover the worst case scenario.

So, the next scenario to focus on will be the low latency remote control scenario.
« Last Edit: February 09, 2017, 09:49:56 AM by WhiteHare »