Node running on supercaps and a solar panel.

[NeverDie]

@alexsh1 said in Node running on supercaps and a solar panel.:

@neverdie you mean like using a mosfet or something?
It is used to supply 3.3v to atmega328p, I do not think I can just turn it off.

Sorry, I meant switch to "pass through".

[alexsh1]

@neverdie Yes, that's on my 'to do' list. However, right now I'd like to see how much I can extract from supercap directly wired to the node.

[gohan]

How much does the supercap self discharge?

[alexsh1]

@gohan I need to implement the voltage measurement (simple voltage divider) as I have a pass though buck converter so 3.3v is shown. All I can say now is that after a few charges self discharge really improved.

[gohan]

That is something you can do by leaving the supercap on a desk for 24 hours and a multimeter 😁

[alexsh1]

@gohan not an option. Sorry :grin:

[alexsh1]

Did anyone try using 3V solar cell for 2.7V caps?
Seems a perfect fit - no need to use two supercaps in series

https://www.ebay.co.uk/itm/3V-x-500-ma-Mini-solar-panel-epoxy-encapsulated-virtually-indestructible-1-5W/171216986698?hash=item27dd53e24a:g:14YAAOxyRhBS1H0O

[alexsh1]

@neverdie Did you make any estimates running your mote on supercap? I cannot understand how come on RFM69HW (lowpowerlab), this guy managed to extract 14,111 packets transmitted https://lowpowerlab.com/forum/projects/a-solar-supercap-powered-moteino/

This is on 15F supercap.

[NeverDie]

@alexsh1 His numbers are 100% accurate.

[alexsh1]

@neverdie I am not questioning his numbers. I am trying to understand why with a normal transceiver (non-H) I am getting so much less. I must admit also that with BQ25570, I have 2x10F in series, which makes it 5F in total only vs 15F.

I should wait until my 30F cap runs out of juice and see how many packets I extracted

[gohan]

If you are happy with your setup to power the node, why are you bothering that much to send that many more packets if you actually don't need them anyway?

[alexsh1]

@gohan This is not about being happy. Engineering mind can never be satisfied ;-)

[alexsh1]

@gohan However, in the nutshell you are right. With a small solar battery (53mm x 30mm, 5V 0.15W), this is what I get:

and today is an extremely cloudy day (no sun at all). Harvesting works really well.

I implemented a voltage divider yesterday with 1.8M / 470k resistors. Now I can monitor voltage in Domoticz:
voltage at 22:00 yesterday - 4.797V
voltage at 12:00 today - 4.887V

[NeverDie]

Try running with your fuses set to xFF xDA xC2, That way your BOD will be turned off, you'll be running from the internal 8mhz resonator, and you'll be waking from sleep in <4uSec.

[alexsh1]

@neverdie I have already all that. I am using this node https://www.openhardware.io/view/76/Stamp-size-MySensor-node

Let's see when I have final results. I have this supercap AVX 30F with ESR = 0.02ohm (vs Vishay 15F ESR=1.8Ohm used with Moteino and BQ25570 at lowpowerlab)

EDIT: Another point is signing/encryption. This would affect supercap charge as the node would be transmitting more data to GW.

[gohan]

@alexsh1 I use a total of 20F 5.5V supercaps and they discharge much more during night most likely because the buck/boost converter is not very low power and supercaps are chinese quality, but as far as I can get 36/48h of runtime without solar charge I should be fine

[alexsh1]

@gohan As you may see from the graph above my self discharge + sleep/tx consumption during the night dropped voltage from 4.851V at 21:00 (I charged supercaps initially) to 4.549V at 08:00. This is the rate of about 0.3V per 12 hours, I estimate that I have around 10 days without any charging before I run out of energy.

It took 4h 25mins on a cloudy day (see another graph) to get 2x10F in series supercaps fully charged to 5.047V

[gohan]

You are using a very efficient circuit and good quality supercaps, I loose about 1V in 12 hours

[alexsh1]

@gohan Did you try to measure your sleeping current?

[NeverDie]

@alexsh1 said in Node running on supercaps and a solar panel.:

I am using this node https://www.openhardware.io/view/76/Stamp-size-MySensor-node

If anything, you should have an advantage, since the nRF24L01 transmits at a faster bitrate, and using significantly less Tx power also.