A solar supercap powered Moteino (15Farad charged by BQ25504)

[WhiteHare]

Er, 1.8V LDO means the regulation voltage is 1.8V and the output won't ever go above 1.8V, so you'll never charge anything up!

This sounds like a disconnect.  Probably my fault for changing things around too quickly and without adequate warning.   With that in mind, is it possible you're referring to Notional Schematic v001 instead of Notional Schematic v003?  If so, then I would agree with you.  However, for context regarding the 1.8v LDO, I'm referring to Notional Schematic v003, posted above.  If you would, please have a look at v003 and see if that changes your mind regarding the appropriateness/inappropriateness of a 1.8v LDO as part of this circuit.

[perky]

Doh! Sorry WhiteHare, I was looking at the previous one. I thought it was a bit odd that you would even make that mistake, of course you wouldn't, sorry I can't see much wrong with it. I think it's about the same power wise specifically because the dominant part which is the radio is constant current, marginally better though if you consider the other parts of the Moteino which will be linearly dependent on voltage.

Mark.

[WhiteHare]

No worries.   I really appreciate all your feedback and suggestions.

[WhiteHare]

I breadboarded notional schematic v001 and, after playing around with it, realized that I could both simplify the circuit and improve its charge speed.  Attached is the new schematic.  Part count is now lower.

[Edit1:
As far as finding the right Goldilocks resistor value, what follows is the fundamental design tension that needs to be balanced:
1.  The lower the resistor value, the faster the supercap will charge.  Broadly speaking, faster charging is better, so that is the motivation to use the lowest resistor value.
2.  However, if the resistor value is too low, then under at least some lighting conditions (such as dimly lit or overcast situations) the solar panel will be unable to deliver as high a current as is being demanded.  Under those circumstances, the voltage across the small cap will fall below 2.7v, and will therefore may be too low to boot a Moteino, especially if the supercap hasn't yet charged up  to something north of 1.8v. ]

[Edit2:  So, the way to pick the resistor value boils down to picking what the minimum startup conditions are: what are the dimmest lighting conditions (within reason, of course) that you still want to almost instantly have around 2.7v across the small cap, when the supercap is uncharged?  Setup the solar panel under those lighting conditions, and then select the lowest resistor value where that is still achieved.  That's all there is to it.  The resistor value is determined by this one worst-case start-up scenario--whatever the minimum conditions are that, together with the solar panel you've picked, you've decided you want to guarantee that the circuit will still work and deliver an "instant-on" to the Moteino.]

[Edit3: There's a trade-off to putting the Schottky before the LDO instead of behind it.  In the v001 circuit, the reverse leakage was limited by the Schottky, so it was very, very low.  On the other hand, in this v004 circuit, there is reverse leakage of about 6ua through this particular LDO.  However, the benefit is that the v004 circuit avoids the very long taper-charge that comes from putting the Schottky after the LDO, as in the v001 circuit.  The saved time can be spent more productively putting additional charge on the supercap, which in turn will pay for the ongoing 6ua reverse current loss when the circuit isn't charging (e.g. there is not enough light to charge).  At the margin that may mean using a larger supercap, but it seems like a win in terms of capability.]

[Edit4: Under difficult cold-start conditions, I have doubts that this circuit would work very well with just a mere BOD enabled Moteino.  Rather, I think it would be better to use a non-BOD Moteino together with a boot circuit set to a very high hysteresis.  ]

[Edit5:  If using the NCP301, you can use a 10K resistor for Rl and a 4.3K resistor for Rh.  In testing my boot circuit with those values, I found that they will set the voltage threshold for atmega328p power-on to 2.69v, and the low voltage cutout was set to 1.80v.  ]

[WhiteHare]

Although using a fixed resistor value has the virtue of being simple, it would be better if it could be replaced by a circuit which does the following: 1.  When current is flowing into the small cap, current flow into the large supercap is shut OFF, and 2. when little or no currentt is flowing into the small cap, then current flow into the large supercap is turned ON with as little resistance as possible.  I'm pretty sure this would be the optimal way to give charging priority to the small cap.

Not sure if there's an easy way to do that with just transistors, but I suppose one might approach it by using a differential comparator chip that allows for an input offset voltage.  Then one could compare the small cap voltage to the LDO output voltage to infer whether or not a non-miniscule current is flowing into the small cap.

[WhiteHare]

I've reverted to version 001 because it allows a smaller smallcap to be used, which is cheaper than adding a couple of Schottky diodes back in.  So, this design trades off performance in exchange for lower cost and lower parts count and much lower reverse current leakage.

[WhiteHare]

Attached is a photo of the assembled version 3 Perky Smart Diode breakout board.

[Edit: I played around with it a bit.  It seems to do its job as a diode, in that it blocks reverse flow of current.  At low voltages on the supercap, it drops about 0.5v.  At higher voltages, it drops about 0.05v.

In contrast, the Schottky diode drops about 0.3v, regardless of what the supercap voltage is.

So, maybe one could get the best of both worlds by wiring the Schottky diode in parallel with the Perky Smart Diode?  I haven't yet tried it, but I'm thinking that at the lower voltages it might drop 0.3v, and then at the higher voltages it would drop 0.05v.

Well, more characerization could be done, but that was just a quick look.]

I measured a reverse current leakage of 1.6uA on the Perky Smart Diode when it was connected to a supercap that was charged up to a voltage of 2.6v.  Pretty good I'd say.

[perky]

That sounds about right, it's the sort of current I was measuring. I was quite happy with it's performance in my system.

Mark.

[WhiteHare]

I think I finally nailed it.  Attached is a schematic for a fast AND simple AND low-cost solar charger AND with a low parts count AND ultra low current leakage AND that's also easy to assemble.  I prototyped it this morning, and it works like a champ.   

[WhiteHare]

Attached are the zipped Gerber and drilling files for a breakout board of the above simple and fast solar charger v001.  To keep PCB fabrication costs low, I made the board as small as I could without sacrificing ease of assembly.

If using OSH PARK, then three PCB's will cost you $0.65, including shipping.

Enjoy!

[ChemE]

Where are these parts coming from?  I found the s1002 on digikey but in lots of 3000.  The mosfet I can't find anywhere.  Also, can you include the .brd and .sch files for eagle?  I've never worked out how to open gerbers using eagle (is this even possible?).

EDIT: Found the mosfet maybe: http://www.mouser.com/search/ProductDetail.aspx?R=0virtualkey0virtualkeyDMP2305U-7
and the diode: http://www.mouser.com/search/ProductDetail.aspx?R=0virtualkey0virtualkey1PS79SB30115

EDIT2: Is this the voltage manager? https://www.digikey.com/product-detail/en/sii-semiconductor-corporation/S-1002CA27I-M5T1U/1662-1084-1-ND/6601224 Not sure which voltage you're using; this one is the 2.7V

[WhiteHare]

I purchased  my parts from digikey.  Here are links for the three parts:

https://www.digikey.com/product-detail/en/sii-semiconductor-corporation/S-1002CA26I-M5T1U/1662-1083-1-ND/6601223

https://www.digikey.com/product-detail/en/nexperia-usa-inc/1PS79SB30,115/1727-4782-1-ND/2531260

https://www.digikey.com/product-detail/en/diodes-incorporated/DMP2305U-7/DMP2305UDICT-ND/2052810

I used diptrace, not eagle.  Probably the biggest shortcoming of diptrace is that the source files aren't easily shareable as they are with eagle. Hopefully the developers will address that someday.

[ChemE]

Thanks for the links.  This project is pretty interesting to me.  I've found some SMD mountable supercaps and wonder if any of them could work as the small/boot capacitor.  I'm always looking to make these sorts of things smaller after all.  Something along these lines:

http://www.mouser.com/search/ProductDetail.aspx?R=0virtualkey0virtualkeyCPX3225A752D

[WhiteHare]

Thanks for the links.  This project is pretty interesting to me.  I've found some SMD mountable supercaps and wonder if any of them could work as the small/boot capacitor.  I'm always looking to make these sorts of things smaller after all.  Something along these lines:

http://www.mouser.com/search/ProductDetail.aspx?R=0virtualkey0virtualkeyCPX3225A752D

Thanks for posting that part.  At first I was put-off by the 25 ohm ESR, but I think one can maybe work with that.  The small size and extremely low leakage make it stand out as unique for its  amount of capacitance.

What's surprising to me about the capacitors in the 1MF to 10MF range is just how expensive they are.  I hadn't really expected that.

[ChemE]

Hey my pleasure.  I'd like to see this project as small and cheap as possible since I plan on having a play at some point.  My hope is you can get the power requirements down low enough that a few F will get it through a few overcast days.  I found something else interesting that might be able to serve as the primary supercap.  http://www.cytech.com/products-ips Scroll down to the bottom and check out the MEC202.  The size is shockingly small and 2.2mAh used a few ms at a time should last a decently long time.  If that could work, then the whole project would shrink to not much larger than a mote.  Pair that with a very small solar panel and you're golden.