Need IC or Circuit for Solar charging for 300 mA Single LiPo Battery Charger

[Gandalf_Sr]

My project requirement is for a solar charger for a small 4.2V single cell Lithium battery (up to 1,000 mAHr) to support a small microcontroller (MCU) based system that's running off grid to gather data, e.g. small weather station, data buoy, mail box open/close detector etc.  The solar panel would likely be a single 9" panel that maxes out at 8.2V at 0.41A. I don't think the MCU can be part of master control system because I envision situations where the battery would get critically low (2.8V) and the MCU would need to throw a 'suicide switch' and disconnect itself from the battery to protect the battery; or maybe the MCU could go into some ultra-low power state where it draws nA and is close to the self-discharge of the battery?  Cost is an issue, I don't want to use a $10 IC as the basis for the design but might have to.  It would also be an advantage to find a device with legs (not QFN or BGA) so that I can hand-build small batches.

There would probably need to be a voltage supervisor device that held the reset of the MCU down until the battery voltage came up to say 3.1V. So a constant voltage, constant current regulated charge circuit is what I need.  I've found various small ICs that do MPPT charge control but they are either too big, too expensive, or too complex e.g.

- LT8611 which is a 2.5A controller that costs $10 and is QFN
- BQ24650 which looks like it would be close to my needs, costs $5, but is in VQFN only

It makes me think that maybe I could design my own small MPPT system using a uCtrllr and have the uCtrllr be the device that does the sensor measurement, battery charge control, and have a simple external system that does the protection based on the commonly-available voltage supervisor ICs.

I don't want to reinvent the wheel though and it seems to me that there must be a missive need for such a circuit; has anyone got any IC or circuit suggestions that would get me closer to my requirements?

Thanks in advance.

[tszaboo]

I've come to the conclusion that tiny solar panel energy harvesting systems dont make any sense. Cost vise. And mechanical construction wise.
The self discharge of a lithium battery is orders of magnitude higher than the usage of most of these low power Bluetooth-y RF systems. And you need the battery, a panel, the BMS. If you keep the battery charged all the time, it kills the capacity.
I'm open for suggestions for small self discharge batteries.

The only reason if you want to have solar panel, if you have GSM or Wifi, which uses a ton of power anyway. Outdoors, because indoors, just throw an USB port on it. Which means that the cost of the system is going to be high, and you can afford the ASIC charging the battery.

Or you can buy a LiSOCl2 or LiMNO2 battery and run the system for a decade.

[Peabody]

I've never tried it, but I wonder if it might be possible to reflow QFN with a hotplate or electric skillet.  It also might be possible to use a QFN as a dead bug.  Finding some solutiuon may be important because I think you'll see a lot of QFN-only parts in your search.  There are QFN-to-DIP adapter boards available for $15 for 10 of them.  If their pads extend  bit, there may be a way to solder them. 

If you could use two 5V solar panels instead of the 8V, it might be possible to pick a  processor that works anywhere between 2.4V and 5.5V, such as the Atmel AVR parts, and then you wouldn't need a regulator.  The Atmega328P, for example, can do that at 8MHz I believe, and goes into deep sleep at something like 100nA.  And it even comes in a DIP option.

The battery itself will probably have a protection circuit built in, which typically shuts it down at 2.4V.  The protection circuit may draw more current than a sleeping MCU, but it's still only a uA or two.  The protection chip is the DW01.

If you can solve the QFN problem, you might look at the MCP73871, which is a couple bucks.  I think Adafruit has a solar charger based on that chip, and there are cheap charger modules using it on Ebay and Aliexpress.  Using a module might get around the QFN problem.  This particular chip has load sharing built in, which you would otherwise have to add.

I'm told that having someone like JLCPCB do the entire build is now actually within reach cost-wise so long as you use their parts.  But I've never priced anything out.

[mark03]

Sorry to continue on a tangent, but I have not found QFN significantly more difficult to solder than QFP.  Essentially all QFNs that I have dealt with from TI, etc. have "wettable flanks" (pads extend up the side a bit).  The same hand-soldering techniques you'd use on QFP work just as well on QFN.  Less bridging, too.  For any kind of power-management or power-conversion application, avoiding QFNs will be extremely limiting.

I guess there is one exception to this... the center ground pad.  Of course, some QFPs have that too.  In any case, you don't always need to connect it---read the datasheet carefully.  On power devices you probably will, but in that case the pad will often extend up and down away from the package.  That provides a place to heat up a tinned/fluxed pad with your iron.  Worst case scenario, you leave a big-ish through-hole beneath the part so you can flow some solder from the other side.

[Gandalf_Sr]

Thanks for all the input and ideas; I have blown up LiPos before but technology has advanced since that day and I'm not (too) scared of them any more.  I have soldered QFNs by hand before, I design my own PCBs with vias under the center ground pad and a square of unmasked copper around the vias on the non-component side; before mounting the IC, I lightly tin the pad on the component side that will be under the IC, then I solder the normal 'pins' around the edge of the IC, and then go to the underside of the PCB and, with a big soldering tip, I apply a bunch of heat and a bit of solder until it flows through the vias and melts onto the IC pad.

I looked at the Adafruit design and it uses a BQ24074 which seems ideal for my needs; this article is actually very good at explaining their design approach although they (understandably) don't give a full schematic.

The obvious approach would be to buy one of their $10 PCBs and play with it before doing my own design.

[Peabody]

You might want to look at one of these too:

https://www.ebay.com/itm/USB-5V-DC-Solar-Lipoly-Lithium-Lon-Polymer-Charger-Board-MCP73871-3-7V-4-2V/182380308651

And here's the schematic for the Adafruit module:

https://learn.adafruit.com/adafruit-bq24074-universal-usb-dc-solar-charger-breakout/downloads

[Mechatrommer]

what about buck converter to 5V and feed to 4.2V lion charger? MC34063 + SLM4900 can be had from lcsc for like 20 cents and soic8...

[Gandalf_Sr]

You might want to look at one of these too:

https://www.ebay.com/itm/USB-5V-DC-Solar-Lipoly-Lithium-Lon-Polymer-Charger-Board-MCP73871-3-7V-4-2V/182380308651

And here's the schematic for the Adafruit module:

https://learn.adafruit.com/adafruit-bq24074-universal-usb-dc-solar-charger-breakout/downloads

Thanks for the schematic.

As for the MCP73871, I looked at the datasheet but the word 'solar' doesn't appear in it making the eBay listing somewhat misleading IMHO.  It seems to be a good LiPo charger that assumes that there will always be enough current available from the input supply which means that it will not deal well with a half-lit solar panel - or have I missed some sneaky way of using it?

[Gandalf_Sr]

what about buck converter to 5V and feed to 4.2V lion charger? MC34063 + SLM4900 can be had from lcsc for like 20 cents and soic8...

Thanks for the suggestion.  The MC34063 is a DC-DC converter so, if it was set up in a buck configuration, the solar panel would have to produce more voltage than the battery before any charging took place, which is OK as long as the converter is set up to self-regulate when the solar panel is loaded and its output voltage falls, how do you suggest that control mechanism is implemented?

[Gandalf_Sr]

I ordered the bq24074 board from Adafruit.

[Mechatrommer]

what about buck converter to 5V and feed to 4.2V lion charger? MC34063 + SLM4900 can be had from lcsc for like 20 cents and soic8...

Thanks for the suggestion.  The MC34063 is a DC-DC converter so, if it was set up in a buck configuration, the solar panel would have to produce more voltage than the battery before any charging took place, which is OK as long as the converter is set up to self-regulate when the solar panel is loaded and its output voltage falls, how do you suggest that control mechanism is implemented?

i dont have solar project experience its just right out from my head (stock i have), if you need to squeeze lower voltage from your solar panel, maybe a boost converter paralleled with buck to give 5V output. if solar goes above say 6V, boost will be shut off by logic level mosfets combination or something... by sensing or biasing the solar or buck output (isolated output from boost through diode). boost like cheap me2109F/me2139 can operate down to 0.9V. when MPPT is concerned, you can do that in MCU, but when mcu is dead (low batt), the boost/buck will do whatever wisdom they have to put any charge to battery to get mcu working again however inefficient it is. SLM4900 charger (just an example of its kind, there are many others compatible IC) has PROG pin to control current (i never tried though) that may deal with MPPT feed from mcu adc or such. but your circuit will expand bigger with discrete solution and many inductors than single IC solution and you may not be interested anyway, just giving some other cheap possibility i maybe talk nonsense. if you only produce a few boards, the mentioned single IC solar charger board solution should be interesting. ymmv.

[Peabody]

As for the MCP73871, I looked at the datasheet but the word 'solar' doesn't appear in it making the eBay listing somewhat misleading IMHO.  It seems to be a good LiPo charger that assumes that there will always be enough current available from the input supply which means that it will not deal well with a half-lit solar panel - or have I missed some sneaky way of using it?

I don't know, but Adafruit has another "solar" charger using the MCP73871:

https://www.adafruit.com/product/390

https://learn.adafruit.com/usb-dc-and-solar-lipoly-charger/using-the-charger

[Gandalf_Sr]

Thanks for the link, the answer seems to be that the MCP73871 does have a feature called VPCC that allows it to back off on current when the input voltage falls.  From the design notes page under Solar Optimization...

The issue we have here is that the [solar panel] voltage collapses during high current draw. We need to find a way to keep the lipo charger from drawing too much current, and backing off when the voltage starts to droop. We looked high and low and finally found a chip that has something like this built in. The MCP73871 calls it Voltage Proportional Charge Control (VPCC) and basically, it does precisely what we want. We can set the voltage to a point just above the battery charge voltage point (say 4.5V) and then instruct the charger to draw as much current as possible. It will automatically increase/reduce the charge rate to keep the voltage higher than 4.5V!

They also add something they call a BFC which looks just like a capacitor 

Then there's the price... $17.50 

[EDIT] So I just ordered 2 of the eBay MCP73871 PCBs from Hong Kong, now I can do a head to head comparison.

[SiliconWizard]

Sorry if I missed the info, but I don't think I have seen any specs regarding how much power your system is supposed to draw (max, min, average), which would be a start to give you proper advice.
Another important spec would be how long your device would be supposed to function between charges (as obviously access to sufficient light for the solar panel may not be guaranteed all the time.)

I know you seem to have made your design choices already regarding the power source, but as some have mentioned already, your choice may not be the most relevant here.

As others have said, I would personally NOT recommend using a LiPo-based solution here (but would be more or less convinced after I get more info about your specs.)

[Peabody]

It's not completely clear, but there may be a difference between the two chips as to what happens if the input voltage rises above, say, 5.5V.  There's a note on the Adafruit MCP part that says when the load is running on the input power, it's just a straight passthrough.  But the datasheet of the BQ seems to say that the output is always regulated to no more than something like 4.4V.  So you'll need to test to see what happens at OUT when IN goes to 6V.  This could be an important difference if you want to power your controller directly from OUT, and not use a regulator.

Other than that, and the package, I don't see much difference between the two chips.  Well, and the fact that the MCP seems to need a big-ass honking capacitor to keep it from oscillating.  Adafruit says the BQ doesn't need that.

It looks like the BQ is a newer design.  It's too bad the module makers of the Far East don't make their version of the Adafruit BQ module.