Solar charger for 2S LiPo. Protection circuit?

[ppTRN]

Hi everyone,
I am designing a circuit that needs to be supplied with 6V, so I am going for a dual LiPo cell (7.4V) and a buck converter using MC33063.
I would like for the batteires to be charger using a small solar panel, but I am struggling finding an adaguate protection circuit.

You can find attached the schematic that I intend to follow and I have a few questions.

1. Should I use a linear regulator? I ask this because I do not know how the buck will behave in absence of drained current when the battery is charged. Maybe I use it anyway with a resistor that drains a few mA? This should not be a problem since it does not waste battery but only solar power.

2. Is applying 7.4V directly to the battery the correct way of charging it?

3. The solar panel will output a current not greater than 500mA. Will the battery drain so much current that will cause a massive voltage drop on the panel? (I actually took a look to the typical V-A curves of a solar panel and the voltage is pretty stable up until the MPPT point)

4. I would love a protection IC just like DW01A, but for two cells. I found a solution online that uses one DW01A for each cell, but I do not really trust it.

5. Nothing bad should happend when the voltage delivered by the solar panel is below 8V, or low enough not to allow the regulator to properly wors, right? I also placed a diode to avoid any chance of current flowing from the batteries to the regulator (I will take into account its voltage drop when designing the regulator output voltage).

6. Is this whole solution to trow away because I am not taking into account cell unbalancement? I would buy a pre made 2 cell lipo, not connecting two in series myself.

this is my first time dealing with batteries, I did my reaserches but if you have any suggestions please feel free to share.

thank you very much!

[tunk]

The 7.4V is the nominal voltage.
Fully charged they are 8.4-8.7V (see cell specs).
Fully discharged they are ~6.0V.

Edit: For a typical discharge behaviour, see e.g.:
https://lygte-info.dk/review/batteries2012/Molicel%20INR18650-P28A%202800mAh%20%28Gray%29%20UK.html

[SiliconWizard]

2. Is applying 7.4V directly to the battery the correct way of charging it?

No. At least not until it has already reached close to this voltage.

[Peabody]

You should use a proper charging IC that will go through the constant current, constant voltage, algorithm, and then terminate charging when the batteries are fully charged.  The TP5100 will do that for either 1S or 2S, but doesn't do balancing, and the TP5100 modules I've seen don't have any protection.  Anyway, I don't know what protection you would use for 2S.

An alternative might be to use a single battery, or two in parallel, with a regular TP4056/DW01 module, then add an MT3608 boost converter to get the 6V.

How much current will your circuit draw?  Will it sleep from time to time at much lower current?

Using a linear charger like the TP4056, the panel will provide as much current as it can at just above the battery voltage.  The charger won't draw the panel voltage below the battery voltage.  It can only draw current to charge the battery when the panel voltage is above the battery voltage.

The panel should probably be a 5V panel.  Maybe 6V.  But not higher.  With a linear charger, higher voltage just dissipates more heat when the sun is bright.  Actually, opinions vary on this.

[sleemanj]

I would go the boost route with parallel cells, it's a lot simpler, you can just grab a protected TP4056 board, and a suitable boost converter, and that's it.

If you go series, then you'll need two protection circuits one for each cell, and either a specific 2S charger, or a CC/CV supply configured appropriatly for the charge requirements, and then add your own buck converter...

[ppTRN]

Thank you all for your advices.

The circuit will spend most of the time sleeping and will be woken up via RTC once a day and it will be active for no longer than one hour. In its working window it will draw short peaks of 400mA, once every couple of minutes.

An alternative might be to use a single battery, or two in parallel, with a regular TP4056/DW01 module, then add an MT3608 boost converter to get the 6V.

Using a linear charger like the TP4056, the panel will provide as much current as it can at just above the battery voltage.  The charger won't draw the panel voltage below the battery voltage.  It can only draw current to charge the battery when the panel voltage is above the battery voltage.

The panel should probably be a 5V panel.  Maybe 6V.  But not higher.  With a linear charger, higher voltage just dissipates more heat when the sun is bright.  Actually, opinions vary on this.

Even if I use an higher voltage solar panel, the wasted power will come directly from the sun, it may stress more the regulator but it should not compromise the charging process.

EDIT: NO, max input voltage for TP4056 is 6.5V

I would go the boost route with parallel cells, it's a lot simpler, you can just grab a protected TP4056 board, and a suitable boost converter, and that's it.

If you go series, then you'll need two protection circuits one for each cell, and either a specific 2S charger, or a CC/CV supply configured appropriatly for the charge requirements, and then add your own buck converter...

I think I will go for this approach, TP4056 for charging, DW01A for protection and a boost to get the 6V.

[NiHaoMike]

If you go with a 12V solar panel due to low cost (e.g. those cheap car battery maintainers), a buck converter with a capped duty cycle will be close enough to proper MPPT for this application. If there's already a microcontroller in the system with some left over resources, you could have it measure the output current of the converter and try to adjust the duty cycle limit for the most output current, that's a simple way to do "proper" MPPT without explicitly measuring power. (It doesn't even need to know the solar panel voltage or current, you're just trying to get the most charging current you can.)

[Peabody]

The circuit will spend most of the time sleeping and will be woken up via RTC once a day and it will be active for no longer than one hour. In its working window it will draw short peaks of 400mA, once every couple of minutes.

I think a lot depends on how long those short peaks of 400mA last, as well as how much current it will draw during the rest of the active hour.  In any case, you have to figure out what your battery requirements are, given that there could be several days of cloudy weather in a row (if it absolutely has to run every day), and keeping in mind that the boost converter will draw almost twice the current from the battery that it supplies to your circuit.  Also, will this hour of run time take place during the active charging period, or at night?

As an aside, a boost converter will draw some material current even when it sees very little load.  So instead of having your circuit sleep for 23 hours per day, you could put a P-channel mosfet ahead of the boost, which would let you switch off all current to the circuit, and the RTC (such as a DS3231) would switch it back on at the alarm time.  So there would be zero battery outflow except during the active hour.  But it would still charge just fine.

Even if I use an higher voltage solar panel, the wasted power will come directly from the sun, it may stress more the regulator but it should not compromise the charging process.

EDIT: NO, max input voltage for TP4056 is 6.5V

A linear charger has to dissipate as heat the difference between the input voltage and the battery voltage, multiplied by the current.  So at 1A they can get quite warm.  I think all of them have built-in provisions to cut back on charging current if they get too hot.  So the higher the panel voltage, the more likely it is that charging current will be reduced.  So for a single lithium cell, a 5V panel works quite well.  The only benefit of a higher voltage panel is that it might give you a little more charging current when it's overcast.  Everything changes when you get into switching stuff, but that doesn't necessarily give you that much improved performance.  In my limited experience, a 5V panel and a linear charger gives you almost all of the performance you can get from fancier circuits.

[Peabody]

I should have said - One thing you should consider is that since the current from the panel will be flowing through the charger, when the battery has fully charged the charger will shut off current, and the battery will then have to supply the circuit even though the sun may be shining brightly.  The charger will resume charging only when battery voltage has fallen to about 4.1V.

A "load sharing" (aka "power path") circuit can prevent that, but is a bit complicated for a solar supply.  I can post one if you're interested.

I think TI and others make charger ICs that have a built-in load sharing circuit.  I have no experience with them, and don't know if they support solar charging.  Others here may be able to offer guidance on them.

Edit:  Looking further, TI's BQ24074 appears to be a good choice for solar charging, and it comes with power path built in.  It comes in a QFN package, which is not so convenient for hobby use.  However, Adafruit makes a module:

https://www.adafruit.com/product/4755

with a user guide:

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

You would still need the boost converter.

[xfce]

I'm interested in this topic too.

I want to make solar charged device to water the garden once a day for 20 minutes.

Here are my requirements:
- I would like to use 1s or 2s LiFePo4
- The output voltage must be more than 7 V at least for the pump.
- pump draws max 3 A @ 8 V
- quiescent current must be low - less than 10 uA
- charging from USB
- Charging from a small solar cell
- the circuit should cut off the battery if the voltage is too low
- output with 3.3 V for the MCU


@Peabody I read your post and the option with BQ24074 looks quite good. The most straightforward solution is to use one LiFePo4 battery and then boost the voltage up to 8 V. But the downside is that the boost converter and the battery must handle high currents. More than 6 A to make 3 A@8V.

Maybe the better solution is to use IC BQ25790. Then I can use it with different inputs solar or USB and power the water pump directly from the batteries. I just need to add LDO for 3.3 V output with low quiescent current for the MCU power source. Or better buck converter.


Well guys, what do you say about it?

[ppTRN]

If you go with a 12V solar panel due to low cost (e.g. those cheap car battery maintainers), a buck converter with a capped duty cycle will be close enough to proper MPPT for this application. If there's already a microcontroller in the system with some left over resources, you could have it measure the output current of the converter and try to adjust the duty cycle limit for the most output current, that's a simple way to do "proper" MPPT without explicitly measuring power. (It doesn't even need to know the solar panel voltage or current, you're just trying to get the most charging current you can.)

The MPPT solution surely would be nice, but I do not think I will implement it due to space constraints, costs (mainly designing costs) and due to the fact that the geographic regions that it will be delivered is irradiated by the sun for most of the time, so I will just forget about MPPT and if needed I will increase the battery capacity.

Intresting solution tho, it might be usefull and I will try to implement it anyway, just for tests.

I think a lot depends on how long those short peaks of 400mA last, as well as how much current it will draw during the rest of the active hour.[...]  Also, will this hour of run time take place during the active charging period, or at night?

As an aside, a boost converter will draw some material current even when it sees very little load.  So instead of having your circuit sleep for 23 hours per day, you could put a P-channel mosfet ahead of the boost, which would let you switch off all current to the circuit, and the RTC (such as a DS3231) would switch it back on at the alarm time.  So there would be zero battery outflow except during the active hour.  But it would still charge just fine.

Unfortunaly I did not have any chance to evaluate the duration of the 400mA spikes. I am taking into account a 2 seconds peak, that is way more than enough (it is a lora transmission burst).

It will likely be active during the day but I was thinking to use just the battery power and not draining power directly from the solar panel, since I do not know how much current i can derive from it. That is also the reason why I am setting a very low current charge for the lipo. When the circuit wakes up, the charger is DISABLED, so that it does not provide any current.

Disabling the boost when the uC is inactive was already in my mind, it surely is a good move.

[...]A "load sharing" (aka "power path") circuit can prevent that, but is a bit complicated for a solar supply.  I can post one if you're interested.

[...]TI's BQ24074 appears to be a good choice for solar charging, and it comes with power path built in.  It comes in a QFN package, which is not so convenient for hobby use.  However, Adafruit makes a module:

https://www.adafruit.com/product/4755

with a user guide:

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

You would still need the boost converter.

As said above, i will just avoid the situation in which the solar charger needs to provide power to the circuit by disabling the charger when the circuit is awaken.
This could still be usefull and if you would like to share any material on the topic it would be very appreciated!

Allow me to post the schematic I came up so far, it is basically the one that you can easily find online when looking for TP4056. I hope I manage to get some LiPo with internal NTC so that i can use the temperature controll embedded in the charge regulator.

Thanks again for all of your answers and help!

[tunk]

Mostly out of curiosity, what will happen when the solar panel
cannot produce 300mA?

The cheap solar power banks has this setup:
solar panel -> (Schottky) diode -> BMS -> cell/battery

Could this be an alternative? A drawback is that you miss one
level of safety: You rely on the BMS to stop charging (and not
the TP4056).

[Peabody]

@xfce, I think both TI chips are only for LIPO or Li-ion, not LIFEPO4.  But perhaps they have other parts for those cells.  The main problem with the BQ25790 is that it's a BGA part (good luck with that), and there are no modules available other than the TI Evaluation Board, available at Digikey for US$178.  What makes the BQ24074 a possibility is the $12 Adafruit module.

But as you note, the big problem is the current you will need.  I wonder if a 12V lead acid battery and charging system might be a better option.  They make small batteries like the ones they put in UPS boxes.  And I think the charging process is simpler.  I would bet you could find such solar-charged designs on Youtube.

[Peabody]

I think a lot depends on how long those short peaks of 400mA last, as well as how much current it will draw during the rest of the active hour.[...]  Also, will this hour of run time take place during the active charging period, or at night?

As an aside, a boost converter will draw some material current even when it sees very little load.  So instead of having your circuit sleep for 23 hours per day, you could put a P-channel mosfet ahead of the boost, which would let you switch off all current to the circuit, and the RTC (such as a DS3231) would switch it back on at the alarm time.  So there would be zero battery outflow except during the active hour.  But it would still charge just fine.

Unfortunaly I did not have any chance to evaluate the duration of the 400mA spikes. I am taking into account a 2 seconds peak, that is way more than enough (it is a lora transmission burst).

I thought Lora was supposed to be low power.

It will likely be active during the day but I was thinking to use just the battery power and not draining power directly from the solar panel, since I do not know how much current i can derive from it. That is also the reason why I am setting a very low current charge for the lipo. When the circuit wakes up, the charger is DISABLED, so that it does not provide any current.

No, I don't think that's right.  The battery is always available to provide load current as needed.  There's no reason to limit the current the panels might also provide.  In bright sun, they may be able to provide all the load current needed, plus charge the battery.  But even with less light, they will provide what they can - at the battery voltage.  I think you should leave the charge current of the TP4056 set to the default 1A.  There is no reason to set it lower unless the battery would not tolerate 1A.

Disabling the boost when the uC is inactive was already in my mind, it surely is a good move.

[...]A "load sharing" (aka "power path") circuit can prevent that, but is a bit complicated for a solar supply.  I can post one if you're interested.

[...]TI's BQ24074 appears to be a good choice for solar charging, and it comes with power path built in.  It comes in a QFN package, which is not so convenient for hobby use.  However, Adafruit makes a module:

https://www.adafruit.com/product/4755

with a user guide:

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

You would still need the boost converter.

As said above, i will just avoid the situation in which the solar charger needs to provide power to the circuit by disabling the charger when the circuit is awaken.

Again, there is no need to do that.

This could still be usefull and if you would like to share any material on the topic it would be very appreciated!

Not sure what "this" refers to, but below is the schematic for load sharing when using solar panels in a 3.3V circuit.  The opamp turns on the mosfet if the output voltage is below the battery voltage.  The schottky diode at the input to the charger is optional, and usually not needed if the circuit sleeps a lot.  For a 5V supply, the regulator would probably need to be buck/boost.

And the other two circuits illustrate the use of the DS3231 to control power to the circuit.  The RTC is powered by its own coin cell which should last for years.  Its alarm brings the INT/SQW open drain output low, which turns on the mosfet.  The processor, after setting the next alarm time, can  turn off its power by clearing the alarm interrupt flag, which lets INT/SQW go back high.  With a single cell and no load sharing, the low-voltage version should work fine.

[Peabody]

Mostly out of curiosity, what will happen when the solar panel
cannot produce 300mA?

If the open circuit voltage of the panel at the current level of sunlight is at least equal to the battery voltage, the output voltage of the panel will rise to the battery voltage level, or slightly above, and it will provide as much current as it can at that voltage (per its I/V curve).  That current will either supply the load or charge the battery, or both.  The charge current setting of the TP4056 is just the maximum current it will pass through.  It will pass less than that with no problem if the maximum current isn't available.  And since it's a linear charger, the panel voltage will not crash.

[Harrow]

Never just connect 4.2V directly to a lipo to charge it as you run a risk of excessive charging current causing the battery to swell and even catch fire.

Typically you want to charge at 1C (or less) constant current until the battery voltage reaches 4.2V/cell, after which you apply a constant voltage of 4.2V/cell until the charge current is less than 0.1C. (These are the current charge rates used for radio control aircraft lipos which have high discharge rates, they might be slightly lower for some other lipos but the charging principle remains the same.)

Also, if you have multiple cells connected in series (such as the 7.4V you describe), you cannot guarantee that the voltage of both cells is the same, so you should be balancing the cells at the same time. (That's why the charge lead from the 7.4V lipo will have 3 wires.)

[xfce]

@xfce, I think both TI chips are only for LIPO or Li-ion, not LIFEPO4.  But perhaps they have other parts for those cells.  The main problem with the BQ25790 is that it's a BGA part (good luck with that), and there are no modules available other than the TI Evaluation Board, available at Digikey for US$178.  What makes the BQ24074 a possibility is the $12 Adafruit module.

But as you note, the big problem is the current you will need.  I wonder if a 12V lead acid battery and charging system might be a better option.  They make small batteries like the ones they put in UPS boxes.  And I think the charging process is simpler.  I would bet you could find such solar-charged designs on Youtube.

I'd rather not use lead-acid batteries because they don't last long in hot temperatures. At 20°C, they survive around 4 years, but at 30°C or higher, it's only 2 years or even less.

The BQ24074 is a suitable option, but you'll need a powerful  DC-DC boost converter to activate it for the pump to function properly.

I found this IC BQ25798. I think it's the best option for my case. I would use 2S lifePo4. Must read all the details about this.

[xfce]

It's quite a complex problem. Adafruit module with bq24074 has battery undervoltage protection. The BQ25798 doesn't have it. Probably I must use a separate battery protection IC.

[Peabody]

I don't see anything in the BQ25798 datasheet that suggests it will charge LiFePo4 cells.  But maybe the constant voltage level is adjustable to allow that.

[xfce]

I don't see anything in the BQ25798 datasheet that suggests it will charge LiFePo4 cells.  But maybe the constant voltage level is adjustable to allow that.

You are right, I was misled by data from the search table on the TI web page.


Probably it works fine if REG01_Charge_Voltage_Limit register is set to lower max voltage.