Looking for CN3801 MPPT (MPPC really) circuit design help.

[tifkat]

Greetings all,

Trying to design a 1S 14500 600mAh LiFePo4 cell charge controller which will run off a 5.0v - 6.0v ~1W solar panel, for use in a wireless doorbell button. I've scoured AliExpress and eBay for a suitable existing module, and can find Li-Ion / LiPo 'solar charge' modules, but not LiFePo4. There are some using the CN3058(E) but they aren't optimised for solar, but for a stable DC input.

After looking through the Consonance website, I think the CN3801 is a suitable chip, as it handles the input voltage I wish to supply, from a solar panel, and is targeted at LiFePo4 cells.  The CN3801 uses the 'Constant Voltage Method' to track the MPP of the solar panel. It has a pin labelled as 'MPPT' and it says this in the datasheet:

The Maximum Power Point Tracking -- (page 6)

CN3801 adopts the constant voltage method to track the photovoltaic cell’s maximum power point. From I-V
curve of photovoltaic cell, under a given temperature, the photovoltaic cell’s voltages at the maximum power
point are nearly constant regardless of the different irradiances. So the maximum power point can be tracked if
the photovoltaic cell’s output voltage is regulated to a constant voltage.

CN3801’s MPPT pin’s voltage is regulated to 1.205V to track the maximum power point working with the
off-chip resistor divider(R3 and R4 in Figure 1).

The maximum power point voltage is decided by the following equation:

        VMPPT＝1.205×(1＋R3／R4)

So, obviously, the controller is looking to see 1.205v on the MPPT pin, and it adjusts the duty cycle of the MOSFET to try to achieve that. Is that a correct reading of this? I am trying to find out if the 6v panel I have ordered is 6v at maximum power point, or if the Vmpp is lower, but let's assume 6v is the Vmpp value. Do I simply select R3 and R4 to setup a voltage divider which gives 1.205v, from the 6v supply coming from the panel? The typical application circuit is on page 2 of the datasheet.

I've had difficulty getting the English language version of the datasheet from the Consonance website, so I have a cached copy on my own web server:

CN3801 Datasheet - English

[Peabody]

If the charger adjusts the duty cycle to keep the input voltage constant, how does it also keep the output voltage at the nominal regulated level?  Could someone please explain this?

I think a "6V" panel will have an open circuit voltage in full sun of something like 6.7, and a voltage at the MPP of a bit less than 6V.  But all the specification numbers are with the panel in direct sun at local solar noon on the equator with no air pollution.  You have a different I/V curve at every level of illumination, but the voltage at the MPP on each such curve doesn't change very much.  The MPP current drops a lot as illumination decreases, but the voltage only drops a little.  So using a constant voltage is a moderately good approximation of MPP.  I think some designers pick a voltage level that matches the MPP on a partial illumination curve - on the grounds that you probably have excess power anyway from the panel in full sun, but it's in partial illumination that you need to stay as close a possible to the MPP to get any useful power at all.

If I read the datasheet correctly, this chip is a buck converter only.

[fourtytwo42]

You need to look at your intended panels data sheet very carefully indeed especially if you have no experience of PV cells. I suspect the panel you are talking about has a VOC of 6V whereas VMPP will be more like 4.5V and even then only at 100% illumination. This controller chip does NOT TRACK VMPP rather it allows a fixed desired VMPP to be selected, furthermore it is a buck only converter meaning it cannot charge a battery who's voltage is near to the panels VOC. A better chip is the LTC3130 that can operate in both buck & boost modes and has a similar VMPP setting pin.

[fourtytwo42]

If the charger adjusts the duty cycle to keep the input voltage constant, how does it also keep the output voltage at the nominal regulated level?  Could someone please explain this?

It is quite simple to combine multiple terms in a feedback loop such that the error signal represents the sign & magnitude required to correct both conditions applying whatever priority might be required as well.  Perhaps you should think of it in terms of energy transfer & the resultant effect on voltages. The highest priority here would be the output voltage BUT if the energy required to maintain it is to low the input voltage would be allowed to rise above the setpoint BUT if the energy required to maintain the output voltage is to high and the input voltage would fall the energy is reduced and the output voltage/current falls.

[Peabody]

If the charger adjusts the duty cycle to keep the input voltage constant, how does it also keep the output voltage at the nominal regulated level?  Could someone please explain this?

It is quite simple to combine multiple terms in a feedback loop such that the error signal represents the sign & magnitude required to correct both conditions applying whatever priority might be required as well.  Perhaps you should think of it in terms of energy transfer & the resultant effect on voltages. The highest priority here would be the output voltage BUT if the energy required to maintain it is to low the input voltage would be allowed to rise above the setpoint BUT if the energy required to maintain the output voltage is to high and the input voltage would fall the energy is reduced and the output voltage/current falls.

Well, if the input power is not sufficient to provide the preferred output power, it seems it should be possible mathematically to maintain both the input and output voltages, but adjust the output charging current lower.  I just don't see how the controller would deliberately do that.  Of course since this is a charger, dropping the voltage below the battery voltage would result in no output current at all.  So maybe it just works automatically - the ouput always provides as much current as it can at the present battery voltage, subject to an upper limit.

[fourtytwo42]

So maybe it just works automatically - the ouput always provides as much current as it can at the present battery voltage, subject to an upper limit.

Yes & that upper limit is to maintain the input > VMPP setpoint.

[tifkat]

Does anyone have any further input on my reading of the datasheet, and how I might set the MPPT pin voltage correctly? I am assuming (for now) that the Rated voltage is the Vmp and not Voc, for now. It doesn't matter in the long run, but understanding how to set the MPPT input level is.

I'm working with essentially garden solar lamp panels, but possibly of bigger dimensions. I may not get any discrete Vmp/Voc figures, but trial and error can help there.

[Peabody]

You'll have to get some sample readings of the panels.  You're in summer now, so you can get good readings of the maximum open circuit voltage the panels are capable of.   But then you have to decide at what illumination level MPP tracking is most important to you - full sun, or some lower level.  Then the MPP voltage for divider purposes will be about 80% of the open circuit voltage.

[tifkat]

You'll have to get some sample readings of the panels.  You're in summer now, so you can get good readings of the maximum open circuit voltage the panels are capable of.   But then you have to decide at what illumination level MPP tracking is most important to you - full sun, or some lower level.  Then the MPP voltage for divider purposes will be about 80% of the open circuit voltage.

Cool. But how about my question? Is my understanding of the datasheet correct? Do I need to supply 1.205v on the MPPT pin of the IC, using the resistor divider? The datasheet says the MPPT pin is set at 1.205v. Possible translation issue. Maybe 1.205v internal reference, and MPPT is compared to that, and the PWM modifies the connect time in order to bring the input up/down to 1.205v.

[Peabody]

The bottom of page 6 of the datasheet gives the formula for the resistors.  But I'm not sure the formula makes any sense.

Edit:  Well, it's just a non-inverting amplifier.   But that gives you the same number as a voltage divider between Vmpp and ground, with the midpoint being 1.205V.  So if you had 100K and 33K resistors, the Vmpp would be 4.8565V.  But I still don't understand how it works in opamp terms.


Edit2:  I found this datasheet for a Linear part that appears to have the same MPPC feature:

https://www.analog.com/media/en/technical-documentation/data-sheets/3129fc.pdf

Page 15 shows the circuit.  It's just a comparator with the MPPC pin (+) and the reference voltage (-) as inputs.  But there's a diode on the output, so the comparator has an effect only if MPPC goes below the reference.

[tifkat]

I have found a design published on EasyEDA. It seems to be designed for a 12.05v input, and uses a 1/10 divider, which pretty much confirms the MPPT pin needs to see 1.205v from the divider, which in turn is configured based on what the expected Vmp is for the attached solar panel. This is the confirmation I was looking for.

https://easyeda.com/wagiminator/y-cn3801-lifepo4-solar-charger_copy_copy

Thanks for all your input guys, it was helpful to read datasheets of those alternatives you mentioned.

Now to identify a suitable MOSFET and inductor to pair with the IC.

[tifkat]

For full closure, Let's consider the following:

I assume my dodgey 6v panel characterises to a 4.5v Vmp. So the resistor divider to get 1.205 on the MPPT pin I might choose to use would be 100k and 36.5k. It would be nice to have 36571, but standard values should be close enough.

Pump that into the stated formula:

1.205x(1+(100000/36500))

And you get 4.50.....

[Peabody]

It might be helpful to others reading this in the future if you would provide more information about what you need at the output of the regulator - voltage and current, and whether it is continuous current or there will be sleep time.

The EasyEDA circuit does not have a "load sharing" circuit, aka "power path", that would allow the panel to directly feed the boost converter after the battery is fully charged and charging has shut down.  So even in full sun, the battery will cycle up and down about 5%.  There's also a potential problem with getting proper termination if the load current is large enough that that "charging current" seen by the 3801 never drops below the termination level.  A load sharing circuit can fix all of that, but is somewhat complicated for solar power.  And there is a problem with load sharing if the panel voltage ever goes higher than the boost converter output, which can require using a buck, buck/boost or sepic converter instead.

I've been working on the problem of load sharing for solar, and believe it would require adding a P-channel mosfet, a Schottky diode and a single opamp.  Below is an example circuit.  It shows a linear charger and LDO regulator, but the load sharing part would be the same for any charger and regulator.

Edit:  Thinking about it, I think MPPT wouldn't work with load sharing.  Having a separate path around the charger would mess up MPPT.  So you would have to have a separate MPPT regulator, followed by a charger, followed by another regulator.