Buck or buck-boost for a MPPT charging of the 12V lead-acid ?

[Oleander]

Hello,

I'm going to design a  MPPT charger fed by 30-60W panel that has a 17.5V max voltage power point and a 22V open circuit
voltage. These are typical values for popular solar panels.
The battery will be of a lead acid type : 12V 20-50Ah.

Will the buck solution be enough for this application ?
Or should I use the boost-buck solution in order to better serve the panel's lower voltage range (below 17V in case of weaker sunlight) ?
I'm looking at a ANALOG (former LINEAR) LTC4000-1 & LT3845A pair (buck) or a LTC4000-1 & LTC3789 pair (boost-buck).
Most examples in LTC4000-1's datasheet suggest a buck solution, however.

[fourtytwo42]

It looks to me that the LTC4000 has not been designed with solar in mind. A solar panel is essentially a constant voltage variable current device, connecting a constant voltage output regulator to a solar panel will simply overload the available current causing the solar voltage to drop to the output voltage of the load and produce very little power.

You need a regulator with either MPPT or a regulated input voltage feature, if you are lucky you may find such a device combined with a battery charger (I am sure some other people here will make suggestions).

[Oleander]

Almost all examples in the LTC4000-1 datasheet are for the typical solar panels ...

[fourtytwo42]

Ahh a matter 0f the -1 sorry I missed that, then in answer to your question you only need buck as the 17.5Vmpp is always greater than the 13.5V lead acid float. Do you have output plots for the panels ? If so you could compare them with the MPP regulation curves given in the LTC4000-1 data sheet. It looks to me that this chip regulates it's input voltage to try for a facsimile of Vmpp without actually actively tracking it if you see what I mean.

In any case this chip seems a poor solution as it also seems to require an LT3845 alongside it, I am sure there are more economical single chip solutions out there

[MagicSmoker]

PV panel efficiency declines as ripple current increases, and both the buck and buck-boost draw current in pulses so require significant input capacitance if used with a PV source. If you need to extract maximum possible power output then use a boost or Cuk converter, instead.

[Oleander]

PV panel efficiency declines as ripple current increases, and both the buck and buck-boost draw current in pulses so require significant input capacitance if used with a PV source. If you need to extract maximum possible power output then use a boost or Cuk converter, instead.

Buck draws real current pulses only from its own input caps. When they're properly dimensioned, PV alone will see almost no pulses at all.

In any case this chip seems a poor solution as it also seems to require an LT3845 alongside it, I am sure there are more economical single chip solutions out there

Maybe this pair is not the best solution, but it simulates well in LtSpice and that speeds up the design process.
Can you direct me to a better MPPT IC's ?

[fourtytwo42]

There are plenty, just use an internet search engine! If your tied to LT then try the LT3652 for example, but there are many others both from Analog, TI, Onsemi, ST & others. Something of this simplicity & integration level rarely needs simulation for a successful design as the data sheets & app notes normally provide everything you need as long as you use sensible layout techniques who's problems are not covered by simulation anyway.  Otherwise many people prefer to use a mixed signal MCU and write there own software for better product control (see Microchip & other app notes).

[MagicSmoker]

PV panel efficiency declines as ripple current increases, and both the buck and buck-boost draw current in pulses so require significant input capacitance if used with a PV source. If you need to extract maximum possible power output then use a boost or Cuk converter, instead.

Buck draws real current pulses only from its own input caps. When they're properly dimensioned, PV alone will see almost no pulses at all.

Sure, that is the reason for the input capacitance on a buck or buck-boost converter - so that the pulses of current aren't drawn from the source directly. But you need an infinite amount of input capacitance to integrate the current pulses into flat DC; anything less than that results in some amount of triangular ripple current reflected back to the source.

Hence why I wrote what I wrote.

[ahbushnell]

PV panel efficiency declines as ripple current increases, and both the buck and buck-boost draw current in pulses so require significant input capacitance if used with a PV source. If you need to extract maximum possible power output then use a boost or Cuk converter, instead.

Buck draws real current pulses only from its own input caps. When they're properly dimensioned, PV alone will see almost no pulses at all.

Sure, that is the reason for the input capacitance on a buck or buck-boost converter - so that the pulses of current aren't drawn from the source directly. But you need an infinite amount of input capacitance to integrate the current pulses into flat DC; anything less than that results in some amount of triangular ripple current reflected back to the source.

Hence why I wrote what I wrote.

The ripple is there on any switching power supply to some level.

[MagicSmoker]

The ripple is there on any switching power supply to some level.

No argument there, but boost-derived converters (including the Cuk and SEPIC) have an inductor in between the source and the switch so will draw current that is already integrated into triangular ripple which makes it much easier to filter to the point of non-existence with an input capacitor. To achieve the same level of ripple reduction in a buck or buck-boost converter either needs a much larger amount of capacitance (with much lower ESR) or else the use of an LC filter on the input (that is, source - series inductor - shunt capacitor - switch).

But again, minimizing the ripple reflected back to the source is only important if the OP is trying to extract maximum power from a PV array, but that is kind of implied by wanting to do Maximum Power Point Tracking, isn't it?

[Oleander]

I've just simulated a LT3845A buck alone with Vin=17.5V , Vout=14.8V , Iout=7.5Amp and a pi filter at the input with a 1.5uH inductor in series.
Filtering caps used are: 10uF 7mohm at the PV side and a 50uF 2mohm at the buck input.
Switching frequency is 300kHz.
The current ripple seen from PV is below 20mA, thanks to the filter used.
Quite nice, I think. Just mathematics.

[Seekonk]

A 60W 12V panel isn't worth fooling around with anything exotic.  Use this basic type of circuit with a buck converter. I do this on a 500W array. Just set to output voltage you want. Isolate with diode so it won't draw at night. This sets minimal panel voltage which quite frankly can't spiral down too far without it. I know, this will finally end up being the Manhattan Project.  Have fun and learn something. 

[David Hess]

The available power is low under low sunlight conditions so not much is lost by using a buck converter and it is easier to achieve high efficiency with a buck converter at high power making up for any loss.

[Oleander]

A little off topic - is it reasonable to parallel 3 typical PV (17.5V MPP) panels to get a 3 times more power from them ?
There must be 3 diodes on each output, I know, but the MPPT charger/converter will be single only.

[Seekonk]

Panels are a current source. Current is the only thing that varies with sunlight.  Using a boost converter is totally wrong in low sunlight.  The power point voltage is the same. Control must be used to vary the load to match the panels.

Panels can be paralleled for more current and the same voltage.  Parallel can work slightly better in shading conditions.  To be proper, fuses should be installed if more than two are in parallel. That is because two panels can supply more current than one shorted panel can take.  Diodes are not needed. Panel voltages should be within 10% of each other.

The panels are just diodes in series. Reverse of what you probably think.  Have snow on your panels?  Feed voltage back till you exceed Vf.  The panels will start conducting and heat up just like diodes and will melt snow. This happens at a bit more than the panel can ever produce.

[David Hess]

A separate MPPT charge controller for each panel series will deliver more power when panels are shadowed but other than that, there is no reason they cannot be used in parallel.  Series operation is more common because it lowers the copper requirements.

[MagicSmoker]

...Using a boost converter is totally wrong in low sunlight.

Why is that? I'll admit I don't have deep experience with PV panels, but it seems to me that the higher the ripple current fraction reflected from a downstream converter the more likely the panel will be pulled below its maximum short circuit current for the intensity of sunlight it is receiving at that time, at which point its output voltage collapses until the converter is cycled off.

[NiHaoMike]

Any converter topology can be used for MPPT. The key is in the controls. One way to do that is to tap into the compensation network of a standard PWM controller and allow a microcontroller to limit the duty cycle. As long as the output voltage is below the setpoint, the duty cycle would try to go to the maximum supported by the controller, which the microcontroller holds back to do MPPT.

What I would like to see is a MOSFET gate driver designed to be driven from the PWM output of any microcontroller, with adjustable built in cycle by cycle voltage and current limiting. Built in signal conditioning would be nice to reduce the parts needed to allow the microcontroller to measure voltages and currents.

BTW, if you take a look at a buck-boost topology, you'll note that by keeping the output high side MOSFET on, the circuit degenerates to a buck topology. That MOSFET would still be needed to prevent a parasitic discharge path in a buck topology, so the added cost really is just one more MOSFET (which can be smaller than the others as it only comes into play at low light) and the control circuit to drive it.

[Siwastaja]

Why is that? I'll admit I don't have deep experience with PV panels, but it seems to me that the higher the ripple current fraction reflected from a downstream converter the more likely the panel will be pulled below its maximum short circuit current for the intensity of sunlight it is receiving at that time, at which point its output voltage collapses until the converter is cycled off.

I think you are overly fixating to the ripple current. Of course, a buck requires properly dimensioned input capacitors, but that's the basic design flow in any case (see any design appnote, and they talk about input capacitors a lot.)

Boost topology is equivalent to the buck, just reversed, and moves the exact same discontinuous ripple current problem to the output side. You need the low-ESR capacitance there, and the total cost is in the same ballpark.

I have hard time believing that a small amount of residual ripple current in a properly bypassed buck converter would play any significant role in panel efficiency. Local low-ESR capacitance is needed for EMI anyway.

[MagicSmoker]

I think you are overly fixating to the ripple current.

Yeah, it probably does seem that way now that you mentioned it. My perspective was more that if you are charging a battery from a PV panel, the battery doesn't care about ripple current so much (it effectively acts like a massive capacitor, in fact) whereas the PV panel does, at least to some extent, and so the "ideal" converter to use would be one which intrinsically reflects a low amount of current ripple at its input. So, really more a case of what topology is best, rather than only one topology works and everything else sucks.

I have hard time believing that a small amount of residual ripple current in a properly bypassed buck converter would play any significant role in panel efficiency. Local low-ESR capacitance is needed for EMI anyway.

I thought the same, but there are numerous papers on it (even worse at tolerating ripple are fuel cells, and it was searching for info on that I came across papers discussing the same for PV panels).

E.g. - https://www.researchgate.net/publication/257415332_Analysis_of_the_effects_of_inverter_ripple_current_on_a_photovoltaic_power_system_by_using_an_AC_impedance_model_of_the_solar_cell

[Seekonk]

...Using a boost converter is totally wrong in low sunlight.

Why is that? I'll admit I don't have deep experience with PV panels, but it seems to me that the higher the ripple current fraction reflected from a downstream converter the more likely the panel will be pulled below its maximum short circuit current for the intensity of sunlight it is receiving at that time, at which point its output voltage collapses until the converter is cycled off.
[/quote]

I admit I just read that quick, but I have no idea what you are talking about.  Maximum power is at power point voltage.  Power point voltage only changes with temperature. Even then ot doesn't change much.  Panel voltage drops as you reach the short circuit current for that light intensity. A boost converter can not boost your way out of a low voltage condition below power point.  Panel voltage drops in low light when it can't even provide enough power to the electronics it is connected to, like just a meter.  Boost is only applicable if tanking a 12V panel and boosting it to a 24V battery while controlling power point voltage.

[MagicSmoker]

...
I admit I just read that quick, but I have no idea what you are talking about.

Well, the messages are all right above this, but to recap, I suggested a boost-derived converter (e.g. - which includes the Cuk and coupled-inductor SEPIC) merely on the basis of it reflecting less ripple back to the PV source than a buck or buck-boost derived converter (e.g. - a forward converter is buck derived, an isolated flyback converter is buck-boost derived, etc.).

...A boost converter can not boost your way out of a low voltage condition below power point. ...

No one - least of all me - argued otherwise. I was merely saying the if the converter reflects excessive ripple back to the PV panel the ripple crests can exceed the maximum power point current (which is a little bit less than the short-circuit current, as I understand it) and that could inadvertently drag the PV voltage over the cliff, so to speak. That is to say, attempting to draw more current than the maximum power point can deliver will cause the PV panel voltage to rapidly decline, which causes the switchmode converter to draw more current from it still, etc.

But the OP managed to get reflected ripple from his buck design down to 20mA or so using a Pi filter so this is rather a moot argument.

[Siwastaja]

I don't understand the problem.

With a buck, you must choose PV and battery voltage such that PV optimal power point V > battery V, in all conditions.
With a boost, it must be PV optimal power point V < battery V, in all conditions.

Boost may need an extra switch to isolate the panel so it doesn't freewheel through the diode. Buck may need an extra switch to prevent energy flowing back to the panel from the battery. (Either condition isn't written in stone. May or may not need, depending on exact conditions.)

If you aim for 50% duty cycle with either type of converter, calculated from nominal panel voltage and nominal battery voltage, so that for buck, Vbat is 50% of VPV, and for boost, Vbat is 200% of VPV, you have a lot of leeway in the power point adjustment feedback over temperature, light level and battery SoC variations.

For the case where VPV and Vbat must be close to each other, a buck-boost, Cuk or SEPIC would be required.

But I don't see any issue with any of these choices. You must choose something. Low PV voltages have an advantage in case of partial occlusion, but higher currents if you cannot place the converters on the roof. High battery voltages increase BMS cost somewhat, but OTOH may make power conversion to mains AC easier, if needed. And there are all kinds of conflicting sweet spots. Actual best topology would then depend on many conditions, but there's nothing fundamentally inferior in any of them IMHO.