A quick and dirty way to implement Solar switching regulator MPPT?

[depot]

I have a small solar-powered side thing where I want to use solar power and supercapacitors and something like LTC3355 to keep a device on for a long time.
When the sun is out, great. But in non-ideal sunlight it's trouble because it draws enough power to apparently drop solar panel voltage and thus power down to almost nothing.

So of course throwing more money at it for a nice MPPT and supercap management is nice, or just a bigger solar panel, fine if it's a hobby thing that I'd only build one of, but of course I'm curious how to "build a bridge that just barely stands up" i.e. a cheaper, simpler system that would still work.

A co-worker suggested that I could use some features of LTC3355 to implement something like MPPT, if only to give it a useful power point in the low-sun daytime case. And honestly, another MPPT thing LTC3130 does basically that with a pin for RUN and another called MPPC. When I try to use LTC3355's PFI input to do something similar though, it seems like the whole regulator shuts down for a longer time, very disappointing.

So what's the difference between the two parts under the hood? It looks like the MPPC and RUN pins are the "quick and dirty" features I'm looking for, is there some external way to implement these? Or, if they're connected to regulator soft start or something else that locks it up, are there things with this kind of feature? Is there anything to my co-worker's suggestion? I do like something like LTC3355 for managing the supercap.

Or maybe I'm overthinking it and a better MPPT is not so expensive and complex and I could still use it.
I was expecting to find more parts for solar + supercap designs in this kind of range for power too, this is strange and maybe I'm missing some perfect device.

[gnuarm]

Rather than supercaps, what about a battery?  One advantage of a battery is that for short time periods a battery is a constant voltage source with small changes in the charging current having little impact on the voltage.  In that case if you maximize current into the battery, you maximize power, both into the battery and out of the solar cells.  So design a battery charger rather than a MPPT and you get the MPPT as a side effect.   

Any MCU can be used to implement this.  You need to sense the current out of the charger circuit and dither the duty cycle of the pass transistor feeding an inductor.  You will constantly change the pulse width slightly.  If the last pulse adjustment results in more current, adjust the next pulse in the same direction again.  If the last pulse results in less current out of the charger change the direction of the adjustment.  This will rapidly seek and track the maximum power point without really trying. 

You will need to monitor the battery voltage and stop charging when it reaches a fully charged voltage or at least change the charger to maintain a fixed current rather than the maximum.  Also, depending on the details of the battery picked, you will want to initially limit the charging current until the voltage reaches the appropriate level.  The MCU can manage all of this easily. 

[depot]

Rather than supercaps, what about a battery? ...

Everything you post is true, thank you, and on top of it there are many lithium battery tailored solutions, but I'm not going to for some dumb reasons:

I've spent most of my career so far doing battery-related things and I feel they're simple.
I want to try supercaps as new (ish) technology and to compare (and squeeze out) performance.
I hear good things about their future and I know some of the farad values get quite big.
Although they have a self-discharge worse than batteries, I want to find out if an overall system life might be longer. Especially, especially if a long time running supercap system will lose capacity and need replacement over time like a battery system does.
In particular I want to try energy harvesting (like solar) with supercaps.

The interesting thing about this little project is that the power reqs are a bit too high for typical tiny energy harvester things. Perhaps that means that a battery is just the most suitable, but I still want to look at it more before calling it.

Maybe is there some way to simulate the battery's natural "side effect" MPPT?

[Peabody]

As I read the LTC3355 datasheet, the input regulator is buck only.  So based on the PFI pin voltage, it shuts down the input regulator, and turns on the ouput boost regulator so the supercapacitor can power the load.

But the MPPC pin on the LTC3130 just limits the current draw so the input voltage doesn't fall below the divider voltage.  I think this is only approximate MPPT, and is based on the idea that the voltage at the MPP doesn't change very much for different levels of illumination.  The MPP current changes a lot, but the MPP voltage doesn't.

It seems the LTC3130 is what you want.  I don't know how you would add the MPP function to a normal switching regulator.

[gnuarm]

Rather than supercaps, what about a battery? ...

Everything you post is true, thank you, and on top of it there are many lithium battery tailored solutions, but I'm not going to for some dumb reasons:

Maybe is there some way to simulate the battery's natural "side effect" MPPT?

Ok, if the supercap is what you want, that's up to you.  I'm not speaking from experience with these circuits, just brainstorming. 

Maybe I'm over thinking this.  The voltage on the supercap will also be stable for large values of capacitance and short periods of time.  So a one stage switcher that dithers the PWM so as to maximize the output current could simply ignore the cap voltage other than to limit it for protection.  I suppose the supercap could work like the battery since even if the voltage is not long term constant, it is short term constant which is all that matters.

[depot]

As I read the LTC3355 datasheet, the input regulator is buck only.  So based on the PFI pin voltage, it shuts down the input regulator, and turns on the ouput boost regulator so the supercapacitor can power the load.

Isn't a buck regulator sufficient for a single or parallel supercap application though?

Maybe there's a test to perform related to this, but from what I've seen so far, a worthwhile power point shouldn't drop voltage so much.

From LTC3130, even though it's capable of boosting the input voltage, I didn't find any useful amount of power below about 4v input. But maybe that's also because the regulator and things wouldn't have enough power to startup at low voltage inputs.

But the MPPC pin on the LTC3130 just limits the current draw so the input voltage doesn't fall below the divider voltage.  I think this is only approximate MPPT, and is based on the idea that the voltage at the MPP doesn't change very much for different levels of illumination.  The MPP current changes a lot, but the MPP voltage doesn't.

It seems the LTC3130 is what you want.  I don't know how you would add the MPP function to a normal switching regulator.

Aww. This is the big question I have now, how does this kind of MPP circuit work to limit current based on input voltage for a switching regulator?

[Peabody]

It seems you would have to be able to reduce the duty cycle of the regulator if the input voltage drops below the MPP voltage.  I don't know of any normal regulator or converter that offers any official way to do that.  But perhaps there's a way to mess with the feedback resistor divider to make that happen.  I've just never seen anything like that done.  Maybe you could feed the output of an opamp through a third resistor and diode into the sense divider midpoint.  Based on its inputs (the input voltage and a reference MPP voltage), it would increase the voltage at the sense point when the input voltage goes low, which would make the switcher reduce the duty cycle.  But that might oscillate.

It's possible to build your own MPPT converter using an Arduino.  But then you would have to add all the other parts - the mosfet(s), the inductor, current sensor, mosfet driver, etc.  I'll attach a circuit I found that shows everything but the Arduino.

Is it possible that a simple linear regulator would work well enough?  Can you provide the specs on your panel, your supercapacitor, and the needed voltage and current for the load?

[gnuarm]

It seems you would have to be able to reduce the duty cycle of the regulator if the input voltage drops below the MPP voltage.  I don't know of any normal regulator or converter that offers any official way to do that.  But perhaps there's a way to mess with the feedback resistor divider to make that happen.  I've just never seen anything like that done.  Maybe you could feed the output of an opamp through a third resistor and diode into the sense divider midpoint.  Based on its inputs (the input voltage and a reference MPP voltage), it would increase the voltage at the sense point when the input voltage goes low, which would make the switcher reduce the duty cycle.  But that might oscillate.

It's possible to build your own MPPT converter using an Arduino.  But then you would have to add all the other parts - the mosfet(s), the inductor, current sensor, mosfet driver, etc.  I'll attach a circuit I found that shows everything but the Arduino.

Is it possible that a simple linear regulator would work well enough?  Can you provide the specs on your panel, your supercapacitor, and the needed voltage and current for the load?

A linear regulator completely defeats the idea of MPPT as it regulates the output voltage by wasting power in resistance.

I think the MCU approach is the best.  Or some sort of programmable logic.  There are chips from Greenpak that have built in logic and analog, I think even including the pass FET.  This should be a breeze for their parts.

[fourtytwo42]

It might be useful to have some actual figures here like what solar panels and load voltage/power you are considering. For small systems where efficiency is not terribly important the T (tracking) part of MPPT is not required, instead the panels are operated at there nominal maximum power point voltage only.
The most flexible arrangement is a buck/boost or sepic converter allowing the panel voltage to be higher or lower than the load and controlling it purely from the panel voltage who's setpoint is nominal MPPV. A buck solution where the panel voltage has to be higher than the load for energy transfer may also be possible according to the situation. Boost only converters have the potential for the load to overload the panel (through boost diode conduction) causing the panel to never achieve MPPV unless the load voltage can be maintained >MPPV (such as a battery). If you want a single chip solution try the TI SN72442 though by the time you add the external components it is hardly simple!

[depot]

Again, this project idea is really flexible. I can try different things to make it work. I have different samples and other things lying around, and I think the cost of a pcb layout is pretty cheap (although slow).

A previous version of this had only LTC3130 and the output voltage was <15v because I had five 10F 3v supercaps in series and then a buck regulator.
I also want to try a system where there's a supercap manager and either parallel caps or one bigger supercap.

I have four solar panels of this kind:
https://www.digikey.tw/en/products/detail/anysolar-ltd/SM531K08L/9990469?s=N4IgTCBcDaIMoFkCsBmAjAaQAwA4AyAtAHIAiIAugL5A
So the "Pmpp" voltage across one is ~4.5v and with four in series I've seen almost 20v with no load in sunlight. When I run some current through it, seems like the "real" peak power point is 18v. But it doesn't reach "peak power point" in cloudy weather or early in the morning. In those cases, I like ~4v and I had the "RUN" point set there so it can trickle in a bit of power.

I've also tinkered around with the solar panels in series or parallel. I think I like the series configuration the most, or at least it makes the most sense to me. If I had a limit to input voltage from some regulator IC, I would have more parallel configurations there. Or if I added more panels for some reason, I would have to parallel them.

My load is something for work, a cellular module that averages half an amp at 3.3v when it's transmitting. It's not always transmitting. 4 of those solar panels seemed to be enough to run continuously on a  sunny day, but that's not as interesting as seeing the system detect lower power modes and stay alive overnight or for a cloudy day.

[fourtytwo42]

The RUN pin is very useful for preventing complete PV voltage collapse and produces a kind of burst mode at low insolations but have you tried the MPPC pin as well to try and get it to regulate it's input voltage around where you expect VMPP to be ? This would prevent it taking excessive current when greater than the run threshold but still at low light conditions. You can only select fixed values for this chip as it has no tracking function so you have to choose where you want peak efficiency then either side of that the efficiency will fall off, so for example you might optimise it for low light knowing that at higher insolations you can afford to loose some efficiency (measuring efficiency as how much power you could have got with perfect PV operating conditions vs what you actually get).

[depot]

Honestly I forget how well MPPC works.

I remember a satisfied feeling about it, because the system could run continuously in bright light, but that's likely because the solar panels give more power than needed.

I think the "dark" trickle charging times are quite interesting because I live in a place with frequent rainy weather. So I think that interacts with "RUN" well, where "MPPC" is intended for finding the real high-power point.


Are there any good sources for using analog parts and a switching controller to implement MPPT?


The next thing I think I'll try is to set up LTC3355 to have two "power points" based on the resistor to set a current limit for "ICHG". Full sunlight and low sunlight power levels, and select between them based on the input voltage. And if solar power is somewhere in between, I don't think I care much if it oscillates a bit, as long as it gets some trickle charging. In the low power level, I would check less frequently than the high power level, so in any uncertain case it's going to be in the low current mode for much longer.