Need help designing load testing methodology for MPPT controller

[cbc02009]

Hello all,

For our senior capstone project, my group is designing a solar tracker / mppt controller. We've got the mppt controller finished and working, and we're nearly done with the solar tracking part. The last part is designing a scientifically rigorous test to compare the solar tracker to a stationary panel and compare the differences in power the panel generates.

So how, in general does one test power transfer over a long period of time?

I have some ideas of how to do it, like seeing how long it takes to fully charge a 12ah lead acid battery, but I'm worried the results of that wouldn't be very consistent, or that the battery charging wouldn't take long enough to be representative of a full day's worth of sunlight.

I could also attach the output to a resistive load, and use another MCU to measure current through a shunt resistor, and voltage across the load, and take the average (or the integral) to find either average power or total energy delivered. I worry though, that with the resistive load, that the energy (and therefore effeciency) output of the MPPT will depend on the value of the resistor I select.

Let me see if I can explain better. I have a 50W solar panel with a maximum power point at V = 17.5V and I = 2.86A. Say that the MPPT algorithm runs its calculations, and decides that 50% duty cycle will cause the solar panel to reach that number. The output voltage will be 17.5V * 50% = 8.75V. If we assume 80% efficiency, the output current that creates the maximum power will be (50W * 80%)/8.75V = 4.5A, which gives us an optimal load resistance of 8.75/4.5 = 1.99 Ohms. Cool that's what we use.

But say the sun goes partially behind a cloud (or something), and the MPPT redoes it's calculation, and decides it now needs 70% duty cycle to make the panel operate at it's MPPT. Our new output voltage is 17.5V*70% =14V. At 14V, the 1.99 ohm resistor from our last calculation would want to draw over 7A through it, but the maximum current we could get at the output (again assuming 80% efficiency) is (50W*80%)/14 = 2.85A. Since the system is obviously current controlled at that point, the actual voltage across the load would be 1.99*2.85A = 5.68V which gives us an output power of 5.68V*2.85A = 16.81 W out of a 50W panel just because of the value of the resistor I chose.


My last thought was to use something like a lab bench power supply as either a constant current or constant voltage load (which I don't actually know if that's a thing that's safe to do. Can those supplies sink current, or only source it? The word power "supply" kind of hints at no.) and then again use a micro to measure current and voltage. I found this https://smile.amazon.com/MakerHawk-Adjustable-Electronic-Intelligent-Resistance/dp/B07F3NHHST/ on amazon, maybe something like that? I could probably design one and build it for cheaper.

If I could somehow design and build a constant current or voltage source, which would be better? I know that solar panels tend to operate as current sources, but the output of the MPPT could be either constant current or constant voltage.

Or am I just thinking myself in circles and confusing things?

Any advice would be greatly appreciated.

[Janne]

Why not just attach a dump load to the battery, for example just divert power to dummy resistors when the battery reaches rated charging voltage. Then get a pair of DC energy meters, one for the tracking panel installment and one for the stationary panel one.

[fourtytwo42]

I think it would be helpful if you posted a schematic of this "mppt" controllers power circuit and told us what the load was designed to be. From your description this sounds like a simple pulse width modulator that would in reality not be able to provide an mppt function for any stable load such as a fixed voltage variable current or constant resistance. So your problem in testing appears to be you have designed an "mppt" controller with no concept of what the load actually is.

[cbc02009]

I think it would be helpful if you posted a schematic of this "mppt" controllers power circuit and told us what the load was designed to be. From your description this sounds like a simple pulse width modulator that would in reality not be able to provide an mppt function for any stable load such as a fixed voltage variable current or constant resistance. So your problem in testing appears to be you have designed an "mppt" controller with no concept of what the load actually is.

It is a full MPPT tracker running the incremental conductance algorithm on an atmel SAML21. The mppt algorithm varies a reference voltage that is fed into a PI controller that adjusts the PWM input of a synchronous buck converter until the desired voltage is achieved. I've attached schematic pictures to this post.

The load is meant to be 12V sealed lead acid batteries, but as I stated, I'm worried about reproducible results, so I was hoping for some ideas for a slightly more rigorous test method.

So no, the MPPT wasn't designed without a load in mind, but yes, I'm not sure the load is what I would like to use for scientifically rigorous testing for the capstone report we will have to write up.

[cbc02009]

Why not just attach a dump load to the battery, for example just divert power to dummy resistors when the battery reaches rated charging voltage. Then get a pair of DC energy meters, one for the tracking panel installment and one for the stationary panel one.

That's an interesting idea. Thank you.

[jbb]

Another approach is to combine a DC supply and a resistor bank to improvise a constant voltage load.

Set the DC supply to desired voltage (eg 13.8V). Choose a resistor bank that will draw a bit more than the expected maximum solar output and put it in parallel. Now the DC supply will keep the voltage constant, the resistor bank will dissipate some energy from the power supply and some from the device under test.

It’s a pain, but I suggest you try testing 3 devices: one with no MPPT and no tracking, one with MPPT only and one with both MPPT and tracking.

[Seekonk]

Add some software to prevent he charge controller from going into various charge modes,.  hose modes will prevent you from getting any real numbers.

[fourtytwo42]

It is a full MPPT tracker running the incremental conductance algorithm on an atmel SAML21. The mppt algorithm varies a reference voltage that is fed into a PI controller that adjusts the PWM input of a synchronous buck converter until the desired voltage is achieved. I've attached schematic pictures to this post.
The load is meant to be 12V sealed lead acid batteries, but as I stated, I'm worried about reproducible results, so I was hoping for some ideas for a slightly more rigorous test method.

Ahh now I understand its a synchronous BUCK converter!! So going back to your original question you can indeed use a fixed resistor providing when absorbing the maximum panel power the terminal voltage does not exceed Vmpp-losses*max duty cycle. Conversely the minimum power it can absorb would be Vmpp-losses*min duty cycle.
As for power measurement you already measure panel voltage and current for the mppt algorithmn so simply export that measurement to an external device for integration/logging.
I think in your original question you have directly related duty cycle to voltage, this is not the case for a buck converter, only for a simple pwm with no low pass filter, I hope this makes sense

[cbc02009]

It is a full MPPT tracker running the incremental conductance algorithm on an atmel SAML21. The mppt algorithm varies a reference voltage that is fed into a PI controller that adjusts the PWM input of a synchronous buck converter until the desired voltage is achieved. I've attached schematic pictures to this post.
The load is meant to be 12V sealed lead acid batteries, but as I stated, I'm worried about reproducible results, so I was hoping for some ideas for a slightly more rigorous test method.

Ahh now I understand its a synchronous BUCK converter!! So going back to your original question you can indeed use a fixed resistor providing when absorbing the maximum panel power the terminal voltage does not exceed Vmpp-losses*max duty cycle. Conversely the minimum power it can absorb would be Vmpp-losses*min duty cycle.
As for power measurement you already measure panel voltage and current for the mppt algorithmn so simply export that measurement to an external device for integration/logging.
I think in your original question you have directly related duty cycle to voltage, this is not the case for a buck converter, only for a simple pwm with no low pass filter, I hope this makes sense

Thank you! Sorry for the confusion. Your suggestion is incredibly helpful. I will have to redo my calculations.

Another approach is to combine a DC supply and a resistor bank to improvise a constant voltage load.

Set the DC supply to desired voltage (eg 13.8V). Choose a resistor bank that will draw a bit more than the expected maximum solar output and put it in parallel. Now the DC supply will keep the voltage constant, the resistor bank will dissipate some energy from the power supply and some from the device under test.

It’s a pain, but I suggest you try testing 3 devices: one with no MPPT and no tracking, one with MPPT only and one with both MPPT and tracking.

I like this idea. thank you.

[max_torque]

surely a practical MPPT controller needs to be a two stage device?

the "Input" stage keeps the panel at the optimum voltage and the "output" stage keeps the output at the set voltage (say 13v for a LA battery charger, or 240Vac for a single phase ac output etc)

Just keeping the panel at it's optimum voltage seems pointless?

[cbc02009]

surely a practical MPPT controller needs to be a two stage device?

the "Input" stage keeps the panel at the optimum voltage and the "output" stage keeps the output at the set voltage (say 13v for a LA battery charger, or 240Vac for a single phase ac output etc)

Just keeping the panel at it's optimum voltage seems pointless?

Optimally it would. Because this is a school project, I was looking for ways to cut costs. I came across this application note from microchip (http://ww1.microchip.com/downloads/en/AppNotes/00001521A.pdf) that suggested a single stage device controlled by a state machine that would alternate what it was tracking (panel power or battery current/voltage) depending on what was needed. So far it seems to work pretty well, but if I were to start from scratch I would probably just add a second stage to the system.

[Hydron]

surely a practical MPPT controller needs to be a two stage device?

the "Input" stage keeps the panel at the optimum voltage and the "output" stage keeps the output at the set voltage (say 13v for a LA battery charger, or 240Vac for a single phase ac output etc)

Just keeping the panel at it's optimum voltage seems pointless?

Actually you only need a single stage as long as the input voltage (to a buck MPPT) from the panel is always higher than the output to the battery. The MPPT will adjust the duty cycle until the maximum power transfer occurs; the output voltage is basically irrelevant other than for overcharge protection etc.

[max_torque]

surely a practical MPPT controller needs to be a two stage device?

the "Input" stage keeps the panel at the optimum voltage and the "output" stage keeps the output at the set voltage (say 13v for a LA battery charger, or 240Vac for a single phase ac output etc)

Just keeping the panel at it's optimum voltage seems pointless?

Optimally it would. Because this is a school project, I was looking for ways to cut costs. I came across this application note from microchip (http://ww1.microchip.com/downloads/en/AppNotes/00001521A.pdf) that suggested a single stage device controlled by a state machine that would alternate what it was tracking (panel power or battery current/voltage) depending on what was needed. So far it seems to work pretty well, but if I were to start from scratch I would probably just add a second stage to the system.

For our non-mppt panel, assuming a purely resistive load, then more current = more voltage, so you can just use a fixed resistor of a suitable rating and integrate the total power that comes out (measure V & I to calc power, integrate to energy over time) The choice of resistor will effect your results, as if you chose it to have a resistance that gives an input voltage close to mppt under your average conditions then you will get a greater total power transfer (chose it to be optimum for conditions you never see, and it'll be really bad!)


For your mppt controller, assuming you have a "current controlled" architecture (ie there is an inductor in the system that stores charge) then your output resistance, as long as it is low enough (ie  sufficiently below the minimum panel optimum voltage at max system current) shouldn't matter.

ie

target panel voltage is always 17.5v, and system current is modulated to keep to that target voltage as solar intensity varies, then the duty cycle of the buck will be modulated to get to that target panel current:

 When solar intensity is low, and current is low, the voltage across the fixed load resistance is low (V=IR), so there is a larger Vdrop, and switch on time is short.

 When intensity is high, and current is high, the voltage across the fixed load resistance is high, so there is a smaller Vdrop, and switch on time is longer

As long as the load resistance is chosen to be suitable to maintain the duty cycle within the system limits then it should work. And you can just measure and integrate output V & I in the same way as for the non-mppt panel to get a comparative performance figure

It is the inductor, which acts to limit the rate of change of current, that allows the system to move energy between systems of different potential at a reasonably constant current (ie low current ripple)

[olsenn1]

I have two folding solar chargers:

        1. Instapark Mercury 10 -- 10 Watt dedicated USB charger
        2. Powerfilm F12-750 -- 12 Watt 15.4V charger

When I connect the Mercury 10 to my electronic load at CV (5V) I get 1.0 Amp of current (5W) and this is the most I can reliably push out of this. However, when I connect my Powerfilm charger via a switching buck converter (8-24V to 5V USB) to the same e-load under the same lighting conditions, I get 0.5 Amp (2.5W) at CV 5V. This is in line with what my devices charge at when I plug them into this USB port. However, I can switch the e-load to CC 1 Amp and get the same 1A@5V output of this solar panel this way.

Obviously my solar panel is capable of outputting more power than it is. Does anyone know what I could do to fix this? Would a MPPT charge controller do the trick (or is this just for ensuring against overcharge of lead-acid batteries)?

Thanks!