Installing serial solar panels

[ramblinrod]

Originally Posted by ramblinrod

I concur 100%.

Whoops, forgot to spoke the rite langwij..

As I be'ed sayin' and a whole bunch of suppository engine ears has be'ed complainin' 'bout. ;-)

Canajun Teknishun

Translation error, "complain' aboot".

Sorry aboot that, eh?

[OceanSeaSpray]

Originally Posted by ramblinrod

...Rather than trying to deflect attention away from your prior errors and attempting to argue that a solar panel is substantially different from a battery or generator or other power producing device...

The proponent was correct and you are wrong, a solar panel is indeed "substantially different", it is a current source regardless of how it is possible to represent it electrically using a combination of ideal elements.

Practical demonstration: it can be short-circuited indefinitely and no damage or overload occurs, unlike a battery or a generator. That is a "substantially different" outcome.

When it comes to learning something from you, well... haha, please do me a favour and hold your breath.

[Paul Elliott]

[NOTE - In all these plots, the green trace is the load current, and the lighter trace is load power.]

I said I would do some simulations, and here are my results. Use them as you please -- I'm doing this for my own amusement, and perhaps to learn something.

First, a SPICE simulation model for a single solar cell, tweaked to roughly approximate the performance of the cells in my 100W 36-cell panels:

I used a voltage-controlled current source so I could set the cell current to simulate degrees of shading. The control voltage is set during simulation, with 1A of cell current for 1V of control voltage.

I then built an 18-cell string with these cells. Two of these strings make up one of my "12V" panels:


Then, I built two test fixtures. One has two 100W panels in parallel, and the other has two panels in series. I added Schottky bypass diodes around each string. These fixtures have per-string control voltages, and a load consisting of a voltage-controlled resistor. During the simulation, this resistor varies from 0.1 Ohm to 1000 Ohms. I plot the load current and power vs load voltage to show the MPPT point (or points) for the configurations:

Parallel Panels


Series Panels

Here is the load-line for the parallel panel connection, with no shading:


We're not quite getting 100W, but the characteristics are close enough for my purposes.

Here is the unshaded series-connected configuration:

Twice the voltage, half the current, same power.

Now I heavily shade one string of one panel, to where it can only deliver 1A max, and the unshaded strings can deliver 6A.


Parallel: MPP = 106W


Series: MPP = 131W

In this heavy shadow case, the series connection is clearly superior (from an available power perspective) than the parallel connection.

Now, I tried a lightly-shaded scenario, where one string can only deliver 5A, and the unshaded strings can deliver 6A.


Parallel: 170W


Series: 165W

Here, parallel slightly beats series.

One more: half-shaded string at 3A, the rest at 6A


Parallel: 139W


Series: 132W, or is it 109W?

Here we see two local MPP maxima. Which one will the MPPT controller find?

The permutations are quite interesting, and it's hard to tell if there really *is* an optimum configuration. What is obvious is that this all depends on shading, and wiring preferences. Low-loss per-panel MPPT controllers probably will be better, but is the difference enough to be worth it? Again, that depends...

[TeddyDiver]

Originally Posted by Paul Elliott

The permutations are quite interesting, and it's hard to tell if there really *is* an optimum configuration. What is obvious is that this all depends on shading, and wiring preferences. Low-loss per-panel MPPT controllers probably will be better, but is the difference enough to be worth it? Again, that depends...

Thanks excellent information.
One has to consider what kind of shading is typical in their installation. Me thinks the most common being shading of the mast or boom, and how the height of the sun affects the shading. Another thought comes to mind is the orientation of the panels, most installations I've seen have the panels transversally which means ones mast can cast a shade over every single panel.

BR Teddy

[Paul Elliott]

Originally Posted by TeddyDiver

Another thought comes to mind is the orientation of the panels, most installations I've seen have the panels transversally which means ones mast can cast a shade over every single panel.

That's why my three top-of-the-dodger panels are oriented long-dimensions fore and aft (three across). That way the boom and mast shadows are less likely to take out all three panels simultaneously. Best way to use a bad location.

[ramblinrod]

Excellent analysis Paul.

I don't think the results will solve the "Which is better debate" and there is always concern that simulation doesn't represent real world, due to some influencing condition not considered, but it sure is nice to see some numbers in an apples to apples comparison.

Thanks for the effort and sharing.

Have you tried running the analysis with a PWM controller?

[ramblinrod]

Originally Posted by OceanSeaSpray

The proponent was correct and you are wrong, a solar panel is indeed "substantially different", it is a current source regardless of how it is possible to represent it electrically using a combination of ideal elements.

Practical demonstration: it can be short-circuited indefinitely and no damage or overload occurs, unlike a battery or a generator. That is a "substantially different" outcome.

When it comes to learning something from you, well... haha, please do me a favour and hold your breath.

Well, the experts have weighed in; sorry to inform you that I was correct using the term "Voltage Source", especially when the original point where it was introduced was not limited to solar panels.

Class dismissed.

[Paul Elliott]

Originally Posted by ramblinrod

Excellent analysis Paul.

I don't think the results will solve the "Which is better debate" and there is always concern that simulation doesn't represent real world, due to some influencing condition not considered, but it sure is nice to see some numbers in an apples to apples comparison.

Thanks for the effort and sharing.

Have you tried running the analysis with a PWM controller?

In a way, my simulations are controller-agnostic. We can look at the load lines and find the MPP, where we hope an actual MPPT controller would operate (although in some of the cases I suspect the controller would stop at the first MPP it found -- possibly not the best one.)

If I'm not mistaken, a PWM controller is going to run at 100% duty-cycle while in bulk-charge mode, and we can look at the solar load lines to see what current is available at various battery voltages. I'm not going to attempt to accurately simulate a charging battery, but we can assume the battery is at (say) a 12.5V average during bulk charge, and compare the amps delivered by a an MPPT controller vs a PWM at 100%.

[Zach]

Question for the solar guru's that focus on current...

Any thoughts on a voltage sensitive relay to bypass a charge controller and clamp direct to the battery in shaded places? If you are making 5v with no connections made at the panel, and not enough to turn on an MPPT, perhaps... Lock over a relay and bypass the MPPT.

My understanding is that since two items on the same circuit can't be different voltage, that if one was to connect a solar panel directly to a battery the voltage of the solar panel at its connection would be that of the battery and any current being produced from the panel would go to charge the battery.

Is this an accurate read? I'm thinking the output still wouldn't amount to spit in a bucket at low voltage off a shaded panel, but spit still beats a parched mouth.

[ramblinrod]

Originally Posted by Zach

Question for the solar guru's that focus on current...

Any thoughts on a voltage sensitive relay to bypass a charge controller and clamp direct to the battery in shaded places? If you are making 5v with no connections made at the panel, and not enough to turn on an MPPT, perhaps... Lock over a relay and bypass the MPPT.

My understanding is that since two items on the same circuit can't be different voltage, that if one was to connect a solar panel directly to a battery the voltage of the solar panel at its connection would be that of the battery and any current being produced from the panel would go to charge the battery.

Is this an accurate read? I'm thinking the output still wouldn't amount to spit in a bucket at low voltage off a shaded panel, but spit still beats a parched mouth.

Per prior discussion, when the parallel connected shaded panel is "reverse biased" by the unshaded panel, ie, prior to load application, the shaded panel open circuit voltage is pulled up to the unshaded panel open circuit voltage, no current can flow through the shaded panel. In simplest (teknishun) terms, electrical current can't try to flow in opposite directions at the same time.

All of this stuff adds complexity and potential failure nodes for little gain. Better to avoid sharing, add more quantity and efficient panels, use MPPT instead of PWM.

[macman4190]

Originally Posted by Paul Elliott

That's why my three top-of-the-dodger panels are oriented long-dimensions fore and aft (three across). That way the boom and mast shadows are less likely to take out all three panels simultaneously. Best way to use a bad location.

I also installed my panels in this orientation for those same reasons. All the YouTube videos I have seen demonstrate that if there will be some shading on the panels, installing in parallel provides better output than installing in serial. For example, the cells in the panels are linked in series, and if you shade just one cell, it take out the entire panel. If two of my three panels are completely shaded late in the day, but one still has full sun (at a low angle), I have enough amps to power my refrigerator. I also noticed that shading kills the amps much more than it kills the voltage. So the shaded panel does not pull the volts down on the unshaded panels. But it also does not contribute any amps to the input.

[Paul Elliott]

Originally Posted by macman4190

I also installed my panels in this orientation for those same reasons. All the YouTube videos I have seen demonstrate that if there will be some shading on the panels, installing in parallel provides better output than installing in serial. For example, the cells in the panels are linked in series, and if you shade just one cell, it take out the entire panel. If two of my three panels are completely shaded late in the day, but one still has full sun (at a low angle), I have enough amps to power my refrigerator. I also noticed that shading kills the amps much more than it kills the voltage. So the shaded panel does not pull the volts down on the unshaded panels. But it also does not contribute any amps to the input.

My tests and simulations (and experience of many other people) shows that a serial-connected panel system can in some cases give better results than a parallel-panel setup, even if shadows are hitting one or more of the panels. The panels need to have bypass diodes for this to work well. Without the diodes, a shaded panel will definitely kill the current through the entire serial string.

My simulations also show that in other shading cases, the parallel panels will do better.

That's why I posted the simulation results -- I wanted to compare serial vs parallel and see how each configuration fared with various shading scenarios.

[hellosailor]

Zack-
If you are only making 5V it won't be any use to the batteries. In fact the MPPT controller only has something like a 2-4% loss internally, so practically speaking it should still be able to charge the batteries BETTER than the panel alone would do the job, even at "12" volts, as long as there is some power (volts or amps) available from the panels that can be converted into what the battery needs.
Just one man's opinion.

[s/v Jedi]

The Spice simulation from Paul was excellent. It has been more than a decade ago that I used Spice, I'm too old :-o

On what I saw about different voltage panels in parallel: assuming they have no blocking diodes, the higher voltage panel will start dumping energy into the lower voltage panel as well as into the controller. With the blocking diode, the voltage difference is right there across that diode like somebody already wrote.

On the panels in series where was has a diffused light that reduces output. Current must be equal but the panels are a current source meaning that the current is equal even if one panel only generates 1/3rd of rated power. I believe the example was 5A and both panels will deliver that so array becomes 5A with reduced power output caused by lower voltage. We saw specs where a panel that delivers 5.25A at Vmp still delivers 5A at Vsc, a shorted panel, i.e. zero Volt. That is less than 5% difference which is spectacular for practical implementation of a current source.

When you have a current source, voltage is determined by the impedance of the circuit. In the case of a MPPT controller, the controller sweeps across the impedance range to sense power output from the PV array across that range, then settles right at that impedance for maximum power transfer. If you have many panels, group them according to shading patterns, then connect each group in series to a separate MPPT controller. If the shading is so complex that each panel is a separate group then so be it and go for the small Genasun controllers.

Experimentation is great when you know what you're doing. You must be able to measure current through strings and diodes; must understand how the controller works and when it performs tracking functions, how it changes circuit impedance to move along the curves etc. If you don't understand that, and many don't, then it will end in a waste of time and rules of thumb like I described above should be used instead. If you have few panels, go for the Genasun because in real life situations nothing beats the electrical performance of that setup. It gets costly for larger panel counts, but will still outperform electrically.

[downunder]

Originally Posted by s/v Jedi

The Spice simulation from Paul was excellent. It has been more than a decade ago that I used Spice, I'm too old :-o

On what I saw about different voltage panels in parallel: assuming they have no blocking diodes, the higher voltage panel will start dumping energy into the lower voltage panel as well as into the controller. With the blocking diode, the voltage difference is right there across that diode like somebody already wrote.

On the panels in series where was has a diffused light that reduces output. Current must be equal but the panels are a current source meaning that the current is equal even if one panel only generates 1/3rd of rated power. I believe the example was 5A and both panels will deliver that so array becomes 5A with reduced power output caused by lower voltage. We saw specs where a panel that delivers 5.25A at Vmp still delivers 5A at Vsc, a shorted panel, i.e. zero Volt. That is less than 5% difference which is spectacular for practical implementation of a current source.

When you have a current source, voltage is determined by the impedance of the circuit. In the case of a MPPT controller, the controller sweeps across the impedance range to sense power output from the PV array across that range, then settles right at that impedance for maximum power transfer. If you have many panels, group them according to shading patterns, then connect each group in series to a separate MPPT controller. If the shading is so complex that each panel is a separate group then so be it and go for the small Genasun controllers.

Experimentation is great when you know what you're doing. You must be able to measure current through strings and diodes; must understand how the controller works and when it performs tracking functions, how it changes circuit impedance to move along the curves etc. If you don't understand that, and many don't, then it will end in a waste of time and rules of thumb like I described above should be used instead. If you have few panels, go for the Genasun because in real life situations nothing beats the electrical performance of that setup. It gets costly for larger panel counts, but will still outperform electrically.

Seems a good summary.