Words of Wisdom: “The response of a series parallel array to shade will be complicated.” – sensij1)
Partial shading can have surprisingly dramatic effects on panel output. Perhaps counterintuivitely, partial shading can have more devastating effects on output than full shading like heavy overcast or evenly shaded forest canopies.
The way to avoid the problems associated with partial shade is to avoid partial shade. Failing that, there are steps one can take to minimize the losses:
There is another approach, which is to bring the shaded cells/strings back online by bringing the rest of the panel down to their level. It sounds counterproductive, but with MPPT controllers and in some higher-voltage series configurations (say Vmp is >=3x bank voltage) it works.
This occurs because the MPPT has a broader range of voltages to sweep and can find other power peaks (panel voltages) that are low enough to bring the shaded cells back online but still high enough to charge the battery bank. It's not reality, but we can think of it as MPPT evenly “shading” the entire panel voltage-wise in order to get max juice from it in partial shade conditions.2)
[draft section]
This section will assume an array of 2x generic 100w 18Vmp panels
For reasons discussed below the string of 36 cells will be subdivided into substrings of cells in series. 2 strings of 18 cells is common3). So we can think of this panel
######### ######### ######### #########
straightened out but electrically identical
################## ##################
Under lab conditions the cells will each make 5.55A at 0.5v, delivering the rated 100w at 18v. In the real world the Vmp will likely be lower due to cell temperature derating and current will likely be much lower due to imperfect sun. But to keep things simple and numbers even we will assume lab conditions.
When the entire panel is lit or shaded evenly the harvest is reduced but no individual cell is running at a lower voltage than its neighbor. So overcast days or dense tree canopies are not a problem. And when people talk about “shade” problems it usually means “partial shade”.
We can't stop the cell voltage from collapsing in shade so we need a way to remove it, electrically, from the circuit. This is typically done with diodes. They are, in effect, electrical valves that allow power to flow in only the intended direction.
Right now we are concerned with the interatctions of panel internals so we are talking about bypass diodes. They can bypass a voltage-collapsed cell or substring and avoids the overheating problem.
Simplest example: we have a string of three cells in series. In full sun all three contribute 0.5v and the total string voltage is 1.5v. In partial shade one of them drops top 0.2v. It gets bypassed so this string is now 1.0v. Without the diode the string would be 1.2v (0.5v + 0.5v + 0.2v) until the shaded cell failed from overheating.
But diodes for each cell would be more expensive in components, assembly, and design. The compromise most manufacturers make is to have a bypass diode at the substring level. Our panel is now like this:
= = ################## ##################
…where the equal sign (“=”) is a bypass path when needed to prevent damage.
Real-world example: we have two substrings, each with 18 cells ins series. In full sun they will make 18v (2s substrings x 18 cells x 0.5v).
In partial shade one of the cells in the 2nd substring are affected.
= = ################## ####0###00########
If their were no diodes we would have 9v from the 1st string and 7.5v from the 2nd string for a total of 16.5v. But the shaded cells in the 2nd string would overheat. So it is bypassed and now the panel makes 9v.
Oops: 9v is not high enough to charge a 12v house bank. We get nothing.
Thought experiment: what if we had 3 substrings of 12 cells but the same cells were shaded? We'd lose one substring and still have 2 6v substrings for 12.0v. That might be enough to charge a 12v bank that was really low. If we had 4 substrings of 9 cells and lost one substring we'd have 3 4.5v substrings for 13.5v. That might be high enough to be actually useful.
Let's start using both our panels in parallel
######### ######### ######### ######### ######### ######### ######### #########
which, unwound and with bypass diodes, would be
= = ################## ################## = = ################## ##################
if we have the same partial shading as before
= = ################## ################## = = ################## ####0###00########
To prevent voltage backing up into a panel from a neighboring panel (or battery bank at night) each panel typically has a blocking diode. So the complete picture is:
= = ################## ################## = = = ################## ####0###00######## =
Note: in most panels the blocking and bypass diodes are identical parts, just installed in different places, in different orientations, and for different purposes.
= = = = ################## ################## = ################## ####0###00######## =