Solar PV Panel Shading Experiments

Shading (and partial shading) of solar panels causes a large loss of output power so I decided to do some tinkering with small panels at ground-level before trying to attempt changes to the panels on the roof.

I obtained two small solar panels. Each output about 8 volts at 85mA in bright sunshine. The two panels were not exactly matched, one having a slightly higher output than the other. These were lashed up on a board and the output fed into a 220 ohn 7W resistor. The figure below shows the basic setup, with one panel feeding the load resistor.

Single Panel

As can be seen, the panel output voltage is 8.56v and 38mA is measured flowing through the load. We can see therefore that the load is limiting the current flow as 8.5/220 = 40mA.

Two Panels in Series

The two panels were then connected in series (in full sunlight), as shown below.

The voltage across the load can be seen to be nearly 15v and the current flowing is 65mA. Again, this is more or less what is expected from Ohm's law: 15/220 = 70mA, so the load is limiting the current.

Two Panels in Series - Partial Shading

Now one panel was shaded. Note that this is shading, not covering-up so the panel should still be producing power. The figure below shows the results.

It is interesting to note that the voltage across the load has dropped to below one-panel's worth (5.6v), and the current flowing has droppped dramitically - down to 24mA. This is far less that the 40mA you would intuitively expect. This means that the shaded panel must be behaving a high-value resistor and restricting the current flow. In this case, that would be: (14.2 - 5.7)/0.024 = 350 Ohms, ie more than the load resistor.

Shading the LH panel gave similar result: Output voltage = 6.0v, Current = 26mA.

Two Panels in Parallel

The previous results have shown that partial shading of series connected panels results in the whole array having bad performance if only one panel is shaded. The output is less than for a single panel on its own.

The next step was to try two parallel connected panels and see if they fare any better. The picture below shows thw two panels now connected in parallel across the same load. In theory, steering diodes should be used to prevent one panel feeding back into the other. However, the panels have built-in series diodes so this was not necessary.

As can be seen, the output voltage is 8.81v and 39mA is flowing. By Ohm's law, we see that this is as expected as 8.81/220 = 40mA so again, the current is limited by the load but the slighly higher voltage is probably because each panel is loaded less heavily. So, for this fixed load, potential power is being 'wasted' but that is not what we are investigating here, it is the effect of partial shading.

Two Panels in Parallel - Partial Shading

One panel is now partially shaded as shown in the picture and the results are quite conclusive.

The output voltage has dropped slightly to 8.41 volts and the current flowing has dropped marginally to 36mA. This tally's with what is expected: 8.41/220 = 38mA so we can conclude that having two panels in parallel means they are virtually unaffected by partial shading. In my roof-mounted array, just splitting the array into two parallel strings should mean I get considerably more power generated in the mornings when one-half of the panels are in shade.

Two Panels in Series - Shunt Diode

Unfortunately, I can't have all the panels in parallel as the inverter needs an input voltage of over 40V to work properly. In fact, something around 80V would be ideal so there is a necessity to use 4 in series - forming two parallel banks. So, although having two paralell banks will solve many of my problems, the problem of series connected partial shading still requires solving.

Fortunately, the solution is a simple one. Placing a diode in paralell with each panel (reverse biased) provides a by-pass route for shaded (high resistance) panels. So the two panels were connected in series again, this time with a shunt diode across the shaded one and the output measured.

As can be see from the picture, the voltage across the load is 7.53v and the current flowing is 32 mA when one panel is shaded. Again, this is more or less as expected (7.5/220 = 34mA). The voltage drop at the output is due to the diode drop (8.2 - 7.5 = 0.7v) but the shaded panel is now no longer knobbling the illuminated panel.

Conclusions

These experiments have convinced me that in order to improve the performance of my grid-tied solar array, the following steps need taking:

• Split the series connected array into two parallel arrays of 4 panels each.
• Place a shunt diode across each panel.

Of these solutions, the first should be straighforward to implement as it involves some minimal rewiring on the roof and a junction box in the garage (There are 4 wires already routed from the roof and into the garage so no new cable runs are required). I believe splitting the array into two parallel arrays will improve output from 70W to approx 200W for the 2 hour period of the day (April) when one set of panels is not fully illuminated but the other is.

The second step will require each panel to be removed and a diode fitted into its junction box. This will require resealing/weatherproofing so may prove to an arduous task. If this gives large benefits when all panels are not illuminated, then it is worth it, if it only gives benefits in the partially shaded case, then it is marginal as to whether it is worth the effort. My free time is a premium so I will need to monitor for a while.