Under partial shading, a solar panel's power curve develops multiple peaks. Standard MPPT algorithms get trapped at the wrong peak. Here's how global MPPT tracking finds the true maximum — and why it matters for your yield.
An unshaded solar panel produces a smooth I-V (current-voltage) curve. As voltage increases from zero (short circuit), current remains relatively flat until a "knee" point, then drops rapidly to zero at open circuit voltage (Voc). The P-V (power-voltage) curve — which is just current × voltage at each point — has a single, well-defined peak. This peak is the Maximum Power Point (MPP).
Standard MPPT algorithms like Perturb & Observe (P&O) and Incremental Conductance (IC) work by slightly adjusting the operating voltage up or down and measuring whether power increased or decreased. If power goes up, keep going in that direction; if power goes down, reverse. With a single-peak curve, this hill-climbing approach always converges on the global maximum. It's simple, fast, and effective — as long as there is only one peak.
When part of a solar panel (or one panel in a string) is shaded, the P-V curve fractures into multiple peaks. This happens because bypass diodes activate to protect shaded cells from reverse-bias damage. Each bypass diode creates a separate "step" in the I-V curve, and each step produces its own local power maximum.
A typical 144-half-cell panel has three bypass diodes. Partial shading on one section activates one diode, creating two distinct peaks on the P-V curve. If two sections are partially shaded, three peaks appear. In a string of panels with varying shade levels, the P-V curve can have as many peaks as there are panels in the string.
Consider a 540 Wp panel with three bypass diode groups (each covering 24 cell pairs):
| Condition | Peak 1 (Local) | Peak 2 (Global) | P&O tracker output | Power lost |
|---|---|---|---|---|
| No shade (uniform) | 540 W at 41 V | — | 540 W | 0% |
| One diode group 50% shaded | 310 W at 28 V | 400 W at 41 V | 310 W (stuck at local) | 22.5% |
| One diode group fully shaded | 180 W at 18 V | 360 W at 41 V | 180 W (stuck at local) | 50% |
| Two diode groups partially shaded | 150 W at 14 V | 280 W at 41 V | 150 W (stuck at local) | 46% |
In every shading scenario, the standard P&O algorithm starts near the low-voltage end (where current is highest) and climbs the first peak it finds. Because it only looks for local gradients — "is the next step up or down?" — it has no way to know that a taller peak exists further along the voltage axis. It locks onto the local maximum and stays there, leaving 20–50% of available power uncaptured.
P&O adjusts operating voltage by a small fixed step (typically 0.5–2 V), measures the change in power, and follows the uphill direction. When it detects a downhill step (power decreasing), it reverses and oscillates around the peak. This oscillation is itself a source of energy loss — typically 1–3% even on a single-peak curve. Under partial shading, P&O settles on whichever peak it encounters first, which is usually the leftmost (lowest voltage) local maximum — almost never the global maximum.
IC uses the derivative of the I-V curve (dI/dV) to find where power change is zero — the mathematical definition of a peak. IC is faster and more stable than P&O around the peak, but it still follows the local gradient. It will find the nearest peak with higher precision, but it cannot detect a peak beyond a valley. Same problem, sharper focus on the wrong answer.
Between any two peaks on a shaded P-V curve lies a valley — a region where power decreases as voltage increases. To get from a local maximum to the global maximum, the algorithm must cross this valley: it must deliberately move in a direction that reduces power. Hill-climbing algorithms by definition cannot do this. They are designed to stop when they see a decline. This is why partial shading is called the "MPPT killer."
A Global Maximum Power Point Tracking (GMPPT) algorithm periodically scans the full voltage range of the panel — from near-short-circuit to near-open-circuit — to map all peaks on the P-V curve, then locks onto the tallest one. The key operations:
During the sweep, the controller isn't operating at any peak — it's momentarily sacrificing power to gather information. A naive sweep of the full voltage range takes 50–200 ms and costs 0.1–0.5 Wh of energy per scan. If the controller sweeps too often, the scan cost exceeds the gained benefit. If it sweeps too rarely, the system stays stuck at a local maximum for minutes during changing cloud conditions.
The engineering challenge is knowing when to sweep. Simple timers (sweep every 5 minutes) waste energy on clear days and miss fast-moving clouds. Smarter algorithms monitor the rate of power change: a sudden drop of >10% triggers an immediate re-scan, while stable output extends the interval. Power change detection on the order of 100–500 ms response time is needed to catch cloud-edge transitions effectively.
Alloy Charge's approach: The SunXtractor's MPPT algorithm incorporates proprietary logic that identifies when the controller is locked onto a local maximum. Rather than waiting for a periodic timer, it detects the characteristic signature of a local peak in a multi-peak curve — based on the shape of the P-V curve gradient around the operating point — and initiates a targeted jump to the global maximum. This event-driven approach minimises sweep losses while ensuring the controller is always operating at the true global peak, not a local trap.
The impact of local vs global MPPT depends on how much partial shading your system experiences. For a rooftop with zero shading on a clear day, global MPPT adds nothing — there is only one peak. But real-world conditions include:
Studies on residential rooftops in Bangalore have measured shading losses of 8–25% of annual yield on string systems. Global MPPT recovery typically claws back 40–70% of those losses — translating to a 5–15% improvement in annual energy harvest compared to a standard MPPT controller.
Global MPPT is critical in string systems because shading on one panel creates multi-peak curves across the entire string's P-V characteristic. The higher the string voltage (more panels in series), the more complex the multi-peak landscape becomes — a 10-panel string can produce a P-V curve with 10 distinct peaks.
In a distributed MPPT architecture (like Alloy Charge's SunXtractor), each controller tracks only one panel's P-V curve. A single panel has at most three peaks (corresponding to its three bypass diode groups). This simpler landscape is faster to scan, faster to resolve, and less prone to false peaks. The combination of per-panel tracking and global MPPT within each controller is the most complete solution to partial shading losses — eliminating both the cross-panel string effect and the within-panel bypass diode effect.
| MPPT Type | Shading Scenario | Power Recovery |
|---|---|---|
| Standard P&O (string) | 1 of 10 panels partially shaded | 60–75% of available power |
| Global MPPT (string) | 1 of 10 panels partially shaded | 85–92% of available power |
| Standard P&O (per-panel) | 1 panel partially shaded | 75–85% of that panel's power; others unaffected |
| Global MPPT (per-panel) | 1 panel partially shaded | 92–98% of that panel's power; others unaffected |
Our team will assess your rooftop's shading profile and show you how much energy global MPPT tracking can recover for your specific site.
Schedule Free Assessment