12KPV MPPT losing tracking? linear drop in power on the mppt (not clouds, nor battery near charged)

Had almost exactly this on my narrowboat setup last summer — drove me absolutely mental for about three weeks before I figured it out.

The linear drop is the key clue here. Clouds give you that jagged, unpredictable dip. Battery approaching full gives you the classic absorption curve taper. But a linear slope like you're describing? That pointed to heat throttling on mine, specifically the MPPT controller itself getting cooked in an enclosed battery compartment.

On my boat the Victron SmartSolar was tucked in a poorly ventilated locker. Fine in spring, but once the bilge temps crept up in July it was derate city every afternoon between roughly 1pm and 4pm. Perfectly correlated with ambient temperature, not cloud cover or state of charge.

Few things worth checking:

• Controller temperature — does it have a temp readout in the app? Mine showed 58°C before it started throttling
• PV string voltage — is it staying consistent or also dropping? Rules out a dodgy connection or panel issue
• Time of day correlation — log it for a few days, does the drop always start at the same time?

Also worth ruling out a failing BMS on the battery side — I've seen a Fogstar battery's BMS get twitchy with heat and start limiting charge current, which would look identical from the MPPT's perspective.

What controller firmware version are you running? There were some tracking algorithm quirks in older builds that could cause weird stepped-down behaviour too.

Would be good to know your panel configuration and whether it's a fixed install or portable — changes what I'd suspect first.

My Victron MPPT did the exact same thing on the boat last August — turned out the temp sensor cable had worked loose and it was "protecting" itself based on phantom readings like a drama queen.

Good points from @Deano and @BoatMark already.

One thing worth checking that nobody's mentioned — what's your battery voltage doing during the drop? If it's genuinely linear rather than stepped, I'd be looking at whether the MPPT is seeing a false high voltage reading and backing off thinking the bank is fuller than it is.

On my shepherds hut setup I had a corroded negative connection between the battery and MPPT ground that was causing exactly this — voltage sense was skewed, controller was throttling itself unnecessarily.

Few questions:

• Are you monitoring panel voltage AND battery voltage simultaneously during the event?
• What cable run length between panels and MPPT?
• Any joints or connectors mid-run?

Voltage drop across a dodgy MC4 connector can also cause the MPPT to lose the peak of the IV curve and hunt downhill. Worth measuring connector temps with a cheap IR thermometer when it's happening.

The linear drop pattern on MPPT trackers is almost always one of two things: partial shading creating a false peak on the IV curve (tracker locks onto secondary peak), or — and this caught me out on my narrowboat last year — a failing bypass diode in one of the panels.

@Deano's narrowboat angle is relevant here. Vibration from the engine works connections loose over time. Worth pulling each panel connector apart, inspecting for corrosion, and reseating firmly.

Specific to the 12KPV: check whether the drop correlates with panel temperature rising through the day. As panels heat up, Voc drops — if your string voltage is already marginal for the MPPT's tracking window, it can lose the curve entirely around midday when you'd expect peak output.

Log your actual panel input voltage at the controller terminals throughout the drop period. That'll tell you everything.

Had this exact issue with my cabin setup last year. Spent three days convinced the MPPT was dying, ordered a replacement, unpacked it... then realised my panel junction box lid had warped slightly and was casting a shadow across the bottom row of cells every afternoon like clockwork. Linear drop, started around 1pm, gone by 3pm. Textbook.