PV Surplus Control: Hysteresis, Timers and Thresholds for Switching Loads

Surplus control starts as one line: if the surplus exceeds two thousand watts, switch on the heating element. It is correct, it is obvious, and it produces a relay that clicks every ten seconds until something gives out.

The reason is in the sentence itself. The load consumes exactly the surplus that permitted it, so the condition that switched it on is false one measurement later. Everything below is about the fact that a control loop cannot use the same number as its trigger and its consequence without saying how far apart the two are allowed to be.

The Feedback Nobody Plans For

A two-kilowatt element switched on at a surplus of 2200 W leaves 200 W. The automation reads 200 W, finds it below the threshold, and switches off. The surplus returns to 2200 W, the condition is true again, and the cycle repeats for as long as the sun holds. Nothing is broken; the rule is simply asking a question whose answer it changes by asking.

The count is what makes it expensive. A contactor rated for 100 000 mechanical operations sounds generous until the loop produces one every few seconds. At the rate in the upper panel it is used up in about two months, and the failure mode of a welded contactor is a load that stays on all night.

Two Thresholds Instead of One

Hysteresis separates the switch-on point from the switch-off point far enough that the load’s own consumption fits between them. The on threshold is the load plus a reserve; the off threshold is not the same number but a small amount of import, which is the only honest sign that the surplus has actually run out.

on    surplus ≥ load + reserve     held for ≥ 3 min
off   import  ≥ 200 W             held for ≥ 5 min
      and runtime ≥ 15 min

2000 W element, reserve 200 W
  on at a surplus of 2200 W
  after switching on: 2400 − 2000 = 400 W surplus  → no reason to switch off
  without hysteresis: 400 W < 2200 W                → off again immediately

The gap between the two thresholds has to be at least the rated power of the load. Anything narrower and the load’s own draw crosses back over the off point, which is the same oscillation with more arithmetic in front of it.

Three Timers, Three Different Jobs

The two delays are not symmetric on purpose. Switching on late costs a few minutes of export at the feed-in tariff; switching off late costs a few minutes of import at the retail price, which is three to four times as much. So the off delay is the one to keep short if either has to give.

Minimum runtime is the parameter that gets left out and the one that matters most for anything with a compressor. A heat pump or a fridge-like load pays for a start with wear and with a burst of inrush current, and hysteresis alone does not stop a rule from starting it four times an hour on a broken-cloud afternoon.

Which Number Is Actually Being Measured

Surplus is a property of the grid connection, not of the roof. The number to read is the meter at the house connection, signed: negative for export, positive for import, or two separate sensors that never both show a value at once. Anything derived from inverter production alone is blind to the household consumption underneath it and will switch a load on into a running dishwasher.

Two details decide how well that reading behaves. Smart meter interfaces report at their own pace – between two seconds and thirty – and a rule that samples faster than the meter reports is comparing a fresh value with a stale one. And short averaging is worth more than precision here: a five-second mean of the grid power is a far better trigger than an instantaneous reading that is mostly kettle.

What a Battery Puts In the Middle

With a battery in the house, the surplus at the meter is what is left after charging, which means the battery and the switched load are competing for the same watts and the winner is whoever the rules let go first. Give the battery priority and the element never runs; give the element priority and the battery never fills before evening.

The arrangement that survives a winter is a state of charge condition on top of the power condition: the load may run only above a certain charge level, so the battery gets the morning and the load gets the middle of the day. That single extra condition also stops the case nobody notices – a load happily consuming surplus that the battery had already stored, discharging it at a round-trip loss of ten to fifteen per cent for no gain at all.

Loads That Modulate Do Not Have a Threshold

A wallbox does not switch, it scales – but only within limits, and the lower limit is the one that turns it back into a switching problem. Six amperes is the minimum charging current, which is about 1.4 kW on one phase and 4.1 kW on three, and below that a car does not charge slowly, it does not charge at all.

So the on threshold for a three-phase wallbox is 4.1 kW of surplus, not zero, and everything between zero and that figure is a range in which the only options are one phase or nothing. Switching between one and three phases is itself a load change of nearly three kilowatts, which needs its own hysteresis and its own minimum runtime – the same three parameters as the heating element, applied to a decision that looks like a smooth dial and is not.