The number a borehole or pool pump needs from an inverter is not the continuous rating on the quote. It is the inverter's peak rating in watts together with the number of seconds the datasheet allows that peak to be held, measured against how long the motor sits at locked-rotor current before it is genuinely turning. A load list built from nameplate watts can add up perfectly and still leave you with a system that trips every time the pump's pressure switch closes.
The morning the pump found the gap
I was standing in the garage just after half past six, torch in one hand, watching the inverter screen because the whole system had been sold to me as "properly sized". The grid had dropped out a few minutes earlier. Changeover relay clicked, house went onto battery, everything calm — until the borehole pump's pressure switch closed.
You hear it before you see anything. That low straining hum of an induction motor pulling hard without turning properly yet. The inverter fan spun up to a note I hadn't heard from it before. Then the beep, and the red line on the display: overload. Everything on the backup side went dark, including the freezer that had been running happily for the previous ten minutes.
That same pump had started every morning for years on municipal supply without a whimper. That is the part that confuses people.
On the grid, you never tested anything
When the grid is up, a hybrid inverter is mostly a pass-through. The transfer relay closes and the load draws from the utility, which can deliver fault current of a size no residential inverter will ever match. The pump starts on that. The inverter is not really doing the work; it is standing next to the work.
So the pump works fine on your system for months, and then load shedding arrives, the grid disappears, and the same pump is suddenly trying to start on a battery and an inverter. The motor's start becomes the inverter's problem, and nobody warned you that it was ever going to be one.
I have stopped trusting anyone's "it worked when we tested it" unless that test was done with the main breaker off.
Nameplate watts are running watts
Every load list I have seen built from appliance labels looks something like this: fridge 150 W, freezer 200 W, borehole pump 750 W, TV, wifi, a few lights. Add it up, call it 2,5 kW, subtract from a 5 kW inverter, plenty of room left.
That arithmetic is correct and useless. The nameplate on a motor is its running rating, measured when the shaft is turning at full speed and the motor is doing the work it was designed for. The start is a separate event. Before the rotor turns, the motor looks close to a short circuit and pulls several times its running current. On a pump, how long that lasts depends on what the impeller is pushing against — a standing head of water in the line, a non-return valve holding pressure, cold water, a long suction run. There is no single multiplier you can look up in a table and apply to every pump in the country. Two pumps with the same nameplate can have very different starts.
Which loads do this to you
Pumps are the loudest offenders because they are big and they start against pressure. But they are not alone. Fridge and freezer compressors, air conditioners, and some gate and garage door motors all draw a start current well above what their labels suggest. Compressor loads are the sneakiest of the lot because they cycle. It is rarely the first start after the power goes off that catches you. It is the fourth, at eleven at night, when the fridge and the freezer have both decided to kick over while the battery is already sitting lower than it was at six.
Two numbers, and one of them has a clock on it
Open the inverter's datasheet and find two ratings. Continuous output, usually quoted at a temperature. And a peak or surge rating, often around twice continuous, with a duration attached — a number of seconds, not a vague promise of "surge capability".
That duration is the whole argument. It is usually qualified by temperature, and sometimes specified into a resistive load rather than a motor. Read it like a contract with an expiry date. If it says five seconds and your pump needs seven, the inverter does not care that the numbers added up on paper. It will go into overload, and it will do it every time the pressure switch closes.
There are two more links in the same chain, because the peak rating on the box is only as good as what feeds it. Inverter topology matters: a unit built around a large transformer behaves differently on a motor start than a compact high-frequency unit carrying the same label figure, and that difference does not show up in the headline watts. And the battery has to be able to deliver the amps. If the bank's continuous discharge limit sits below what the surge needs, the DC bus sags, the inverter drops out on low DC voltage, and you spend the afternoon blaming the wrong component entirely.
It is the surge plus whatever was already running
Here is where my load list changed shape. I stopped listing loads in watts and started listing them in two columns: running watts, and how the load starts.
The number that matters at the moment of the trip is not the pump's surge on its own. It is the pump's surge plus everything already drawing on the backup side. If the geyser element is on, that is a couple of kilowatts of resistive load that was perfectly fine five minutes ago and is now eating the headroom the pump needs. Same story with the kettle, the iron, the tumble dryer. The inverter sees one total, and it does not distinguish between the load that was there before and the load that just slammed in.
Stagger the starts before you buy anything
The cheapest fix I have found is a clock, not a bigger inverter. The pump, the geyser element, the kettle and the air conditioner should not be allowed to begin together. Put the geyser on a timer so its element never comes on in the first stretch of an outage. Give the pump a clear run — let it start, come up to pressure and settle before anything else switches on. Keep the heavy resistive loads for when the grid is back.
Do this first and a surprising number of "undersized inverter" problems simply stop happening. I have watched a system go from tripping on every second outage to never tripping, with the same inverter still on the wall.
Measure it instead of arguing about it
A clamp meter with a true-RMS inrush or peak-hold function, clamped around the pump's live conductor, will show you the start current as a number. Watch what it reads in the first second. Then look at the inverter's fault log, which usually records what the load was doing when it tripped. Between the two, you stop guessing.
Two cautions. A meter without inrush capture will miss the part you care about, because the spike is over in milliseconds. And the current you measure at the pump is not always what the inverter sees — a long cable run, a loose terminal or an undersized conductor all change the picture. If the readings look strange, that is worth an electrician's eyes before you buy a bigger box. The wiring on the backup side, and whether that pump circuit belongs on it at all, falls under SANS 10142-1 and is not a decision to make from a forum thread.
When no sensible inverter will do it
Sometimes the honest answer is that the pump's start is genuinely hard, and no inverter of a size you would want to live with will carry it. A three-phase borehole pump on a variable speed drive ramps up gently by design, and you may never see the problem at all. Some single-phase pumps can be softened too, but that is a change to the motor circuit — a contractor's job, with its own trade-offs in cost and reliability — not something to bolt on because someone online said it worked for them.
And oversizing the inverter is not free. A bigger unit draws more at idle, costs more, and often wants more battery and more panel behind it to be useful. Add in what the connection rules and your municipal meter allow, and the case for simply going bigger gets thinner. If the real constraint is the battery's discharge current, or the stagger, a bigger inverter buys you nothing at all.
What I keep coming back to is this. The datasheet's peak rating and the duration printed beside it are the real specification for a pump. Everything else — the continuous kW, the quote, the neat load list — describes a house for the ninety-nine percent of the day when nothing is starting.
