What we actually sold, and to whom

On 13 November 2024, Company Importação e Exportação Ltda., of Rua Patrimônio in Vila Velha, Espírito Santo, placed two orders with us. Between them: sixty Atlas turbine bodies and sixty blade sets, invoiced at just under €57,000. That is the part we can vouch for from our own records, and it is the reason this page exists at all.

Vila Velha sits beside Vitória, one of Brazil's major ports and the service base for a great deal of offshore oil work in Espírito Santo. A distributor there taking sixty vertical turbines is not a mystery. What happened to four of them is the interesting part, and that part comes from our distributor rather than from our own instruments: they were fitted to Petrobras vessels.

We are flagging that distinction deliberately. Petrobras is not our customer — they are our distributor's customer. We have no meter readings from those decks and no survey of how the machines were installed. Everything below about ships is engineering reasoning and honest caveat, not a performance claim.

Why a vertical axis suits a moving deck

The wind a turbine feels on a ship is not the wind on the weather chart. It is apparent wind — the vector sum of the true wind and the wind the vessel makes by moving through it. Change course and the apparent wind swings. Change speed and it swings again. A vessel under way alters both all day.

A horizontal-axis turbine has to point into the wind to work, so it needs a yaw mechanism and it needs time to turn. That is a fine trade on open land, where the wind holds a bearing for minutes at a stretch; it is why every utility-scale wind farm is horizontal. On a deck that is changing heading, the machine spends its life turning towards wind that has already moved.

A vertical-axis machine has no bearing to find. It accepts wind from any direction, which is precisely the property you want when direction is the one thing you cannot rely on. It also means there is no yaw drive, no slip rings and no rotating electrical joint — and in salt air, every mechanism you delete is one that cannot seize.

On land you choose a turbine for the wind speed. On a moving vessel you choose it for the wind direction, because direction is what will not sit still.

Everything is in the one sealed housing — generator, IoT charge controller, display, terminals. On a deck, that matters more than it does on a roof.

What a deck asks that a roof does not

The Atlas body is a single sealed aluminium housing holding the generator, the built-in IoT charge controller, the display and every terminal. On a house that is a convenience — no second box to find a wall for. On a working deck it is closer to a requirement, because there is no dry cupboard to put a separate controller in. The body weighs 22.30 kg on our own workshop scale, without blades or mast.

Noise is the other thing that reads differently at sea. The machine runs at 35–40 dB. On a suburban roof that is a question of neighbours. On a vessel, crew sleep in accommodation directly below the deck the equipment is bolted to, and they cannot go anywhere else.

Now the part that argues against a straightforward sale. There are three Atlas blade sets, each laser-cut from 1.2 mm high-strength aluminium and matched to a wind profile rather than to a price tier: low wind runs 2–20 m/s for 2 kW peak, moderate runs 4–25 m/s for 3.8 kW, and high wind runs 5–35 m/s for 4.5 kW in a six-blade configuration built for exposed coastal sites. For a sheltered garden the answer is usually the low-wind set. At sea it is the opposite. Apparent wind on a vessel under way is frequently higher than the true wind, and an exposed deck has nothing upwind of it for a thousand miles. Specifying a marine installation the way you would specify a back garden is the surest way to get it wrong.

A still from our own footage: gusseted pedestals bolted through the deck, cross-braced, with drip loops on the cable runs. All of that is the operator's engineering, not ours.

And two things we will not pretend away. Salt is relentless, and an aluminium housing helps but does not exempt anyone from a maintenance regime. A hull pitches and rolls, which imposes cyclic and gyroscopic loads that a rooftop never sees — loads we have not characterised for this machine. Those blue pedestals in the photograph were fabricated by the operator, and the mounting engineering on a vessel is properly theirs, not ours.

What we are not going to tell you

We are not going to give you a kilowatt-hour figure for these vessels. We have no metered data from them, and inventing one would be worth less than nothing — the first marine engineer to read it would check the arithmetic and stop trusting the rest of the page. Our published annual range for a small Atlas is 180 to 3,500 kWh, and that range exists precisely because output depends on a site we have not measured.

Nor are we going to market the Atlas as certified marine equipment. It is not sold as such. Four of them ended up at sea because a customer looked at an omnidirectional machine in a sealed case and drew a sensible conclusion, and we would rather describe that accurately than dress it up. If you are considering something similar, talk to us before you order, and measure the wind where the machine will actually stand — on a fixed installation, an anemometer settles in one season what a brochure cannot settle at all.

Our thanks to Company Importação e Exportação for the order. The sixty units, the two order dates and the invoice total come from our own sales records. The specifications are our own published figures for the Atlas and are checkable on the product page. The deployment aboard Petrobras vessels is reported to us by our distributor and shown in our own footage; we have not surveyed it ourselves, and we have said so rather than let you assume otherwise.

Wound stators in their housings, before assembly. Copper, laminated steel, three phases.

What “15 kW continuous, 22 kW peak” actually means

Two ratings, two entirely different limits. The 15 kW continuous figure is what the machine will sustain indefinitely without cooking itself, and it is set by heat: Class 200 °C enamelled wire to IEC 60317-13 GR 2, four kilograms of copper acting as a thermal reservoir, parallel-strand bundling that doubles current capacity and spreads the heat across thin conductors, and a housing that cools passively. The 22 kW peak is short-term overload for transients, and it is bounded by physics of a different sort — the demagnetisation threshold of the N42 magnets and saturation of the stator iron.

Here is the part that matters more than either number. A generator rating describes what the machine can absorb without damage. It says nothing about what it will produce. That depends entirely on what you connect to the shaft. Maximum speed is 2,000 rpm, efficiency is about 95% at rated load, and it runs bidirectionally — but a modest rotor turning slowly will draw a small fraction of 15 kW out of it, and no amount of generator headroom changes that.

A rating is a ceiling, not a promise. What turns the shaft decides the number you actually get.

We would rather say that plainly than sell one to somebody expecting fifteen kilowatts from a small rotor in light air. The headroom is still worth paying for — a generator run permanently at its limit is the most reliable way to kill a machine early, and this one at a typical few kilowatts is loafing — but headroom and output are not the same thing.

What is inside it

A 12-pole stator with 36 slots, three-phase distributed winding in a fractional-slot configuration for minimal cogging and a clean sinusoidal output. Ninety wires per slot — two parallel strands of forty-five turns, 0.70 mm enamelled copper, the insulation grade used in industrial servomotors and EV traction motors. Four kilograms of copper in total.

The rotor carries 24 NdFeB N42 magnets, bread-shaped, 60 × 25 × 5 mm and nickel-coated — 180 cm³ of magnet in total. They are mounted skewed, which cuts cogging torque by around 80%: smoother torque, easier starting, longer bearing life. The core is not a casting but a stack of 240 individual 0.5 mm silicon-steel laminations built up to 12 cm, about 16 cm in outside diameter, which is what keeps eddy-current losses down. Machining is held to ±0.05 mm, and the unit is EMC compliant to EN 61000-6-3. All in, it weighs 20 kg.

None of that photographs well and none of it appears on a box. It is, however, the entire difference between a generator that is still turning in ten years and one that is not.

Olympia — state capital, county seat of Thurston County, at the head of South Puget Sound.

Where they are going

Olympia is the capital of Washington and the county seat of Thurston County, a city of 55,605 people at the 2020 census anchoring the South Puget Sound region, about fifty miles south-west of Seattle. Its downtown is partly built on filled tideland from the dredging that created the deep-water port, and its climate is warm-summer Mediterranean — dry Julys, and the rain arriving between November and January.

We do not know what these motors are for, and we are not going to invent a story about it. This model goes into electric vehicles, industrial machinery, marine installations, robotics and off-grid power systems, and a port city is a plausible home for most of those. One thing is worth saying regardless: if the intended use is wind, the head of Puget Sound is a sheltered inland waterway, and it is a considerably weaker wind resource than the Washington coast or the Columbia Gorge. That is not a reason against the motor. It is a reason to measure before sizing anything around it.

What it costs, and what it does not include

The motor is AU$899 on this store as this is published, on the product page, itemised, with nothing held back for the checkout. That buys the machine on its own: generator, housing, shaft. It does not buy a rotor or blades, a mount or a tower, a rectifier, charge controller, inverter or batteries, the wiring between any of them, or the electrical work and permissions that apply where you are. A component sale is honest about being a component.

If you are building something wind-driven around it, one more purchase comes first: an anemometer at AU$399. Power in the wind goes with the cube of speed, so a site that is 20% slower than you assumed gives roughly half the energy. A season of real measurements at your own address is worth more than any figure we could publish about ours.

Our thanks to the customer in Olympia for the order. Every specification above comes from our own published data for this model and is checkable on the product page; the price is the one on this store on the day of publication. If you are designing around a generator, hold us to all of it.

Crates on the line in our own factory, each packed to the same list.

Where it is going, for anyone picturing the Aegean

Ilioupoli takes its name from Heliopolis — the sun city. It was settled from 1924 by Greek refugees from Asia Minor, and it is now a dense suburb of about 12.7 km², its built-up area continuous with the suburbs around it, with the forested ridge of Hymettus covering its eastern half. That is a very different place from the postcard. It is not a headland, it is not an island, and it is not the open Aegean.

This matters because the wind that reaches a roof in a continuous built-up area has already been through several kilometres of buildings. It arrives slower, more turbulent, and constantly changing direction — and with a forested mountain immediately to the east, the direction it arrives from depends on the day. The etesians, the meltemia, are real: strong dry north winds that blow across the Aegean from roughly mid-May to mid-September. But they are a sea wind, and a suburban rooftop under a ridge sees a considerably rougher, more broken version of whatever is happening offshore.

A suburb is a harder wind site than a cliff. That is not a reason to skip the turbine — it is a reason to pick the right kind.

One thing that does not matter much here: altitude. Ilioupoli climbs the foot of Hymettus, but at a couple of hundred metres the air is only about 2% thinner than at sea level, and power scales directly with air density. It is a rounding error. We mention it only because we made a great deal of the same correction when we wrote about a customer at 1,600 m in South Africa, where it costs about 15%. Here it does not, and it would be dishonest to imply otherwise.

Why a vertical machine for a site like this

A horizontal turbine has to point into the wind to work. In clean, steady air that is an advantage — it is why every large wind farm is horizontal. On a suburban roof, where the wind changes direction every few seconds, the machine spends a meaningful part of its life turning to face air that has already moved on.

A vertical machine has no direction to find. It takes wind from any bearing, which is exactly the property you want where the flow is broken and shifting. It starts turning at 2 m/s with the low-wind blade set — about a third of the cut-in of a typical horizontal machine — which matters on a site whose average is modest and whose good hours are scattered. And it runs at 35–40 dB, quieter than the conversation you are having beneath it. In a neighbourhood where the houses are continuous with each other, noise is not a specification, it is a relationship with the people next door.

Athens at dusk. Ilioupoli sits six kilometres south-east of the centre, under Hymettus.

There are three blade sets, each laser-cut from 1.2 mm high-strength aluminium and matched to a wind profile rather than to a marketing tier. Low-wind blades run from 2 to 20 m/s for a 2 kW peak. Moderate-wind blades run from 4 to 25 m/s for 3.8 kW. High-wind blades run from 5 to 35 m/s for 4.5 kW, in a six-blade configuration built for exposed coastal sites. Choosing the wrong one is the most common way to be disappointed by a small turbine, and for a sheltered suburban roof the answer is usually the low-wind set — the one with the smallest headline number.

What is actually in the crate

The machine going to Athens is an Atlas, and its body is one sealed aluminium housing containing the generator, the built-in IoT charge controller, the display and every connector. There is no second box to find a wall for and no separate enclosure to weatherproof. Weighed on our own workshop scale it read 22.30 kg — a Teknika TCS-K2+R, 300 kg capacity, 50 gram divisions, which is a workshop instrument and not a certified trade one. That is the body alone, without blades or mast, and it is the number your roof structure has to start from.

Mounting flange, cabling, fixings, tools and the manual, packed around the body.

Inside the housing is the same 15 kW continuous-rated motor we put in the Magnum, which at a typical 1 kW output is running at under 7% of its rating — cold, unstressed, and a long way from the limit. The stator is a 12-pole, 36-slot three-phase distributed winding in Class 200 °C enamelled copper to IEC 60317-13, the insulation grade used in industrial servomotors and EV drivetrains. None of that photographs well. All of it is why the machine is still there in ten years.

What it costs on this store, and what it does not cover

It is AU$1,799 on this store as this is published, and that price includes the built-in IoT charge controller — there is no separate controller to buy, which is the main reason this bill is shorter than the Magnum one. The blade set is chosen to match your site rather than picked off a shelf, so please talk to us before you order rather than guessing from this page: the right set for a sheltered suburban roof is not the one with the largest number on it.

What the price does not buy: a mast or roof mount, the foundation or fixings into your structure, the cable run down to wherever the power is going, batteries and an inverter if you are storing it, and any electrical work, connection agreement or planning permission that applies where you live. None of that is hidden — it is simply site-specific, and anyone quoting you a single all-in figure for a Greek rooftop has not seen your roof.

Before any of it: measure your wind

Every figure above is worthless without one we cannot supply: the wind at your address. Our own published range is 180 to 3,500 kWh a year, and that span is not vagueness — it is the honest width of the answer once you account for site and blade set. Power goes with the cube of wind speed, so two roofs on the same street can differ by a factor of two in speed and eight in energy. In a dense suburb under a ridge, that variation is the rule rather than the exception.

So measure first. An anemometer costs a fraction of a turbine and settles the question inside a season. Then put your own measurements into the calculator on the product page — your wind, your tariff, your roof. For some of you it will say the numbers work comfortably. For others it will say a turbine is the wrong purchase at your address, and we would far rather it told you that before you bought one than after.

Our thanks to the customer in Ilioupoli for the order. Prices are those on this store on the day of publication and are checkable on the product page at any time. The weight came off our own scale and the wind facts are the ones we would want if we were the ones buying. Measure your site — then hold us to every figure above.

Where this is, for anyone picturing it wrong

Kempton Park sits on the Highveld at around 1,600 metres, beside OR Tambo International, on the road network that carries traffic between Johannesburg and Pretoria. It is about 500 km from the sea. If you had a beach in mind, that is the wrong picture — this is high, dry, open country, and it is windy in the way that open country at altitude usually is.

Sandton, Johannesburg — the same province, about twenty kilometres from Kempton Park.

Altitude matters to a turbine, and not in your favour. Air at 1,600 m is roughly 15% less dense than at sea level, and the power in the wind is directly proportional to air density. The same wind speed on the Highveld therefore carries about 15% less energy than it would on the coast. Any output figure you read anywhere — ours included — is quoted at sea-level density unless it says otherwise. It is a small correction, but it is the sort of thing you would rather hear from us now than work out for yourself later.

Thin air costs you about 15% of the energy in the same wind. Nobody puts that on a spec sheet.

What is actually in a TESUP motor

The generator is the part that decides whether a small turbine lasts. Ours is rated 15 kW continuous — considerably more capacity than the machine is ever asked to deliver, which is deliberate, because headroom is what stops a generator cooking itself on the gusts. Inside it: a 24-magnet rotor running NdFeB N42 magnets, four kilograms of pure copper winding, and a silicon-steel laminated stator core rather than a cheap casting that overheats.

The details that photograph badly are the ones that matter. Skewed magnet mounting cuts cogging torque by around 80%, which is what lets a rotor start in light air instead of standing still. The winding uses Class 200 °C enamelled wire to IEC 60317-13 — the insulation grade used in industrial servomotors and EV drivetrains. Machining is held to ±0.05 mm. All of it comes out of our own European factories, where we make over 270 components in house.

The N1 — the road that carries most of what South Africa moves.

What a complete machine costs here

A generator on its own is not a wind turbine. Here is every line of a working one, at the prices on this store as this is published.

That is roughly AU$141 for each kilowatt of peak output, or about AU$447 per kilowatt of the rated figure. What it does not buy is a mast, a foundation, cabling, batteries, an inverter, or the electrical work and permissions that apply where you are. None of that is hidden — it is simply site-specific, and anyone quoting you one all-in number has not seen your site.

Before you buy anything: measure your wind

Sunrise over open country. Two sites a street apart can differ by a factor of two.

Every number above is worthless without one we cannot supply: the wind at your address. Power goes with the cube of wind speed, so a modest difference in site is an enormous difference in output — and on the Highveld you have that 15% density correction sitting on top. An anemometer costs a fraction of a turbine and settles the question inside a season.

Then put your own measurements into the calculator on the Magnum page — your wind, your tariff, your roof or your yard. For some of you it will say the numbers work comfortably. For others it will say a turbine is the wrong purchase at your address, and we would far rather it told you that before you bought one than after.

Our thanks to Africa Truck Stop for the order. Prices are those on this store on the day of publication and are checkable on the product page at any time. If you are weighing this up wherever you are, measure first — and then hold us to every figure above.

What it makes, and the wind it takes

A Magnum can put 10 kWh into your system in a single hour. That is not a figure we are nervous about — it is on our own published curve, and you can read it off the chart at the top of this page. What we would rather you took away is the second half of the sentence: the wind that hour needs.

Nine kilowatts at 20 m/s, twelve at 22. Ten therefore lands around 21 m/s — roughly 76 km/h, a Beaufort 9 gale. That is a storm hour, not a Tuesday afternoon. The everyday numbers sit further down the curve: 3.8 kW at 15 m/s, about 2 kW at 12, and the rotor already earning from 6 m/s. Across a year that works out at somewhere between 2,500 and 9,000 kWh, depending entirely on where you put it.

Why the curve climbs so steeply is not marketing, it is the power equation: available wind power scales with the cube of wind speed. Double the wind and there is eight times the energy in it. That is why a turbine's headline figure tells you almost nothing on its own, and why the full curve tells you nearly everything.

Anyone can quote one number. Ask for the whole curve — that is the document that cannot be dressed up.

Why we think it is the most powerful of its kind

The motor inside a Magnum is rated 15 kW continuous — considerably more capacity than the machine is ever asked to deliver. That is deliberate. Headroom is what stops a generator cooking itself on the gusts, and running a generator permanently at its limit is the single most reliable way to kill a small turbine early. Feeding it is a rotor of 4.34 m² swept area on carbon fibre composite blades, driving a 24-magnet rotor with NdFeB N42 magnets and four kilograms of pure copper winding.

The details that do not photograph well are the ones that decide how long it lasts. Skewed magnet mounting, which cuts cogging torque by around 80% and lets the rotor start in light air instead of sitting stalled. Class 200 °C enamelled wire to IEC 60317-13, the insulation grade used in industrial servomotors and EV drivetrains. A silicon-steel laminated stator core rather than a cheap cast one that overheats. Machining held to ±0.05 mm. Put together, the blade aerodynamics reach a power coefficient of 0.45 in good wind, and with the electrical efficiency of that winding the machine converts around 43% of the energy passing through the rotor into electricity — at the top of what a small horizontal-axis turbine can do.

And very probably the most affordable in its class

A specification like that usually arrives with a price to match. Ours does not, and the reason is unglamorous: we make it ourselves. Over 270 components come out of our own European facilities — generators, electronic cards, carbon fibre blades, silicon-steel stator laminations, mounting hardware. There is no chain of distributors between the factory and your roof. That is the whole trick, and it is not one anybody can copy quickly.

Rather than leave you to guess at what a real machine costs, here is every line of it, at the prices on this store as this is published.

Divide that however you find useful: roughly AU$141 for each kilowatt of peak output, or about AU$447 per kilowatt of the rated figure. If you already own a controller, take AU$399 off. Mounting hardware depends on where you are — in several of our markets the mast ships with the blade set, and elsewhere customers fabricate to our fitting, which is covered further down.

We are not asking you to take our word for either claim, because we are the ones selling it and every manufacturer says this about their own product. So go and compare — but compare properly. Ask for the whole power curve, not a single headline kilowatt. Ask the price of a complete working machine, not the price of whichever part sounds cheapest. Then put the two side by side. Ours are on the product page, itemised, with nothing held back for the checkout. If you find something that genuinely beats it on both counts, we would like to know — that is the sort of thing that changes what we build next.

What a turbine still needs from you

Being confident about the machine means being honest about everything around it. That AU$1,697 buys the turbine. Your site supplies the rest: a mast, a foundation or fixings into your structure, cabling down to wherever the power is going, batteries and an inverter if you are storing it, and whatever electrical work and permissions apply where you live. None of it is hidden — it is simply site-specific, and anyone who quotes you one all-in figure has not seen your site.

The mast deserves a paragraph of its own, because it is where most self-builds go wrong. The push on a rotor rises with the square of wind speed, so the 10 kWh hour is also the hour carrying roughly four and a half times the sideways force of a 10 m/s day. If you are making your own pole, spend on diameter before wall thickness: bending stiffness follows the fourth power of diameter, so widening a 50 mm tube to 60 mm at the same 3 mm wall buys about 78% more stiffness for 21% more steel, while doubling that wall to 6 mm buys 66% more for 87% more steel. Height costs you more than either — stiffness falls with the cube of unsupported length — so buy height with guy wires, not optimism. If a mast hums, the answer is a stiffer mast rather than rubber packing, because the noise is resonance and not roughness. And the one dimension you cannot improvise is the fitting: the Magnum head seats on a 37 mm inner diameter at the top of the pole.

Before any of it: measure your wind

Every figure above is worthless without one number we cannot supply: the wind at your address. Two houses on the same street can differ by a factor of two, and because power goes as the cube of speed, a modest difference in site is an enormous difference in output. An anemometer costs a fraction of a turbine and settles the question inside a season.

Then put your own measurements into the calculator on the Magnum page — your wind, your electricity price, your roof. For some of you it will say the numbers work comfortably. For others it will say a turbine is the wrong purchase at your address, and we would far rather it told you that before you bought one than after.

The most powerful of its kind, and very probably the most affordable. We believe both, and we have published the whole curve and itemised every part so you can decide for yourself instead of taking our word for it. Prices are those on this store on the day of publication and are checkable on the product page at any time. That is the only version of a claim like this worth making.

The Atlas body is the part that does the work. The generator, the smart charge controller, the display and every connector sit inside one sealed aluminium housing built for permanent outdoor life. There is no separate control box to find a wall for, and no second enclosure to weatherproof. That is why the figure is what it is: nearly the whole electrical system of a wind turbine, in a single piece you can lift.

What 22.3 kg does not include

The blades. The pole or the mast. The mounting hardware. The Atlas takes swappable blade sets, and which set you choose changes the total. So please do not size a roof bracket or a mast on 22.3 kg — that is the body alone. Your structure has to carry the whole assembly, plus every load the wind puts through it, on the worst day of the year rather than an average one.

One Atlas body on the workshop scale. The reading is 22.30 kg.

Output figures sell turbines. Weight decides where you are allowed to put one.

Why the weight is the first real constraint

Ask a prospective owner about a turbine and they ask about kilowatts. Ask the person who has to fix it to a building and they ask about kilograms, then about where the load lands, then about wind class. A rated output is a promise about good conditions. A weight is a fact your rafters have to live with in every condition, for years.

This is also why we would rather publish the number than be asked for it. If 22.3 kg plus blades is too much for the structure you had in mind, that is far better discovered now than after delivery.

About the scale, because the number should be checkable

It is a Teknika TCS-K2+R platform scale: 300 kg maximum, with a division of 50 grams. So it resolves to the nearest 0.05 kg, and 22.30 is a clean multiple of that. You can read the panel in the photograph above, including the line that says it is not a certified trade instrument. It is a workshop scale, which is exactly what it looks like.

That is why we write about 22.3 kg rather than implying laboratory precision. It is accurate enough to size a mount and honest about its own limits, which is the most any single measurement can offer.

And measure your wind before you buy anything

Weight is the constraint you can check with a tape measure and a structural opinion. Wind is the one nobody can check for you. Two houses on the same street can differ by a factor of two, so any annual output figure we quote is a guess about your roof — and a guess we have an obvious interest in.

So measure first. An anemometer costs a small fraction of a turbine and will tell you within a season whether your site is worth it. Then put your own numbers into the calculator on the Atlas and Magnum pages — your wind, your electricity price, your roof. It will tell some of you that a turbine will not pay for itself where you live, and we would rather it told you that now.

Twenty-two and a bit kilograms is a small fact. But it is the fact that decides whether an installation is a morning's work or a conversation with a structural engineer — and you should have it before you spend anything, not after the pallet arrives.

A static high-pressure system parked itself over Western Europe. High pressure brings heat, and heat brings air conditioning. It also brings still air. Those two things arrive together, and that is the whole problem in a sentence.

What a hot, still week does to a grid

In Great Britain on 23 June, system frequency slipped below 49.8 Hz for close to two hours — the point at which the operator starts spending serious money to hold the network together. Exports to the Netherlands were cut. The next day, the National Energy System Operator secured around 1.7 GW of imported electricity at roughly £1,400 per megawatt-hour. A normal summer afternoon trades in the tens of pounds.

In Belgium, wholesale prices passed €1,000/MWh — the highest in six years.

Heat doesn't break one thing. It breaks everything at once.

This is the part that got the least attention, and it matters the most.

France's nuclear fleet cools itself with river water and returns that water warmer than it found it. When the Rhône, the Garonne and the Seine climb towards their regulatory ceiling of roughly 28°C, EDF is legally obliged to throttle back. In late June, air temperatures at France's river-cooled sites reached 42.5°C. EDF cut about 4 GW, around 6% of the national fleet. The afternoon surplus France normally exports to its neighbours fell from 11–12 GW to under 3 GW.

The cushion France normally exports to its neighbours, before and during the heat. Sources: Montel News; pv magazine.

Gas was no better, for a reason that has nothing to do with fuel supply. A gas turbine's output depends on the density of the air it draws in, and hot air is thinner. Run the same machine at 40°C instead of 20°C and it loses roughly 13% of its capacity and about 7% of its efficiency. British gas plants ran derated through the worst of it — at precisely the hours they were most needed.

And wind, our own product, was the quietest contributor of all. In a quarter when British renewables had just set a record — 53.1% of generation, with wind alone at 35.6% — the heatwave days saw wind fall away to a fraction of that.

Nuclear, gas and wind all faltered in the same week, for the same reason, at the same hour. That is not a wind problem. It is a one-source problem.

The part that concerns us directly

We would rather you heard this from us than worked it out later. Peer-reviewed work published this February, modelling European wind under CMIP6 climate projections, finds that wind droughts — long spells of unusually low wind — are becoming more frequent and longer-lasting, with German summers singled out. Anyone selling wind turbines in Europe ought to be able to say that out loud.

So here is what we think it actually means. It is not “buy a turbine”.

One source is the risk — not one technology

The useful thing about a heatwave is that it is exactly when solar is at its best. Still air and cloudless skies are the same weather. The conditions that quiet our turbines are the conditions that push a solar panel towards its peak. They are close to opposites, which is precisely why owning both beats owning either.

That is not a sales pitch bolted onto a news story. It is the reason we sell flexible solar panels alongside turbines rather than instead of them.

And measure before you buy anything

Wind is site-specific in a way solar simply isn't. Two houses on the same street can differ by a factor of two. Any annual output figure we quote is a guess about your roof — and a guess we have an obvious interest in getting wrong.

So measure first. An anemometer costs a small fraction of a turbine and will tell you within a season whether your site is worth it. Then put your own numbers into the calculator on the Atlas and Magnum pages — your wind, your electricity price, your roof. It will tell some of you that a turbine won't pay for itself where you live. We would rather it told you that now than we did after you had bought one.

The last week of June was not an argument against wind, and it was not an argument for it either. It was an argument against depending on any single thing — one fuel, one technology, one river, or one grid that has to pay £1,400 a megawatt-hour on the hottest afternoon of the year. The households that barely noticed it were the ones generating something of their own, from more than one source.

Hero photograph: Adobe Stock. Illustrative — not a photograph of the plants affected in June.

A customer climbed onto his roof, wired an anemometer up next to his Atlas so you could see the wind speed as it turned, pointed his phone at it and pressed record. That clip has now been watched 5.4 million times.

Here it is. No music, no voiceover, no edit.

The gap is the interesting part

It is not that the customer clip did well. It is how far ahead of everything else it is.

The next most-watched video on the channel — one of ours, about the low-wind blades — has 282,000 views. After that: 265,000. Then 253,000. Then 210,000. Perfectly respectable numbers for a company that sells wind turbines.

One customer with a phone out-performed our entire back catalogue by roughly nineteen to one — and reached about 340 times as many people as we have subscribers.

We have lighting. We have a factory floor we are proud of. We have people whose job is to make our products look good, and they are good at it. None of it came close.

Why we think it happened

The honest answer is uncomfortable: it beat our videos because they are our videos.

When a manufacturer films its own product working perfectly, you discount it automatically, and you are right to. Of course it works in our footage — we chose the day, the wind, the angle and the edit. Everybody watching knows that, which is why the polish works against us rather than for us.

The man on the roof has no such problem. He paid for it. He has no reason to make it look better than it is. And he did the one thing we would probably have been too nervous to do: he put a live wind reading in shot, so you can judge the output against the conditions yourself instead of taking anyone's word for it.

What people are really asking

We suspect the 5.4 million are not asking whether the Atlas is a good turbine. They are asking something far more specific: would that work on my house?

A studio cannot answer that. A slightly shaky clip of an ordinary roof, in ordinary weather, somewhere that looks a bit like where you live — that answers it. The one thing a production budget cannot buy is somebody else's roof.

So we are going to stop competing with it

The obvious reaction to a customer video beating your marketing is to make marketing that looks like a customer video. We are not going to do that. It would be a lie, and it would show.

Instead: if you own a TESUP turbine and you have filmed it — running well, running badly, being installed, riding out a storm, sitting still on a flat calm day — we would like to see it, and with your permission we would like to publish it. Unedited is fine. Unflattering is fine. It is worth more to the person deciding than anything we can produce ourselves.

If you were one of the 5.4 million

And you are now wondering about the practical side: the payback sum depends entirely on your wind and your electricity price, so there is a calculator on the Atlas page rather than a number here that would be wrong for most of you.

Five million views is a strange kind of feedback. It is not a survey and it is not a sales figure. It is a very large number of people looking at a turbine on somebody else's roof and wondering about their own — and it took a customer with a phone to show them.

Inside the factory

From raw copper to finished turbine — 100% in-house.

The copper stator on the left becomes the finished Atlas on the right — photographed on the same floor, the same week. Drag to see it happen.

Before After

Our factories — Slovakia, Europe ← drag to compare →

We buy metal by the tonne, not by the unit

Copper wire, electrical steel, magnets and housing steel are purchased as bulk commodity orders covering a full production run — not topped up batch by batch as turbines are built.

The difference this makes is larger than most people expect. A small manufacturer buying a reel of copper at a time pays the small-order price, and pays it again on every unit it builds. That premium is baked into the cost of each turbine, permanently. Order the run's copper in one purchase and the premium disappears from every machine that comes off the line — the same metal, the same specification, at commodity pricing rather than retail.

Bulk buying also removes something less obvious: uncertainty. When the material for a production run is already secured, a spike in the copper market does not become a sudden price rise on the website halfway through the year.

Scale compounds, so we made it compound on purpose

The same Atlas generator ships to Ireland, Japan, Brazil, Türkiye and dozens of countries in between. One design, one production line, one spare-parts inventory.

That is a deliberate choice, and it feeds itself. Selling one design everywhere creates the volume that justifies buying metal by the tonne — and buying by the tonne is what keeps the price low enough to sell in that many markets. Manufacturers who build a different variant for each region never reach the volume where either half of that loop starts working.

Nothing sits between our factory and your door

A turbine that reaches a customer through an importer, a wholesaler and a retailer carries three margins by the time it arrives. Ours goes from our factory to our own storefront to your address. That is not clever engineering — it is simply a shorter chain, and every link removed is a cost the customer never pays.

Where the saving does not go

Buying well is only worth it if the savings are not immediately spent cutting corners. Two specifics:

Solid copper windings. The cheap substitute across the small-generator industry is copper-clad aluminium — aluminium wire with a thin copper skin. It weighs less and costs less, and it carries roughly 60% of copper's conductivity, so it runs hotter for the same output and ages faster. We use solid copper. It is invisible in a product photo, which is exactly why it is worth stating.

No gearbox. A 24-pole permanent magnet generator produces usable electricity at low rotational speed, so there is no gearbox between rotor and generator. That removes the component most likely to fail on a small turbine, along with its oil changes and its noise.

Why any of this matters beyond our price list

Renewable energy has a distribution problem more than a technology problem. The engineering to generate power from moderate wind has existed for decades. What has been missing is a machine at a price an ordinary household can reach, in enough markets at once to matter.

Economies of scale are usually discussed as a way for a company to protect its margin. Applied deliberately, they are also the mechanism that moves a technology from a small number of wealthy roofs to a large number of ordinary ones — and small, distributed generation reaches houses that a grid extension never will.

That is the whole strategy, and there is nothing mysterious about it. Buy the metal properly. Build one design for everywhere. Sell it directly. Do not cheapen the parts that decide how well it works.

The slider above shows both ends of the process — a copper stator, and the finished Atlas generator. Both photographed on our own factory floor in Slovakia, one week apart.

What is a permanent magnet generator?

A permanent magnet generator produces electricity using fixed magnets rather than electromagnets that must first be energised. Because the magnetic field is always present, the machine generates from the very first rotation and needs no excitation current, no slip rings and no external power to get started.

TESUP's generator is a 24-pole permanent magnet machine using NdFeB N42 (neodymium-iron-boron) magnets, each 60 × 25 × 5 mm, nickel-coated and bread-shaped. There are 24 of them, 180 cm³ of magnet in total, turning inside a stator measuring 16 cm × 12 cm.

Why 24 poles, and why no gearbox?

Pole count sets the relationship between rotational speed and output frequency. A generator with few poles has to spin fast to be useful, which is why many turbines put a gearbox between the rotor and the generator. A gearbox is also the component most likely to fail, the one that needs oil changes, and the one that makes noise.

Twenty-four poles produce usable voltage at low rotational speed, so the rotor can drive the generator directly. No gearbox means no gear oil, no gear wear, and one fewer thing that can seize on a roof in February.

What is cogging torque?

Cogging torque is the resistance you feel when you turn a permanent magnet motor by hand — that notchy, step-by-step reluctance, as though the shaft wants to settle into fixed positions. It is caused by the magnets attracting toward the steel teeth of the stator. Each tooth is a preferred parking spot, and the rotor has to be pushed out of it.

In a fan or a power tool, cogging is a minor annoyance. In a small wind turbine it is the single most important obstacle to low-wind performance, because cogging torque is a fixed threshold that the wind must overcome before anything turns at all. Below it, output is not small — it is zero.

This is why cut-in wind speed matters so much more than rated power for a domestic turbine. Rated power describes a windy day. Cut-in speed describes every other day.

How does skewing the magnets reduce cogging torque?

If every magnet sits exactly parallel to the stator teeth, all 24 poles reach their attraction peak at the same instant. Those peaks add together, and the result is a strong, notchy pulse the wind has to break through on every step.

Skewing means mounting the magnets at a slight angle to the axis of rotation, so that each magnet engages a stator tooth progressively rather than all at once. The peaks no longer coincide — they spread out and partially cancel. The total torque the generator produces is unchanged; what disappears is the ripple.

TESUP uses skewed magnet mounting for exactly this reason. Lower cogging torque means a lower breakaway threshold, and a lower breakaway threshold is what allows the Atlas to self-start at 2 m/s with its low-wind blade set — roughly one third the cut-in speed of a typical horizontal-axis turbine.

Skewing is not free: it slightly reduces peak flux linkage, so a skewed machine gives up a little top-end output in exchange for starting earlier. For a domestic turbine, where light wind is the common condition and storms are rare, that is the right trade to make.

Why does a turbine that makes about 1 kW use a 15 kW motor?

Because a generator's rating is a thermal limit, not a promise of output. The 15 kW continuous rating is what the machine can sustain indefinitely without overheating, and it is set by three things: the Class 200 °C enamelled wire insulation (IEC 60317-13 GR 2, the same grade used in industrial servomotors and EV drivetrains), the 4 kg of copper winding acting as thermal mass, and passive cooling through the housing.

At the Atlas's typical output of around 1 kW, the motor is running at under 7% of its continuous rating. It runs cold. Copper losses rise with the square of current, so a machine operating far below its limit wastes very little as heat — and heat is what ages insulation and eventually ends a motor's life.

The 22 kW peak figure is the short-term overload capacity for gusts. It is bounded by two physical limits: the demagnetisation threshold of the N42 magnets, and saturation of the stator iron.

Why tooling is what actually decides cogging torque

Skewing only cancels cogging torque if every magnet sits exactly where the design says it should. The cancellation is a geometric effect: 24 magnets, each offset by a precise angle, so their attraction peaks fall out of step with one another. If the seats those magnets sit in vary from one rotor to the next, the skew angle varies with them — the peaks drift back into alignment, cogging rises, and the wind speed needed to break the rotor free rises with it.

That is why ±0.05 mm across all machined components is not a boast on a spec sheet. It is the tolerance the physics requires. A rotor built to loose tolerances can carry exactly the same magnets, the same pole count and the same copper, and still start later in the wind.

Holding that tolerance on one prototype is straightforward. Holding it on every unit, for years, is a tooling problem — which is why this arrived at our facility:

A press tool like this is not bought for speed, though it is faster. It is bought for repeatability. The first rotor formed on it and the ten-thousandth come out of the same steel, located by the same pillars, to the same dimensions. That is the only way a tolerance figure means anything once you are building at volume rather than one at a time.

It also means one more component that no longer depends on somebody else's quality control. Our European facilities already produce over 270 components in-house — generators, electronic cards, blades and mounting hardware. Every tool that lands makes that list one item longer.

Where the energy actually comes from

The power available in wind follows:

P = ½ · ρ · A · V³ · C_p · η

Where ρ is air density, A is swept area, V is wind speed, C_p is the power coefficient and η is electrical efficiency. The Atlas rotor sweeps 1.035 m² (1030 mm diameter × 1005 mm height), reaches a C_p of 0.28 in good wind, and converts at 92% electrical efficiency.

The term that dominates is V³. Wind speed is cubed, so a site with 30% more wind yields roughly double the energy, not 30% more. This is why mast height and siting change results more than any specification on this page — and why an honest answer to "how much will I get?" always begins with a question about your location rather than a number.

Full generator specification

What this does and does not tell you

Every figure above is a property of the machine, and machine properties are the part we control. Cogging torque, pole count, magnet grade, insulation class and machining tolerance are the same in every unit we ship.

Annual energy yield is not. That depends on your wind, and published yield for the Atlas spans 180–2,900 kWh per year precisely because the range between a sheltered suburban garden and an exposed coastal site is that wide. Tell us your location and mast height and we will give you a realistic range rather than a best case.

See the Atlas specification →