What my engineer is actually doing to a box of light

There's a working Helder 2 on Arjen's bench right now. And when I say working, I mean the whole thing — the enclosure, the heat sink, the fans, the circuit boards, all of it, mounted on the arm exactly the way you'd use it. Arjen built every piece of it himself, on his own machines. This isn't a rough breadboard with wires hanging off it. It's a real, complete device that happens to be the first of its kind in the world.

The only thing it isn't yet is finished. The industrial design — the final shape and finish that'll make you want one on your desk — comes later. What Arjen has made is the engineering truth of the product: everything that decides whether Helder 2 works, is safe, and does what we promised. The pretty part gets layered on top once we know the hard part is right.

And now he's testing it. When I tell people that, they usually picture him flicking it on and going "yep, it lights up." That's not it at all. So I wanted to write down what's really happening, because I think it's kind of beautiful, and because it explains why we do things in the order we do.

The whole point is the light you can't see

Helder 2 is a wellness light. Most of what it puts out is near-infrared — a wavelength your eyes physically cannot see. There's a warm amber glow too, but the main event is invisible.

That one fact changes everything about testing. Here's why: when a normal light is too bright, you squint. You flinch. Your body has a built-in alarm. But there's no alarm for infrared. You could be sitting in front of a lot of it and feel nothing at all, because there's nothing to feel. Your eyes don't know to look away from something they can't see.

So we can't judge this thing by looking at it. "Seems fine" is worthless here. The only way to know what's coming out of it is to measure it. And that's really what Arjen is doing at that bench — turning something invisible into hard numbers.

Weighing light on a scale

The first real test uses a device I love explaining: an integrating sphere. Picture a hollow ball, white on the inside. You put the light inside, and the ball bounces every single photon around until a sensor can count all of it. It's basically a scale for light. You put light in, you get a number out: this is exactly how much, at exactly these colours.

Then he takes a second instrument and measures the light at 20 centimetres — the distance you'd actually hold it from your skin — because that's the number that matters for a person, not the number at the surface of the device.

Until those measurements exist, we genuinely don't know how strong our own light is. Everything else waits on this. It's the key that unlocks the next door.

The door it unlocks: is it safe?

Once we know precisely how much light lands on skin over a session, we can answer the question that actually matters: is that safe? There's an international rulebook for exactly this — how much light is fine on skin and near eyes, and for how long. With real numbers in hand, Arjen can check our 20-minute session against that rulebook and see how much margin we've got.

Notice the order. We can't declare it safe and then measure. We measure, and the safety answer falls out of the measurement. That's the whole reason to test now rather than guess.

Everything else he's poking at

While that's going on, he's running the light through a gauntlet of other checks. In plain terms:

Does it get too hot, and does it protect itself? He heats it up on purpose and watches whether the fan kicks in when it should, and whether it shuts itself off before anything gets dangerous — even if he sabotages the fan to see what happens in a fault.

Does the light wobble? Cheap lights flicker in a way that's too fast to consciously see but that your brain still notices — it's part of why bad office lighting gives people headaches. He measures whether ours is rock-steady. We want it boring, in the best way.

Is the power side safe and well-behaved? It runs off USB-C. He checks it plays nicely with normal chargers, that there's no dangerous voltage anywhere you could touch, and that if something shorts or spikes, it fails safely instead of dramatically.

Does it interfere with your other gadgets — and does it have any radios? He checks it isn't spewing electronic noise into your Wi-Fi, and confirms what's deliberately not there: no Bluetooth, no Wi-Fi, no wireless anything. It's a light, not a spy.

Does it actually do what the buttons promise? The modes, the timers, the tap-versus-hold — he confirms Body mode really runs 20 minutes and stops, Head mode really runs 10, and every control does the one thing it says.

"Did we build it right" vs "did we build the right thing"

There's a distinction I've come to really like. There are two different questions you can ask about a product.

One is: did we build it to spec? Did the thing we drew on paper come out true in the metal? That's measurable — it's numbers against targets, and it's almost everything Arjen is doing right now.

The other is: did we build the right thing? Is it actually pleasant to hold against your face, easy to position, nice to live with? That one you can't answer with instruments — you answer it with real people using it, once it's wearing its final industrial design. Arjen's prototype already positions on the arm and works exactly as intended, so we'll learn a lot from living with it. But the formal "is this lovely to use" verdict waits for the finished form. It's coming, just not at this stage.

I think that's an honest way to work: be precise about what you can prove today, and don't pretend to have proven the rest.

Why test before it's finished

It'd be tempting to wait until Helder 2 has its final look before testing it. But almost none of this depends on the final industrial design. The light is the light. The heat is the heat. The safety numbers don't care about the colour of the finished plastic.

And here's the thing that makes this prototype special: because Arjen built it complete — real enclosure, real heat sink, real fans, real boards, all on the arm — the numbers he gets are real numbers, not lab approximations. The heat flows through an actual heat sink. The fans move actual air. When he measures how hot it runs, that's how hot it runs. A half-built rig would give you half-trustworthy answers; a fully built one gives you the truth.

So we do the hard, invisible, number-heavy work now — while changes are still cheap and nothing's been locked for production. Every test gets written down before it's run, measured properly, marked pass or fail, and recorded. Not "looked good on the day." Written down, with the evidence attached.

That's what's actually happening on that bench. One person who designed and machined an entire working device from scratch, now turning its invisible light into numbers, and using those numbers to make sure that when Helder 2 eventually lands on your desk in its final form, the boring, invisible, important stuff was true all along.