Ten years of winding
The first coil was wound on a saucepan lid, driven by a cheap generator off a marketplace, and got warm in the hand after five minutes. It worked — the tester showed a field. But building a second one the same turned out to be impossible: winding by eye gives you a different coil every time.
That is where this started. Below is what broke over the years and what it led to.
The wire
At first we bought ordinary enamelled copper wire from the electronics market. The problem was not the copper but the enamel: on cheap wire the coating is uneven, and in a dense winding neighbouring turns short in places. The coil seems to work, but the current is higher than it should be and it heats up.
Now we buy wire with a controlled coating thickness and check the insulation resistance before winding. A dull step, invisible in the finished product — and the reason two devices of the same model give the same numbers.
The pitch
The second lesson: in a dense winding, «as tight as possible» is not the goal. Turns pressed flat against each other add capacitance between neighbours, and at higher frequencies the coil stops behaving like a coil and starts behaving like a capacitor. The signal sags, the waveform distorts.
We went through several pitches and settled on one where the frequency response stays flat to the top of the range. It is a trade-off: a tighter winding would give a stronger field at the low end and ruin the top.
The power supply
The longest story. The early generators were off-the-shelf modules. They produced a clean sine with nothing connected; put a coil on the output and the oscilloscope showed something with its top sheared off.
We had to build our own. The current generator holds its waveform under load across the whole working range, and that — not the copper — is the expensive part of the device.
The housing
Here we went against convention. It is simpler and cheaper to pot everything in compound: the product looks more solid and nobody sees how it is made.
We left the build open. The winding is visible, the terminals are exposed, you can put an oscilloscope on it. That decision costs us the customers who want a nice sealed box, but it is honest: you see what you are paying for and can check our numbers with your own instruments.
What has not changed
We wind by hand. We tried a semi-automatic winder — on a 14-ray torus it gives a more even pitch, but it loses the ability to guide the wire by feel, and at the crossings that is exactly what matters.
So the batches stay small and every coil passes through someone's hands. That is not romance, it is a limit of the method.
What comes next
Twenty-four designs are on the bench right now; nine have reached the shop. The rest either failed to hold their parameters from sample to sample or turned out to cost more than they are worth.
If you have a winding idea you want tested, write to us. We wind trial coils to other people's specifications happily — it is the cheapest way to learn something new.