One project over the last several weeks has been winding coils for the atomic fountains. There are two different requirements, one is the so-called “C-field” coils and the second is the MOT (magneto-optic trap) coil pair.

The “C-field” is the bias field in an atomic clock that essentially tells the atoms which way is up, i.e. it defines the quantization axis. It also shifts the frequency of the transition, so in a frequency standard you need to know what the field is. In a clock (there is a subtle difference) you care about the stability, i.e. you don’t want it to change, so it’s enough to feed this with a precision current source to give a bias field of a milligauss or so. Two layers, up and back, so the pitch on each layer should tend to cancel and leave you with a vertical field, and about 600 turns per layer. There are also extra shim windings at each end to better simulate an infinite solenoid — a real solenoid’s field drops off at the ends, so we boost it back up a little. The drift region, where the atoms oscillate between the two hyperfine states (the “tick” of an atomic clock), sees a very stable field.

Pretty easy, but time-consuming (as it were); the basic winding took more than four hours. What you see is the jig I used, which has a stepper motor and a home-built feed system that wets the wire with alcohol to activate the bonding material. Square wire is used so it doesn’t have any gaps.

The MOT coils, obviously, are for trapping the atoms; they’re arranged to give a quadrupole field, i.e. one that is zero at the center and increases ~linearly as you move away, at least when you are close to center, and mounted on the trapping chamber. These have to be compact, so there are several layers of wire. They are shown here, along with a beautiful titanium vacuum chamber, stunningly contrasted with sable optical couplers.

These turned out to be a bit of a pain. Wire doesn’t like to bend at 90 degrees, so the bottom a layer always had an extra bulge that propagated through the coil. And square wire is worse — when it bends around a corner, it has a tendency to twist 45 degrees (which must minimize its energy) into a diamond profile, and this made the bulging worse and two-dimensional. A couple of attempts had the winding error become bad enough that there were gaps, so I scrapped those and started over. I finally re-drilled the entrance hole (by hand, so I didn’t ruin the coil form) to be at ~45 degrees so the bend wasn’t as severe, and that improved the situation.

When the MOT coils were wound I had to double-check that the number of turns was equal, so that the field zero would be at the mid-point. This turned out to be a little tougher than I expected, because our fluxgate was too sensitive and our Hall probe was analog and didn’t give a stable enough reading on the meter. Then I remembered we have a clamp-style Ammeter that even measures DC (though I was prepared to use AC if it didn’t), so I was able to run an amp through the coils and see if the currents were the same or differed by an amp, and unwind the offending coil one turn to equalize them.