Bobbin Probe Pull

Pull the probe through the calibration standard and watch each window build the signal. Change the frequency to see what lower frequencies do.

Calibration standard · side cutaway

Pull: cable leads, probe travels left to right

Eddy current into the pageOut of the pageCrowding around a flawField reach (no current in air)Brightness = current density, fading with depth

2.000 in

Animate AC field Speed

Frequency Tube

Coil spacing 0.060 in Wear scar depth 40% Wear scar length 0.25 in

400 kHz

Coil A alone

Diff verticalDiff horizontalAbs vertical

Each frequency normalized: 4×20% FBH = 4.0 V (diff), 100% TW = 40°

Indication	Diff Vpp	Diff angle	Depth from curve	Abs Vpp

Values fill in once the probe has passed each one. The depth curve is rebuilt from the cal flaws at each frequency.

Stop just past the 100% hole. The first loop is drawn. Keep pulling and the second loop forms on the other side of null.
Stretch the wear scar past the coil spacing. The differential splits into two small signals at the scar's ends while the absolute stays one hump.
Park coil A under the 60% hole. The eddy currents can't cross the hole, so they crowd through the thin ligament under it and glow white.
Drop to 25 kHz and pull past the ID groove and the support ring. The ID groove nearly disappears and the support takes over the screen.

400 kHz · Inconel 600

Solid: this frequency. Dashed: 400 kHz Inconel 600.

Lower frequency goes deeper. Standard depth of penetration grows as 1/√frequency, so the field reaches the OD and past it.
Deeper field, less phase spread. Phase lag builds with depth measured in penetration depths. When the whole wall is a small fraction of one penetration depth, flaws at different depths land at similar angles and the curve bunches up.
High frequency favors the ID. Current crowds the inside surface, so ID flaws read strong and the angles spread out for sizing.
Low frequency favors what's outside. Supports, deposits and OD conditions grow, which is why the low channel is for locating supports and feeding the mix, not sizing.

Teaching model: simplified coil response; angles and amplitudes are illustrative. Not for sizing real data.