I originally used 16 turns
on an FT37-43 toroid for about 100uH, but it wouldn't oscillate until I
increased supply voltage from 5 to about 12 volts, and applying the
measurement current would stop oscillation. The frequency was
over 150kHz. I did a new coil using a slightly larger FT37-43 with 50+
turns to give 1.63mH. I wound four turns onto the toroid for the
measurement winding. Frequency would move around (rise) after each
adjustment, either due to some time constant or because current was
changing below the 0.4 A-t level. (100mA on the meter times four
turns.) Supply voltage was 5.07 VDC. In some cases it seems to do
better with higher supply voltage.
The article hinted that this technique wouldn't be repeatable enough
for an accurate current measurement technique. That's probably
true. For order-of-magnitude measurements or detection of
overcurrent, it would be fine. It could also be used in a
feedback loop (PLL) to control frequency using current and inductance
as opposed to the traditional varactor method.
This measurement could also be used to infer the inductance (and
permeability) changes with current. In my example, 4 ampere-turns
approximately doubled the frequency. Since frequency is
proportional to the square root of inductance, the inductance was
reduced by a factor of four (4).
I also experimented with bringing a fairly powerful magnet close to
the toroid and was able to move the frequency up to 200 kHz with it.
The graph below shows how frequency changed with amp-turns through the core
Nick Kennedy, WA5BDU