In my message to Yahoo MEG-Builders #1207 dated 03 NOV 2003,
I reported the results of an experiment to measure the magnetic
field inside the output coil at locations on the inside and
outside of the core opening. That experiment indicated that
there is a large difference in the magnetic field based on
location which I attribute to the difference in magnetic-vector-
potential (A) at those locations. Those findings prompted me to
want to know if there are differences in even smaller regions
around the output coil since there will be differences in the
gradient of A for small differences in location. Hence I built
another setup using ferrite inductors as sense coils spaced
evenly around the circumference of the core leg. I used short
inductors to reduce coupling to adjacent core surfaces at the
top and bottom of the core opening. A drawing of the test setup
is in the file "ACorTst2.bmp".

Go to "Files" then go to the folder "MESSAGE ATTACHMENTS", go
to the folder "Magnets Make a Difference", and open
"ACorTst2.bmp".

The core is again a Honeywell AMCC-320, cut core. Honeywell
cores can be purchased from Eastern Components,
www.eastern-components.com. A layer of electrical insulating
tape 0.020-inch thick is placed on the core. On this tape are
placed 31 inductors side-by-side, centered on the core leg.
These inductors are Jeffers Electronics part number 19A472-J,
and are 4.7 mH, 0.194-inch diameter, and 0.450-inch in length.
Similar inductors can be purchased from Digi-Key
(www.digikey.com) and other electronics distributors. There is
nothing special about these inductors, I had a package of them
in my "junk" box and used them. A layer of insulating tape
covers the inductors and holds them firmly in place. Over the
insulating tape is placed a single layer of 0.002-inch brass
shim stock to provide an electrical shield between the
inductors and the voltage in the output coil. The ends of this
brass shield are insulated to prevent it from becoming a
shorting turn which would be a low-resistance block to any
changing magnetic field in the core or output coil.

Two layers of insulating tape 0.0022-inch thick are placed
over the brass layer, followed by 6 layers of 0.0015-inch teflon
tape commonly used by plumbers. The teflon layer reduces the
capacitance between the first layer of the output coil and the
brass shield. Such capacitance would reduce the resonant
frequency of the output coil.

The output coil is wound using #24 enamel-coated magnet wire
wound bifilar (two wires side-by-side) creating a large
capacitance between the two wires. The layers are wound 12
turns of this bifilar configuration per layer, for a total of
five layers and 112 bifilar turns. Four layers of teflon tape
0.0015-inch thick are used to insulate between layers. When
the end of one wire is connected to the beginning of the other,
the output coil is a total of 224 turns and measures 314 mH.
The capacitance between the two wires is 4.83 nF, and the
series resonance of the output coil with the nearest drive-coil
is 33.5 kHz.

On the other core half is an output coil of 224 turns wound
similarly, but not identically, whose series-resonance with its
drive-coil is 29.5 kHz. Having two resonant output coils
provides some balance to the changing flux in the core.

A resistor of 12k ohms was placed on each output coil. A
filter is placed between the output of the 10x probe (used to
measure the voltage on the inductors) and the oscilloscope
input. The 3dB cut-off frequency of this filter is 90 kHz.
Its purpose is to reduce the "ringing" voltage of the
inductors (about 300 kHz) to make it easier to measure the
voltages on the oscilloscope. This ringing is the response
to the sharp changes in flux induced by the square-wave drive
pulse.

There are two drive circuits, each composed of two MOSFETs
in a half-bridge configuration so that the full supply
voltage can be applied across the drive coil. Each drive
circuit produces a half-square wave to its respective drive
coil to provide large values of dB/dt in the core, and
consequent large values of dA/dt outside the core.

The drive logic turns on each drive circuit for about 49%
of the total period. Thus there is no overlap of drive
signals, and during most of the time there is drive to the
core. Providing a small amount of "off" time allows the
core to discharge if there is any asymmetry in the applied
drive.

The voltage at the output coil was kept constant at 1,000
volts peak-to-peak for all measurements. This level of
output provides a reasonable amount of dB/dt in the output
coil, encourages the formation of surface charge and
requires about five watts of drive power.

Measurements were made during a single test session to
reduce any possible changes due to changing environmental
conditions (temperature, humidity, etc). These conditions
will slightly change the response because the metal of the
core will expand or contract which changes the coupling
between drive coil and output coil, resonant frequency,
and related effects.

For the results posted here, the measurements were made at
a fixed frequency of 31.75 kHz without and then with the
magnets in place. Because the permeability of the core changes
slightly with the large magnetic field of the Neodymium magnets
used, other measurements were made at the different resonant
frequencies without and with magnets. The results were similar
to those reported here.

With the magnets in place, a gauss-meter was used to measure
leakage flux near the output coil. In all areas, especially
those closest to the magnet stack, the static field was
measured at less than 100 gauss. When the magnet stack was in
place, prior to placing transformer laminations to eliminate
the air gap between magnets and core, placing the probe in that
gap showed a magnetic field greater than 8,000 gauss. The magnet
stack is 1.5-inch in length, 0.75-inch width, and height of 1.25-
inch.

The drawing, "ATstRslt.bmp" plots the results of my
measurements versus location on the core leg. As can be seen,
there is a significant difference in measured B field based on
the location. Of particular note is that with the magnets in
place, there is a significant difference in the measured flux in
the inside core space whereas there is almost no change for the
measured flux outside the core space. In the inside core space
there is much more A, and a large gradient in A, as detailed in
my message #1207.

This drawing is done so that if the voltage were zero, the
test point would occur on the dotted line surrounding the core.
The location of each test point is directly proportional to
its measurement. Thus the drawing becomes a contour of flux
measurements versus location on the core leg. It is
interesting to note the large difference between the sense
inductors near a corner of the leg, and those in the center of
the leg. The results are not perfectly symmetrical probably
due to slight differences in placement of the inductors (they
may not be the same distance from the core surface due to
irregularities in the insulating tape, and perhaps forced up
and away from the core due to force from their connecting
leads). Also, there are slight imperfections in the bifilar
winding, sometimes the adjacent wires do not lay against one
another, slightly changing the circumstances for surface
charge (whose motion through the gradient of the permanent-
magnet-induced A is what I believe is responsible for these
results).

In message #1207, I noted that I had observed, but not
quantified, the difference when the magnets were placed in
the core. Also, I had not observed much difference in that
test between an inductor at the center of the leg, and one
placed near the edge. Here, obviously, with better resolution
of location, and more attention to the measurements,
differences are very clear.

As noted earlier, these tests were run under different
operating conditions and similar or greater differences were
noted without and with the magnets in place.

Several questions arise:
1. What is the back-reaction to the drive circuit from
this increased magnetic field in the output coil ?
2. What should be done with this increased magnetic
field ?
3. How can this effect be increased ?

More exploration, hopefully more discovery :-) !!

David J.

Files:
ACorTst2.bmp
ATstRslt.bmp