version assumes that the op-amp’s
output and supply currents change in
step and that the driver stage has a
well-defined gain. It uses modified
PNP and NPN current mirrors.
Each power transistor (Q2 and
Q4) has a large emitter resistor that
drops over a volt at the current
peaks. These transistors are driven by
opposite polarity emitter followers so
the base-emitter voltages of each pair
roughly cancel. The drivers’ bases are
connected to the amplifier’s power
pins and to the power rails via 51 ohm
resistors (R12 and R13).
The current gain of this combination is 23. Apart from about 3 mA
of static current, the amplifier’s
power currents equal the current
flowing through its combined
100 ohm load. That is, a 100 mV
amplifier input generates 23 mA at
the driver output. SW1 selects a
suitable drive voltage for each
current range. C6 reduces noise and
stray feedback and makes the current
ramp much cleaner.
The op-amp’s static current
causes the output transistors to pass
nearly 100 mA when the output
current is zero. I tried reducing
this, but only introduced cross-over
distortion. We must just live with
about 1.6 wasted watts. The
regulators’ and transistors’ peak
dissipation is around 5W, but no
heroic heatsinking is required.
The output transistors’ base
currents are adjusted to balance out
offsets. Short the inductor terminals
and set the switch to its grounded
input position. Measure the voltage
across R18 with a multimeter on its
200 mV scale and adjust VR1 until
the meter reads zero. The specified
output transistors have current gains
exceeding 150 at 500 mA. Many
popular power transistors have lower
gains and won’t work unless R14,
R15, and VR1 are reduced to supply
more base current.
■ FIGURE 5. The unboxed
B-H tester board.
pliance, so bipolar 5V power
rails might have sufficed.
However, at a current peak
the input to the current mirror is 1.3V less than the positive or negative power rail. If
the main power were ±5V,
the net supply to the op-amp
would be about 8.5V. Since
the TL082 is rated for 10V
minimum, I boosted the
power rails to ±8V to avoid risking odd
non-linearities. This means hunting
down 7808 and 7908 three-terminal
regulators. By the way, I used 2,200 µF
reservoir capacitors for space reasons,
4,700 µF ones would be better.
One end of the inductor under
test is grounded. This makes measuring the voltage across it easy, but you
need a floating current-sensing resistor
(R18). A differential amplifier (U5a)
drives the scope’s X input with a scale
factor of 5V per amp. The inductor
voltage goes to the integrating
amplifier (U5b) whose output drives
the scope’s Y input. As mentioned
above, an adjustable amount of the X
voltage is fed to the integrator input
to compensate for the inductor’s
The control switch has an off
position that shuts off the ramp
generator. Ideally, it should be put in
this position before removing the
inductor under test. The
output voltage swings
from one power rail to
the other if you don’t.
plastic or metal box at a later date. I’d
already cut a 2. 5” by 4. 6” board, so I
juggled things to fit. To simplify
making changes, I used wire-wrap
sockets for all the small components.
This adds considerably to the board
height and you might be more
comfortable using PC sockets and
sleeved bus wire.
The back edge of the board is
fitted with a length of 0.5” by 0.75”
aluminum angle. This supports the
power input socket and the two BNC
sockets that connect the unit to the
oscilloscope. It also acts as a heatsink
for the power regulators and the two
output transistors. If you put the board
in a case, this angle fits flush against
one wall so use a metal box.
Power comes from a 9 VAC
wall transformer. Mine is an old
RadioShack one rated at 780 mA.
These ratings are about the minimum
acceptable for this application. If
you use a 12 VAC or higher voltage
transformer, you’ll need a bigger
I built this project on
a piece of perforated
board (see Figure 5),
intending to put it into a
Why Eight Volts?
The current driver doesn’t need
more than about +-2V of output com-
■ FIGURE 6. As suspected,
my metal tape cores show
a square loop.
June 2007 49