be 10 mV (5/5/100) and the current
through the diode will be 1 mA. I
designed the filter for a 100 μs period; you could make C1 larger and
use a longer period but you have to
wait five time constants before taking
the Vtp measurement. With 1 μF, the
time constant is: R1*C1 = 0.1 Meg *
1 μF = 0.1 second. You must wait at
least 0.5 seconds for good accuracy.
I was going to run it from the five
volts of the micro supply, but realized
that there is not enough head room
to measure a blue LED, so that is
why VCC = nine volts. When
measuring a red LED, Vtp is about
three volts. That leaves six volts drop
across Q1, and at 100 mA the power
dissipation is 0.6 watts. That is a bit
too much for a 2N3904, so I am
recommending 2N2219. I don’t think
you will need a heatsink.
SOLENOID DRIVER
QI need help making a circuit
that activates a solenoid
with a switch. The amount
of time that it stays activated needs to be adjustable and it
needs to activate as fast as possible.
Can you help me?
— Dalton Jager
AThis circuit (Figure 7) will
work with input from 5 to
15 VDC. I had to make
some assumptions about
the solenoid: operating voltage is 12
VDC; current draw is one amp.
Most mechanical relays will pull in
15 milliseconds, so I used that
number for the time the solenoid
■ FIGURE 7
■ FIGURE 8
will have to pull in.
Assuming that the solenoid will
stay pulled in down to eight volts
input, then the capacitor to energize
the solenoid should be: C = I*dt/dV
= 1 amp* 15 ms/4V = 4000 μF. You
can use this same formula if you have
a different solenoid.
The way this works is: The 555
turns on Q1, which turns on Q2. Q2
pulls C1 up to double the voltage on
the solenoid. When the charge on
C1 is depleted, current flows through
D1 to keep the solenoid pulled in as
long as the 555 output stays high. I
would use a fast diode for D1, like a
1N4936. For Q1, use Si3442DV or
similar; for Q2 use TIP42.
TELEPHONE LINE
MONITOR AGC
QI need an audio amplifier
that will provide an equal
level of output for both
sides of a telephone
conversation. I searched the web for
a schematic or a kit that
will provide automatic
gain control in this kind of
application, but could not
find one. We currently
use a 600:600 ohm audio
transformer (in series with
capacitors) bridged on the
POTS line on the input
side and an LM386 on
the other side of the
transformer. This allows
monitoring (and/or
QUESTIONS & ANSWERS
recording) of both sides of the conversation for quality control and order
confirmation purposes. However,
many times the caller can barely be
heard while the agent’s voice is
extremely loud. Can you suggest a
circuit that will output both sides of
the conversation at the same level?
— Stan Grupinski
AIn the circuit in Figure 8, the
2N3904 is acting as a
variable resistor. It is very
non-linear so it is necessary
to keep the signal level small; in this
case, about 10 millivolts. The amplifier
gain is 101 so the output is about
one volt p/p. R4 and the capacitors
C1 and C2 are a low pass filter to
provide some averaging and not
respond to peaks. This slows the
attack time but the attack time is
about 200 ms, which should not be
objectionable. R5 discharges the
capacitors so that low level signals
can be attenuated less. The decay
time is about 300 ms. I assumed
that you were using a single supply,
so used IC1B to provide VCC/2 and
prevent clipping of the signal. The
circuit should work from 5 VDC to
12 VDC.
PROTECTION DIODES
QI have an H-bridge driving a
motor at 310 VDC and 3. 5
amps. The internal body
diode of the FET is rated
535 ns recovery time. I think that is
December 2008 27