Re: Burned Parts on Weed Whip Charger,
December 2012, page 20.
In your December column, you analyzed a small
battery charger in a weed whip. While your circuit
analysis was correct (as always) based on a TO92 NPN
transistor for the pass element, I believe the original part
was an SCR. The give-away was the remaining letters
"MCR." Motorola used to make a line of SCRs in TO-92
packages. The MCR100 series were .8A, and the MCR22
series were 1.5A. These are sensitive gate types requiring
only 200 microamps of gate drive.
Looking at the circuit again with an SCR this time, it
certainly looks more plausible. The zener clamps the gate
to 15 volts. There are two diode drops (VGK and the
pass diode) between this reference and the battery.
When the battery voltage drops to this level (approx
13.6V), current flows through the gate-cathode junction
and fires the SCR for the remainder of the cycle. This
continues at line frequency, until the battery voltage is
pushed up closer to the reference and no gate current
flows. The SCR remains off until the battery voltage falls
again. Now would be a good time to mention that the
circuit will only function with a pulsating DC input! The
input must approach zero to commutate the SCR off
every cycle. So, NO filter capacitors in the wall wart! The
SCR has an on-state voltage drop of around one volt at
400 mA, combined with around a 90° conduction angle;
dissipation should not be an issue.
The main drawback to the original circuit is the
backward (positive) temperature coefficient. The
reference zener voltage rises with temperature on one
side, and VGK with the pass diode’s drops becomes
smaller on the output side. The result is the charging
voltage goes UP with increasing temperature, leading to
overcharging. The opposite scenario leads to
undercharging — exactly opposite of what a lead acid
battery wants for a long life.
I have actually built similar circuits as drop-in
additions to small unregulated
chargers, and given them to friends for
their classic muscle cars. They are
used to float a charged battery
between infrequent drives; see the
schematic. The main difference in my
circuit is the substitution of a TL431
"adjustable” zener for the fixed zener
diode. This allows for a precise voltage
adjustment to match the battery type
(i.e., wet, gel, maintenance free, etc.).
Temperature compensation is
accomplished with the TL431 being
essentially flat, along with D2 in series
with the reference terminal. R5
provides a small current in D2. The
battery voltage is divided down to the
reference by R2, R3, and R4. R1
provides gate drive and operating
current for the TL431. S1 and R6 were
an absorption charge level. The optional "charged"
indicator lights up when the battery is at the set point.
Calibration should be done close to 25C/77F — that is
where many battery manufacturers specify their
charge/float voltages. The tempco isn't exactly textbook,
but it is about negative 10 mV/C — about half of what it
really should be. An interested reader could add another
diode in series with D2 (and fiddle R3), and it should
bring it in.
I hope this helps Mr. Stenlund and the many other
readers of my favorite N&V column.
Thanks for writing, Tim. I had not considered that Q1
was an SCR, but it makes a lot of sense as you describe it.
I am sure readers will appreciate your schematic.
Re: Voltage Converter question, January 2013, page 24:
The reason Q1 shows a four volt drop is that it is not
turned on all the way. The transistor as shown in Figure 1
is an NPN transistor that is located on the high side of
the load. To turn on, it needs more voltage than is on the
collector. By putting it on the low side of the load (i.e.,
the emitter to negative and the load between positive
and the collector), the transistor will be turned on. It will
also need a 1K or so resistor on the base. It would be
easier to use an N-channel MOSFET and not have to
worry about biasing.
I suspect you are right, Mark. I would expect a 1.2
volt drop across Q1 in the best case, and more if the drive
were not sufficient. The circuit is switching, so I stand by
my original answer, but your suggestion to connect Q1
from the load negative to negative voltage is a good one.
The MOSFET idea is also good. Thanks for writing.