BY DOUG MALONE
■ FIGURE 2. 4%
accurate, short-circuit proof, 5V
current flowing through the device?
Well, fortunately, things are not as
critical as they might first appear. The
reverse impedance of the zener is quite
low, which means that the current can
vary quite a bit and the voltage will still
remain relatively constant. As part of
the research for this article, I tested
five 1N751A zeners from Fairchild
Semiconductor. I used a 200 ohm, 5%
series resistor whose measured value
was 197 ohms. When I varied the supply from 8V to 10V, the current through
the least accurate zener of the group
varied from approximately 14. 3 mA to
24. 3 mA, but the zener voltage stayed
within 2.3% of 5.1V. The best of the five
devices was within 0.2% of 5.1V! Not
bad for such a simple and inexpensive
component. Your mileage may vary.
If the current through the device
varies from 1 mA to 15 mA, the
voltage drop is guaranteed to change
a maximum of an additional 10 mV.
Figure 3 shows a typical circuit.
So far, accuracy has been stated as
a percentage of the nominal voltage
with no additional qualifying conditions.
One such condition is temperature.
Unfortunately, changes in ambient temperature will cause a change in the reference voltage, regardless if the device
is a zener or a precision reference.
For instance, the LM4040 DIZ- 5.0
has a worst-case temperature coefficient of 150 ppm/°C. This means that
for every degree up (or down) from
25°C, the breakdown voltage of the
device will change by 150/1,000,000
or .015%. Obviously the smaller the
temperature coefficient, the better.
For general indoors lab work, the
temperature coefficient will probably
not be as important as it would be if
the voltage reference were exposed to
more extreme temperatures. Your
specific application will determine if
you need to use a device with a low
temperature coefficient. As you
might suspect, devices with a low
temperature coefficient tend to cost
more than those with a higher
■ FIGURE 3.
Next up, we have the common
7805 three-terminal regulator. The
MC7805 from Fairchild (and others)
costs about $. 50, gives 4% accuracy
under light load, is short-circuit proof,
and can more easily accommodate
wide swings in load current than
the simple zener regulator. Higher
accuracy devices are available such as
the 2% KA7805AE for $. 80. Figure 2
shows a typical application circuit.
For 1% accuracy, you might want
to consider the LM4040DIZ- 5.0 from
National Semiconductor. This device
costs about a dollar and is a “shunt”
reference similar in general concept to
a zener except that it isn’t nearly as
picky as to the current flowing
through it. With a 1% accurate device
at room temperature, the output
voltage is 5.00V plus or minus 50 mV.
usual, you get what you pay for. This
increased performance will approximately double the cost to a little over $2.
Next, we can consider the Analog
Devices AD586L shown in Figure 4.
This device provides 5.000V within
2. 5 mV (.05% accuracy) and has a low
temperature coefficient of 5 ppm/°C,
an aging rate of 15 ppm/1,000 hours,
and a supply current of 3 mA. It will
cost you a little over $9, however.
At this point, I need to explain a
little about aging rate — also known as
long-term stability — specifies how the
reference voltage will change over an
extended period of time. In other words,
even though the temperature, applied
voltage, and load current may be constant, all voltage references will slowly
drift over time. A 15 ppm/1,000/hour
aging rate means that after 1,000 hours
(about 42 days), the voltage can be
expected to be 15/1,000,000 or .0015%
higher or lower than its original value.
Now for the pièce de résistance!
Intersil has recently come out with a
series of voltage references that significantly improve upon the historical
Going Up the
Next up on the accuracy
scale, you could try a
LM4040AIZ- 5.0; this device is
accurate to 0.1% and has
a moderate temperature coefficient of 100 ppm/°C. Now we
are getting somewhere! But, as
■ FIGURE 4. .05%
accurate, 5 ppm,
September 2007 53