LINEAR TECHNOLOGY LTC6652 Technical data
Reliable Precision Voltage Reference
OUTPUT VOLTAGE (V)
2.4985
0
NUMBER OF UNITS
40
60
80
180
140
2.5005
20
160
120
2.4995 2.5015
LTC6652A LIMITS
1004 UNITS
DRIFT (ppm/°C)
0 0.5 1.5 2.5 3.5
0
NUMBER OF UNITS
15
20
5
10
30
25
1.0
2.0 3.0
TEMPERATURE (°C)
–50
2.496
V
OUT
(V)
2.500
2.498
2.504
2.502
0
50 100 150
GUARANTEED
with 5ppm/°C Drift is Factory Trimmed
and Tested at –40°C, 25°C and 125°C
Introduction
High precision requirements are no
longer limited to the most exotic and
expensive measurement equipment.
Designers of industrial monitors
and automotive monitor and control
circuits are using precision circuits
to maximize the performance and
uninterrupted operating times of their
products. Improved precision allows
for more accurate assessment of sensor outputs that measure ambient
conditions, equipment position, battery condition, component wear and
many other system indicators. Precise
and consistent measurements are the
key to managing system elements that
change very little over their operating
lives. Recognizing these slight changes
can allow estimation of the remaining
lifetime of lamps, motors, and other
components, or allow control of battery charge and discharge to maximize
operating life. These applications not
only require high accuracy, low drift
and low noise, but also a wide operating temperature range and reasonable
cost.
Factory Calibration Means
Dependable Precision
The LTC6652 reference is a precision
low drift voltage reference that includes
advanced curvature compensation
circuitry and post-package trim. To
guarantee reliable performance, these
parts are tested at –40°C, 25°C and
125°C to verify they meet specification
across the entire temperature range.
This comprehensive testing ensures
that the LTC6652 can be used with
confidence in demanding applications. One result of this testing is
demonstrated in Figure 1. The output
voltage versus temperature for several
randomly chosen parts shows a drift
characteristic that is consistent from
part to part. This is a result of a propri-
Linear Technology Magazine • January 2009
DESIGN FEATURES L
by Michael B. Anderson and Brendan Whelan
etary curvature compensation circuit
that tracks the operating conditions
and the manufacturing process, yielding consistent results. Figure 2 shows
a typical temperature drift distribution of randomly selected production
tested LTC6652s, illustrating how well
the design and testing methodology
works. Finally, the initial accuracy
distribution is tightly controlled, as
shown in Figure 3.
Compare the Real Specs:
Is the Temperature Range
Operating or Functional?
When comparing voltage references for
use in demanding environments, it is
important to know, with confidence,
how the voltage reference performs at
the extremes. When it is important for
the reference to provide precision (not
just survive) in a harsh environment,
the LTC6655 leaves most competing
voltage references behind.
For example, many applications
requiring a precision reference are
designed to work over the industrial
temperature range (–40°C to 85°C).
If the ambient temperature reaches
85°C, the interior of the enclosure and
the temperature of the reference will
Figure 2. Drift distribution (–40°C to 125°C)
Figure 1. Typical drift characteristics of
production trimmed and tested parts
likely exceed 85°C. It is not uncommon in this case for the interior of a
circuit enclosure to climb above 100°C
due to the thermal dissipation of its
components. In addition, any comparable voltage reference fully loaded at
5mA with a 13.2V input voltage would
self-heat an additional 18°C, raising its
own internal junction temperature to
118°C. This temperature is well outside the useful range of most voltage
references. The LTC6652, however,
maintains exceptional performance in
these conditions, despite the extreme
environment. By comparison, the drift
of a reference specified only to 85°C will
Figure 3. Typical V
for LTC6652-2.5
distribution
OUT
7
L DESIGN FEATURES
DISTRIBUTION (ppm)
–250 –150 –50
0
NUMBER OF UNITS
5
10
15
35
25
30
20
15050
125°C TO 25°C –40°C TO 25°C
TIME (HOURS)
0
LONG TERM DRIFT (ppm)
160
3000
80
120
–120
–80
40
1000 2000 50004000 6000
–160
–40
0
LTC6652-2.5
likely exhibit substantial additional
error, or it may even fail to operate.
Other references that claim similar
performance to the LTC6652 often are
only “functional”, meaning they don’t
fail, but they don’t meet specification
either at temperatures exceeding 85°C
or below 0°C. These competing parts
are rarely tested across their entire
specified temperature range. The reality of industrial circuit design is that
in many cases, component specifications over “industrial” temperatures
are woefully inadequate.
In contrast, every LTC6652 is fully
tested at 25°C, –40°C and again at
125°C for every device, not just a
small sample. This extensive testing
proves the consistently high quality
of the LTC6652 over its entire wide
temperature range.
Further, the LTC6652 was designed
from the ground up to accommodate
a wide temperature range. Figure 1
clearly illustrates its consistent performance over the entire range. There
is no need to question or derate the
performance of a system that uses
the LTC6652 at its temperature extremes. The consistent, guaranteed
performance makes it easy to design,
specify and calibrate a high performance system. This is not the case
with some competing products.
Eliminate Field Calibration
After any precision reference is soldered onto a printed circuit board,
thermal hysteresis will likely shift the
output from its factory trimmed value.
Further temperature cycling will also
contribute to a shift in the output voltage. Over time, the output will tend to
drift slightly as well due to aging of the
circuit. The circuit design, fabrication
process and mechanical design of the
LTC6652 is optimized for low thermal
hysteresis and excellent long-term
stability, reducing the need for field
calibration. Thermal hysteresis is
caused by differing rates of expansion
and contraction of materials within
a packaged semiconductor device as
the device experiences temperature
changes. As the package material and
the semiconductor die expand and
contract at different rates, mechani-
8
Figure 4. Hysteresis plot (–40°C to 125°C)
cal force changes device parameters
(ever so slightly) and cause the output
voltage to change. This change is
measured in parts per million (ppm)
and is shown in Figure 4.
The LTC6652 boasts a typical
thermal hysteresis value of 105ppm
over its full temperature range. What
this means is that when a device goes
The LTC6652 reference
family is designed and
factory trimmed to yield
exceptional drift and
accuracy performance. The
entire family is guaranteed
and production tested at
–40°C, 25°C and 125°C
to ensure dependable
performance in demanding
applications. Low thermal
hysteresis and low long-term
drift reduce or eliminate the
need for field calibration.
Figure 5. Example of long-term drift
from room temperature to 125° and
back again to room temperature, the
output will typically shift 105ppm.
For a 2.5V voltage option, the output
would shift –260µV. A cold excursion
would shift the most recent room temperature measurement +260µ. Typical
hysteresis of 105ppm is equivalent to
0.0105%; just a small fraction of the
initial accuracy.
It may be convenient to compare
typical values for thermal hysteresis
when choosing a voltage reference.
It is important to remember that
these numbers do not tell the whole
story. It is the distribution of expected
hysteresis that must be used to determine the expected error caused
by temperature cycling. Referring to
Figure 4, some parts will have better
or worse hysteresis. This chart helps
to estimate a realistic error budget.
This is something our competitors
don’t always include, yet is critically
important when specifying precision
systems.
Over time a reference is likely to
shift on its own even if it’s kept at a
constant temperature. This is known
as long-term drift. Long-term drift
is measured in ppm/√khr and has
a logarithmic characteristic where
the change in output voltage decays
as time passes. Figure 5 shows the
long-term drift of the LTC6652. Note
that most of the drift occurs within
the first 1,000 or 2,000 hours as the
PCB and package settle. Afterward
the drift tends to settle, and the slope
is reduced over time as a function of
khr. Direct measurement on soldered
√
down parts is the only way to determine
Linear Technology Magazine • January 2009
Loading...