LT1461-2.5 Linear Technology, LT1461-2.5 Datasheet - Page 9

• Manufacturer Part Number: LT1461-2.5
• Description: Micropower Precision Low Dropout Series Voltage Reference
• Manufacturer: Linear Technology

The results of slotting the PC boards of Figures 5a and
5b are shown in Figures 6a and 6b. In this example the
slots can improve the output shift from about 100ppm to
nearly zero.
Long-Term Drift
Long-term drift cannot be extrapolated from acceler-
ated high temperature testing. This erroneous technique
gives drift numbers that are wildly optimistic. The only
way long-term drift can be determined is to measure it
over the time interval of interest. The erroneous tech-
nique uses the Arrhenius Equation to derive an accelera-
tion factor from elevated temperature readings. The
equation is:
where: E
To show how absurd this technique is, compare the
LT1461 data. Typical 1000 hour long-term drift at 30 C =
60ppm. The typical 1000 hour long-term drift at 130 C =
120ppm. From the Arrhenius Equation the acceleration
factor is:
The erroneous projected long-term drift is:
For a 2.5V reference, this corresponds to a 0.39 V shift
after 1000 hours. This is pretty hard to determine (read
impossible) if the peak-to-peak output noise is larger than
this number. As a practical matter, one of the best labora-
tory references available is the Fluke 732A and its long-
term drift is 1.5 V/mo. This performance is only available
from the best subsurface zener references utilizing spe-
cialized heater techniques.
APPLICATIONS
120ppm/767 = 0.156ppm/1000 hr
A
A
F
F
K = Boltzmann’s Constant
T2 = Test Condition in Kelvin
T1 = Use Condition Temperature in Kelvin
e
e
A
E
0 0000863
K T
= Activation Energy (Assume 0.7)
A
.
0 7
1
.
1
–
T
1
2
303
U
1
–
403
INFORMATION
1
U
767
W
U
The LT1461 long-term drift data was taken with parts that
were soldered onto PC boards similar to a “real world”
application. The boards were then placed into a constant
temperature oven with T
scanned regularly and measured with an 8.5 digit DVM. As
an additional accuracy check on the DVM, a Fluke 732A
laboratory reference was also scanned. Figure 7 shows the
long-term drift measurement system. The long-term drift
is the trend line that asymptotes to a value beyond 2000
hours. Note the slope in output shift between 0 hours and
1000 hours compared to the slope between 1000 hours
and 2000 hours. Long-term drift is affected by differential
stresses between the IC and the board material created
during board assembly.
Hysteresis
The hysteresis curves found in the Typical Performance
Characteristics represent the worst-case data taken on 35
typical parts after multiple temperature cycles. As ex-
pected, the parts that are cycled over the wider – 40 C to
125 C temperature range have more hysteresis than those
cycled over lower ranges. Note that the hysteresis coming
from 125 C to 25 C has an influence on the – 40 C to 25 C
hysteresis. The – 40 C to 25 C hysteresis is different
depending on the part’s previous temperature. This is
because not all of the high temperature stress is relieved
during the 25 C measurement.
LABORATORY
PCB1
REFERENCE
FLUKE
PCB2
732A
Figure 7. Long-Term Drift Measurement Setup
PCB3
SCANNER
A
= 30 C, their outputs were
8.5 DIGIT
DVM
LT1461-2.5
COMPUTER
1461 F07
9