MCP651 Input Offset Evaluation Board User’s Guide
DS51834A-page 8
©
2009 Microchip Technology Inc.
Analysis of this simplified circuit gives the following nominal circuit outputs:
EQUATION 1-2:
R
1
and R
2
(R
12
) balance the circuit at the DUT’s input. These resistors are small, and
are oriented on the Printed Circuit Board (PCB) to cancel their thermoelectric voltages.
The parallel resistances R
1
||R
2
and R
3
||R
4
are equal to minimize the contribution of the
DUT’s input bias currents to the measured V
OST
(contributions by R
5
through R
8
do not
affect V
M
); the typical value of I
OS
at +125°C is ±100 pA, which produces a change in
V
OST
of ±0.02 µV.
The unity gain buffer (+1 gain on the bottom right) isolates the V
COX
input filters from
the following attenuator and integrator. Although it’s not shown here, the resistor R14
at the input to the “+1 Buffer” ensures its output voltage is 0V when the V
COX
connector
is left open.
The attenuators (1/G
INT
) scale V
COX
and V
OUTX
so that they do not overdrive op amps
U2 and U3 (“+1 Buffer” and (“Integrator”). For instance, when V
OUTX
= 5.6V (given
V
SSI
=0.3V and V
DDI
= 5.8V), the voltages at the outputs of the attenuators (1/G
INT
) is
1.80V.
The differential integrator accumulates the scaled difference between V
COX
and
V
OUTX
, which slowly forces this difference to zero (the I part of the PI controller).
Resistor R
56
injects the integrator’s output at the DUT’s input through resistors R
4
and
R
3
; it minimizes the error at V
OUTX
.
A proportional term (the P part of the PI controller) is also injected at the DUT’s input
through resistor R
78
; it stabilizes the control loop (the integrator term becomes
negligible above 16 Hz). It also sets a low frequency DUT noise gain of about 505 V/V.
This proportional term is rolled off by C
2
starting at 0.18 kHz; this is high enough to not
interact with the integrator term, and low enough to keep the DUT stable. Thus, C
2
minimizes noise gain at higher frequencies, which reduces the chance of unwanted
feedback effects.
With the overall gain G
A
G
M
of either 201 V/V or 1998 V/V, this circuit can measure
V
OST
values up to either ±12.4 mV or ±1.25 mV. A voltmeter with 1 mV resolution can
distinguish steps of either 5 µV or 0.5 µV, respectively.
The DUT’s noise seen at the input to G
M
has a noise power bandwidth (NPBW) set by
R
78
and C
2
(0.28 kHz). This implies that this noise is dominated by the 1/f noise. The
Lowpass Filter (f
BW
≈
1.6 Hz) reduces this 1/f noise a little more before it is seen at V
M
.
The measured noise, over a 140 second period of time with a typical part, was about
19 µV
P-P
referred to input (RTI). This compares favorably with the MCP651’s calibrated
V
OS
specification (±200 µV, maximum at +25°C).
V
OUTX
V
COX
≈
V
M
G
A
G
M
V
OST
≈
Where:
G
A
= 1 + R4/R3
≈
51.00 V/V
G
A
G
M
≈
201.0 V/V, S2 (position 1) closed
≈
1998 V/V, S2 (position 1) open
