FEATURES
Low Offset Voltage: 50 mV max
Low Noise Voltage at 100 Hz, 1 mA: 1.0 nV/√Hz max
High Gain (h
Excellent Log Conformance: r
Low Offset Voltage Drift: 0.1 mV/8C max
Improved Direct Replacement for LM194/394
Available in Die Form
PRODUCT DESCRIPTION
The design of the MAT02 series of NPN dual monolithic transistors is optimized for very low noise, low drift, and low r
Precision Monolithics’ exclusive Silicon Nitride “TriplePassivation” process stabilizes the critical device parameters
over wide ranges of temperature and elapsed time. Also, the high
current gain (h
range of collector current. Exceptional characteristics of the
MAT02 include offset voltage of 50 µV max (A/E grades) and
150 µV max F grade. Device performance is specified over the
full military temperature range as well as at 25°C.
Input protection diodes are provided across the emitter-base
junctions to prevent degradation of the device characteristics
due to reverse-biased emitter current. The substrate is clamped
to the most negative emitter by the parasitic isolation junction
created by the protection diodes. This results in complete isolation between the transistors.
The MAT02 should be used in any application where low noise
is a priority. The MAT02 can be used as an input stage to make
an amplifier with noise voltage of less than 1.0 nV/√
Other applications, such as log/antilog circuits, may use the excellent logging conformity of the MAT02. Typical bulk resistance is only 0.3 Ω to 0.4 Ω. The MAT02 electrical characteristics approach those of an ideal transistor when operated over
a collector current range of 1 µA to 10 mA. For applications re-
quiring multiple devices see MAT04 Quad Matched Transistor
data sheet.
REV. C
Information furnished by Analog Devices is believed to be accurate and
reliable. However, no responsibility is assumed by Analog Devices for its
use, nor for any infringements of patents or other rights of third parties
which may result from its use. No license is granted by implication or
otherwise under any patent or patent rights of Analog Devices.
): 500 min at IC = 1 mA
FE
300 min at I
) of the MAT02 is maintained over a wide
FE
= 1 mA
C
. 0.3 V
BE
Hz at 100 Hz.
BE
.
Dual Monolithic Transistor
MAT02
PIN CONNECTION
TO-78
(H Suffix)
NOTE
Substrate is connected to case on TO-78 package. Substrate is normally connected to the most negative circuit
potential, but can be floated.
ABSOLUTE MAXIMUM RATINGS
Collector-Base Voltage (BV
CBO
Collector-Emitter Voltage (BV
Collector-Collector Voltage (BV
Emitter-Emitter Voltage (BV
Collector Current (I
Emitter Current (I
vs. Collector Current10 µA ≤ I
Bulk Resistancer
Collector Saturation VoltageV
CB
(@ 258C for VCB = 15 V and IC = 10 mA, unless otherwise noted.)
CEO
OS
OS
B
FE
FE
/∆V
OS
BE
CE (SAT)
10 µA ≤ IC ≤ 1 mA
VCB = 0 V34nA max
IC = 1 mA, VCB = 0 V400min
I
= 10 µA, VCB = 0 V300
C
10 µA ≤ IC ≤ 1 mA, VCB = 0 V4% max
CB
0 V ≤ VCB ≤ 40 V50µV max
10 µA ≤ IC ≤ 1 mA
C
VCB = 050µV max
≤ 1 mA
C
100 µA ≤ IC ≤ 10 mA0.5Ω max
IC = 1 mA0.2V max
1
1
1
MAT02N
40V min
150µV max
1.2nA max
IB = 100 µA
NOTES
1
Measured at lC = 10 µA and guaranteed by design over the specified range of IC.
Electrical tests are performed at wafer probe to the limits shown. Due to variations in assembly methods and normal yield loss, yield after packaging is not guaranteed
for standard product dice. Consult factory to negotiate specifications based on dice lot qualification through sample lot assembly and testing.
TYPICAL ELECTRICAL CHARACTERISTICS
(VCB = 15 V, IC = 10 mA, TA = +258C, unless otherwise noted.)
MAT02N
ParameterSymbolConditionsLimitsUnits
Average OffsetTCV
OS
Voltage Drift0 ≤ V
Average OffsetTCI
OS
10 µA ≤ IC ≤ 1 mA0.08µV/°C
≤ V
CB
MAX
IC = 10 µA40pA/°C
Current Drift
Gain-Bandwidthf
T
VCE = 10 V, IC = 10 mA200MHz
Product
Offset Current Change vs. V
CB
∆IOS/∆V
CB
0 ≤ VCB ≤ 40 V70pA/V
DICE CHARACTERISTICS
1. COLLECTOR (1)
2. BASE (1)
3. EMITTER (1)
4. COLLECTOR (2)
5. BASE (2)
6. EMITTER (2)
7. SUBSTRATE
Die Size 0.061 × 0.057 inch, 3,477 sq. mils
×
(1.549
1.448 mm, 224 sq. mm)
CAUTION
ESD (electrostatic discharge) sensitive device. Electrostatic charges as high as 4000 V readily
accumulate on the human body and test equipment and can discharge without detection.
Although the MAT02 features proprietary ESD protection circuitry, permanent damage may
occur on devices subjected to high energy electrostatic discharges. Therefore, proper ESD
precautions are recommended to avoid performance degradation or loss of functionality.
–4–
WARNING!
ESD SENSITIVE DEVICE
REV. C
MAT02
Figure 1. Current Gain vs.
Collector Current
Figure 4. Base-Emitter-On
Voltage vs. Collector Current
Figure 2. Current Gain
vs. Temperature
Figure 5. Small Signal Input
Resistance vs. Collector Current
Figure 3. Gain Bandwidth
vs. Collector Current
Figure 6. Small-Signal Output
Conductance vs. Collector Current
Figure 7. Saturation Voltage
vs. Collector Current
REV. C
Figure 8. Noise Voltage
Density vs. Frequency
–5–
Figure 9. Noise Voltage Density
vs. Collector Current
MAT02
Figure 10. Noise Current
Density vs. Frequency
Figure 13. Collector-to-Collector
Leakage vs. Temperature
Figure 11. Total Noise vs.
Collective Current
Figure 14. Collector-to-Collector
Capacitance vs. Collector-to
Substrate Voltage
Figure 12. Collector-to-Base
Leakage vs. Temperature
Figure 15. Collector-Base
Capacitance vs. Reverse Bias Voltage
Figure 16. Collector-to-Collector
Capacitance vs. Reverse Bias Voltage
Figure 17. Emitter-Base Capacitance
vs. Reverse Bias Voltage
–6–
REV. C
MAT02
Figure 18. Log Conformance Test Circuit
LOG CONFORMANCE TESTING
The log conformance of the MAT02 is tested using the circuit
shown above. The circuit employs a dual transdiode logarithmic converter operating at a fixed ratio of collector currents
that are swept over a 10:1 range. The output of each transdiode
converter is the V
is the product of the collector current and r
resistance. The difference of the V
of the transistor plus an error term which
BE
is amplified at a gain of
BE
, the bulk emitter
BE
×100 by the AMP01 instrumentation amplifier. The differential emitter-base voltage (∆V
) consists of a temperature-
BE
dependent dc level plus an ac error voltage which is the deviation from true log conformity as the collector currents vary.
The output of the transdiode logarithmic converter comes
from the idealized intrinsic transistor equation (for silicon):
kT
I
C
=
In
where(1)
I
q
S
-23
J/°K)
-19
°C)
V
BE
k = Boltzmann’s Constant (1.38062 × 10
q = Unit Electron Charge (1.60219 × 10
T = Absolute Temperature, °K (= °C + 273.2)
= Extrapolated Current for VBE→0
I
S
= Collector Current
I
C
An error term must be added to this equation to allow for the
bulk resistance (r
) of the transistor. Error due to the op amp
BE
input current is limited by use of the OP15 BiFET-input op
amp. The resulting AMP01 input is:
kT
I
C1
∆
VBE =
In
+ IC1 r
I
q
C2
BE1
– IC2 r
BE2
(2)
A ramp function which sweeps from 1 V to 10 V is converted by
the op amps to a collector current ramp through each transistor.
Because I
is made equal to 10 IC2, and assuming TA = 25°C,
C1
the previous equation becomes:
∆
VBE = 59 mV + 0.9 IC1 rBE (∆rBE ~ 0)
As viewed on an oscilloscope, the change in ∆V
change in I
is then displayed as shown below:
C
for a 10:1
BE
REV. C
–7–
MAT02
With the oscilloscope ac coupled, the temperature dependent
term becomes a dc offset and the trace represents the deviation
from true log conformity. The bulk resistance can be calculated
from the voltage deviation ∆V
and the change in collector cur-
O
rent (9 mA):
r
=
BE
This procedure finds r
provide the r
R
= R2.
1
for Side B. Differential rBE is found by making
BE
∆V
9mA
for Side A. Switching R1 and R2 will
BE
1
O
×
100
(3)
APPLICATIONS: NONLINEAR FUNCTIONS
MULTIPLIER/DIVIDER CIRCUIT
The excellent log conformity of the MAT02 over a very wide
range of collector current makes it ideal for use in log-antilog
circuits. Such nonlinear functions as multiplying, dividing,
squaring, and square-rooting are accurately and easily implemented with a log-antilog circuit using two MAT02 pairs (see
Figure 19). The transistor circuit accepts three input currents
(I
, I2, and I3) and provides an output current IO according to
1
I
= I1I2/I3. All four currents must be positive in the log-antilog
O
circuit, but negative input voltages can be easily accommodated
by various offsetting techniques. Protective diodes across each
base-to-emitter junction would normally be needed, but these
diodes are built into the MAT02. External protection diodes are
therefore not needed.
For the circuit shown in Figure 19, the operational amplifiers
make I
= VX/R1, I2 = VY/R2, I3 = VZ/R3, and IO = VO/RO. The
1
output voltage for this one-quadrant, log-antilog multiplier/divider is ideally:
R3R
VXV
O
=
V
O
R1R
If all the resistors (R
V
. Resistor values of 50 kΩ to 100 kΩ are recommended
XVY/VZ
O
Y
(VX, VY, VZ > 0)(4)
V
2
Z
, R1, R2, R3) are made equal, then VO =
assuming an input range of 0.1 V to +10 V.
ERROR ANALYSIS
The base-to-emitter voltage of the MAT02 in its forward active
operation is:
kT
I
C
In
=
V
BE
q
+ rBEIC, VCB ~ 0(5)
I
S
The first term comes from the idealized intrinsic transistor
equation previously discussed (see equation (1)).
Figure 19. One-Quadrant Multiplier/Divider
–8–
REV. C
Figure 20. Compensation of Bulk Resistance Error
Extrinsic resistive terms and the early effect cause departure
from the ideal logarithmic relationship. For small V
CB
, all of
these effects can be lumped together as a total effective bulk resistance r
logarithmic relationship. The r
than 0.5 Ω and ∆r
. The rBEIC term causes departure from the desired
BE
between the two sides is negligible.
BE
term for the MAT02 is less
BE
Returning to the multiplier/divider circuit of Figure 1 and using
Equation (4):
V
BE1A
+ V
BE2A
– V
BE2B
–V
+ (I1 + I2 – IO – I3) rBE = 0
BE1B
If the transistor pairs are held to the same temperature, then:
kT
I
kT
1I2
In
=
I3I
q
O
I
S1AIS2A
In
I
q
S1BIS2B
+ (I1 + I2 – IO – I3) r
BE
(6)
If all the terms on the right-hand side were zero, then we would
have In (I
1 I2/I3 IO
) equal to zero which would lead directly to
the desired result:
I1I
I
2
=
O
, where I1, I2, I3, IO > 0(7)
I
3
Note that this relationship is temperature independent. The
right-hand side of Equation (6) is near zero and the output current I
will be approximately I1 I2/I3. To estimate error, define ø
O
as the right-hand side terms of Equation (6):
MAT02
approximately 26 mV and the error due to an rBEIC term will be
r
/26 mV. Using an rBE of 0.4 Ω for the MAT02 and assum-
BEIC
ing a collector current range of up to 200 µA, then a peak error
of 0.3% could be expected for an r
the MAT02. Total error is dependent on the specific application
configuration (multiply, divide, square, etc.) and the required
dynamic range. An obvious way to reduce I
duce the maximum collector current, but then op amp offsets
and leakage currents become a limiting factor at low input levels. A design range of no greater than 10 µA to 1 mA is generally
recommended for most nonlinear function circuits.
A powerful technique for reducing error due to I
Figure 20. A small voltage equal to I
sistor base. For this circuit:
R
C
V
=
V1 and ICrBE =
B
R
2
The error from r
R
. Since the MAT02 bulk resistance is approximately 0.39 Ω,
1
an R
of 3.9 Ω and R2 of 10 R1 will give good error cancellation.
C
is cancelled if RC/R2 is made equal to rBE/
BEIC
In more complex circuits, such as the circuit in Figure 19, it
may be inconvenient to apply a compensation voltage to each
individual base. A better approach is to sum all compensation to
the bases of Q1. The “A” side needs a base voltage of (V
V
) rBE and the “B” side needs a base voltage of (VX/R1+VY/
Z/R3
R
) rBE. Linearity of better than ±0.1% is readily achievable with
2
this compensation technique.
Operational amplifier offsets are another source of error. In Fig-
ure 20, the input offset voltage and input bias current will cause
an error in collector current of (V
amp, such as the OP07 with less than 75 µV of V
less than ±3 nA, is recommended. The OP22/OP32, a programmable micropower op amp, should be considered if low power
consumption or single-supply operation is needed. The value of
frequency-compensating capacitor (C
amp frequency response and peak collector current. Typical values for C
. . .
range from 30 pF to 300 pF.
O
error term when using
BEIC
error is to re-
CrBE
is applied to the tran-
CrBE
r
BE
V
R
1
) + IB. A low offset op
OS/R1
) is dependent on the op
O
CrBE
1
and IB of
OS
is shown in
(10)
+
O/RO
I
ø = In
For the MAT02, In (I
Ø
ø, ε
~ 1 + ø and therefore:
S1AIS2A
I
S1BIS2B
q
+
(I1 + I2 – IO – I3) r
kT
) and ICrBE are very small. For small
SA/ISB
I1I
2
= 1+ø
I3I
O
BE
(8)
(9)
I
1I2
(1 – ø)
I
3
The In (I
I
~
O
) terms in ø cause a fixed gain error of less than
SA/ISB
±0.6% from each pair when using the MAT02, and this gain
error is easily trimmed out by varying R
. The ICrBE terms are
O
more troublesome because they vary with signal levels and
are multiplied by absolute temperature. At 25°C, kT/q is
REV. C
FOUR-QUADRANT MULTIPLIER
A simplified schematic for a four-quadrant log/antilog multiplier
is shown in Figure 21. As with the previously discussed onequadrant multiplier, the circuit makes I
input currents, I
and I2, are each offset in the positive direction.
1
= I1 I2/I3. The two
O
This positive offset is then subtracted out at the output stage.
Assuming ideal op amps, the currents are:
V
I1=
V
X
+
R
R
1
R
,I2=
2
V
V
Y
R
+
R
R
1
2
(11)
V
V
V
IO=
X
Y
+
R
R
1
1
V
R
+
+
R
R
2
O
O
,I3=
V
R
R
2
From IO = I1 I2/I3, the output voltage will be:
ROR
VXV
2
=
V
O
R
1
Y
2
V
R
(12)
–9–
MAT02
Collector-current range is the key design decision. The inherently low r
collector current. For input scaling of ± 10 V full-scale and using a 10 V reference, we have a collector-current range for I
and I2 of:
Practical values for R
100 kΩ. Choosing an R
collector-current range of approximately 39 µA to 283 µA. An
R
of 108 kΩ will then make the output scale factor 1/10 and
O
V
= VXVY/10. The output, as well as both inputs, are scaled
O
for ±10 V full scale.
Linear error for this circuit is substantially improved by the
small correction voltage applied to the base of Q1 as shown in
Figure 21. Assuming an equal bulk emitter resistance for each
MAT02 transistor, then the error is nulled if:
The currents are known from the previous discussion, and the
relationship needed is simply:
The output voltage is attenuated by a factor of r
plied to the base of Q1 to cancel the summation of voltage
drops due to r
nearly zero which will thereby make I
rate relationship. Linearity of better than 0.1% is readily achievable with this circuit if the MAT02 pairs are carefully kept at
the same temperature.
of the MAT02 allows the use of a relatively high
BE
–10
R
(I
+ I2 – I3 – IO) rBE + ρVO = 0
1
BEIC
10
+
R
1
2
and R2 would range from 50 kΩ to
1
1
V
=
O
terms. This will make In (I1 I2/I3 IO) more
≤I
≤
C
10
R
10
+
R
1
2
of 82 kΩ and R2 of 62 kΩ provides a
r
BE
V
O
R
O
and ap-
BE/RO
= I1 I2/I3 a more accu-
O
1
(13)
(14)
MULTIFUNCTION CONVERTER
The multifunction converter circuit provides an accurate means
of squaring, square rooting, and of raising ratios to arbitrary
powers. The excellent log conformity of the MAT02 allows a
wide range of exponents. The general transfer function is:
m
V
V
= VY
O
V
, VY, and VZ are input voltages and the exponent “m” has a
X
practical range of approximately 0.2 to 5. Inputs V
Z
V
X
X
(15)
and VY are
often taken from a fixed reference voltage. With a REF01 providing a precision +10 V to both V
and VY, the transfer func-
X
tion would simplify to:
m
V
V
= 10
O
Z
10
(16)
As with the multiplier/divider circuits, assume that the transistor
pairs have excellent matching and are at the same temperature.
The In I
will then be zero. In the circuit of Figure 22, the
SA/ISB
voltage drops across the base-emitter junctions of Q1 provide:
R
B
RB+ KR
is VZ/R1 and IX is VX/R1. Similarly, the relationship for Q2 is:
I
Z
R
B
RB+ 1– K
()
is VO/RO and IY is VY/R1. These equations for Q1 and Q2 can
I
O
A
VA=
kT
VA=
R
A
I
Z
In
I
q
X
kT
I
O
In
I
q
Y
(17)
(18)
then be combined.
Figure 21. Four-Quadrant Multiplier
RB+ KR
RB+ 1– K
()
A
R
A
In
I
I
Z
O
= In
I
I
X
Y
(19)
–10–
REV. C
MAT02
Substituting in the voltage relationships and simplifying leads
to:
m
R
O
V
=
V
O
R
1
V
Z
Y
, where
V
X
(20)
m =
RB+ KR
RB+ 1– K
A
R
()
A
The factor “K” is a potentiometer position and varies from zero
to 1.0, so “m” ranges from R
Practical values are 125 Ω for R
/(RA + RB) to (RB + RA)/RB.
B
and 500 Ω for RA; these val-
B
ues will provide an adjustment range of 0.2 to 5.0. A value of
100 kΩ is recommended for the R
resistors assuming a full-
1
scale input range of 10 V. As with the one-quadrant multiplier/
divider circuit previously discussed, the V
, VY, and VZ inputs
X
must all be positive.
The op amps should have the lowest possible input offsets. The
OP07 is recommended for most applications, although such
programmable micropower op amps as the OP22 or OP32 offer
advantages in low-power or single-supply circuits. The micropower op amps also have very low input bias-current drift, an
important advantage in log/antilog circuits. External offset nulling may be needed, particularly for applications requiring a
wide dynamic range. Frequency compensating capacitors, on
the order of 50 pF, may be required for A
A
is likely to need a larger capacitor, typically 0.0047 µF, to as-
1
and A3. Amplifier
2
sure stability.
Accuracy is limited at the higher input levels by bulk emitter resistance, but this is much lower for the MAT02 than for other
transistor pairs. Accuracy at the lower signal levels primarily depends on the op amp offsets. Accuracies of better than 1% are
readily achievable with this circuit configuration and can be better than ±0.1% over a limited operating range.
FAST LOGARITHMIC AMPLIFIER
The circuit of Figure 23 is a modification of a standard logarithmic amplifier configuration. Running the MAT02 at 2.5 mA per
side (full-scale) allows a fast response with wide dynamic range.
The circuit has a 7 decade current range, a 5 decade voltage
range, and is capable of 2.5 µs settling time to 1% with a 1 V to
10 V step.
The output follows the equation:
R3+ R
V
=
O
kT
V
2
R
2
REF
In
V
q
IN
(21)
The output is inverted with respect to the input, and is nominally –1 V/decade using the component values indicated.
LOW-NOISE 31000 AMPLIFIER
The MAT02 noise voltage is exceptionally low, only 1 nV/√Hz
at 10 Hz when operated over a collector-current range of 1 mA
to 4 mA. A single-ended ×1000 amplifier that takes advantage of
this low MAT02 noise level is shown in Figure 24. In addition
to low noise, the amplifier has very low drift and high CMRR.
An OP32 programmable low-power op amp is used for the second stage to obtain good speed with minimal power consumption. Small-signal bandwidth is 1 MHz, slew rate is 2.4 V/µs,
and total supply current is approximately 2.8 mA.
REV. C
Figure 22. Multifunction Converter
–11–
MAT02
Transistors Q2 and Q3 form a 2 mA current source (0.65 V/
330 Ω ~ 2 mA).
OP32 inputs are 3 V below the positive supply voltage (R
Each collector of Q1 operates at 1 mA. The
LIC
~ 3 V). The OP32’s low input offset current, typically less than
1 nA, and low offset voltage of 1 mV cause negligible error
when referred to the amplifier input. Input stage gain is g
which is approximately 100 when operating at I
R
of 3 kΩ. Since the OP32 has a minimum open-loop gain of
L
of 1 mA with
C
mRL
,
500,000, total open-loop gain for the composite amplifier is
over 50 million. Even at closed-loop gain of 1000, the gain error due to finite open-loop gain will be negligible. The OP32
features excellent symmetry of slew-rate and very linear gain.
Signal distortion is minimal.
Frequency compensation is very easy with this circuit; just vary
the set-resistor R
for the desired frequency response.
S
Gain-bandwidth of the OP32 varies directly with the supply
current. A set resistor of 549 kΩ was found to provide the best
step response for this circuit. The resultant supply current is
found from:
V +
–V–
–2V
()
()
BE
I
SET
,ISY=15I
of 549 kΩ, is ap-
SET
SET
(22)
The I
SET
=
()
R
, using ±15 V supplies and an R
SET
proximately 52 µA which will result in supply current of 784 µA.
Dynamic range of this amplifier is excellent; the OP32 has an
output voltage swing of ±14 V with a ±15 V supply.
Input characteristics are outstanding. The MAT02F has offset
voltage of less than 150 µV at 25°C and a maximum offset drift
of 1 µV/°C. Nulling the offset will further reduce offset drift.
This can be accomplished by slightly unbalancing the collector
load resistors. This adjustment will reduce the drift to less than
0.1 µV/°C.
Input bias current is relatively low due to the high current gain
of the MAT02. The minimum β of 400 at 1 mA for the
MAT02F implies an input bias current of approximately 2.5 µA.
This circuit should be used with signals having relatively low
source impedance. A high source impedance will degrade offset
and noise performance.
This circuit configuration provides exceptionally low input noise
voltage and low drift. Noise can be reduced even further by raising the collector currents from 1 mA to 3 mA, but power consumption is then increased.