Application Note 11
September 1985
Designing Linear Circuits for 5V Single Supply Operation
Jim Williams
In predominantly digital systems it is often necessary
to include linear circuit functions. Traditionally, separate
power supplies have been used to run the linear components (see Box, “Linear Power Supplies—Past, Present,
and Future”).
Recently, there has been increasing interest in powering
linear circuits directly from the 5V logic rail. The logic
rail is a difficult place for analog components to function.
The high amplitude broadband current and voltage noise
generated by logic clocking makes analog circuit operation difficult. (See Box, “Using Logic Supplies for Linear
Functions”.)
Generally speaking, analog circuitry which must achieve
very high performance levels should be driven from dedicated supplies. The difficulties encountered in maintaining
the lowest possible levels of noise and drift in an analog
system are challenging enough without contending with
a digitally corrupted power supply.
Many analog applications, however, can be successfully
implemented using the logic supply. Combining components intended to provide high performance from the
167Ω
Q1
–
+
2M
L, LT, LTC, LTM, Linear Technology and the Linear logo are registered trademarks of Linear
Technology Corporation. All other trademarks are the property of their respective owners.
RATIOMETRIC
OUTPUT FOR
A/D CONVERTER
150Ω
A2
1/4 LT1014
+
200k
1.5k
Figure 1 shows a circuit which provides complete, linearized signal conditioning for a platinum RTD. One side of
the RTD sensor is grounded, often desirable for noise
considerations. The Q1-Q2 current source is referenced
to A1’s output. A1’s operating point is primarily fixed by
the 2.5V LT®1009 voltage reference. The RTD’s constant
current forces the voltage across it to vary with its resistance, which has a nearly linear positive temperature
coefficient. The nonlinearity causes several degrees of
error over the circuit’s 0°C to 400°C operating range. A2
amplifies RP’s output, while simultaneously supplying
nonlinearity correction. The correction is implemented by
feeding a portion of A2’s output back to A1’s input via the
10k-250k divider. This causes the Q1-Q2 current source
–
A3
1/4 LT1014
Q2
Linearized RTD Signal Conditioner
2M
200k
499Ω
logic rail with good design can give excellent results (see
Box, “High Performance, Single Supply Analog Building
Blocks”). The examples which follow show this in a variety of precision measurement and control circuits which
function from a 5V supply.
5k
LINEARITY
TRIM
3.01k
A4
1/4 LT1014
+
OUTPUT
0 – 4V =
0 – 400°C
±0.05°C
2.4k*
RP
5V
274k
LT1009
2.5V
50k
ZERO
TRIM
–
A1
1/4 LT1014
–
1k
GAIN
TRIM
8.25k
10k
250k
Q1, Q2 = 2N4250
RP = ROSEMOUNT-118MF-100Ω
= TRW-MAR-6
*
= 5% CARBON
RATIO MATCH 2M-200k ±0.01%
AN11 F01
Figure 1. Linearized Platinum RTD Signal Conditioner
an11f
AN11-1
+
Application Note 11
output to shift with RP’s operating point, compensating
sensor nonlinearity to within ±0.05°C. A3, also referenced
to the LT1004, voltage sums an offsetting signal at A2’s
negative input, allowing 0°C to equal 0V at A2’s output.
The resistive divider in A4’s input line sets circuit gain,
and the circuit’s output is taken at A4.
To calibrate this circuit, substitute a precision decade
box (e.g., General Radio 1432K) for RP . Set the box to
0°C value (100.00Ω) and adjust the zero trim for a 0.0V
output. Next, set the decade box for a 140°C output
(154.26Ω) and adjust the gain trim for a 1.400V output
reading. Finally, set the box to 400°C (249.0Ω) and trim
the linearity adjustment for a 4.000V output. Repeat this
sequence until all three points are fixed. Total error over
the entire range will be within ±0.05°C. The resistance
values given are for a nominal 100.00Ω (0°C) sensor.
Sensors deviating from this nominal value can be used
by factoring in the deviation from 100.00Ω. This deviation, which is manufacturer-specified for each individual
sensor, is an offset term due to winding tolerances during
fabrication of the RTD. The gain slope of the platinum is
primarily fixed by the purity of the material and is a very
small error term.
Linearized Output Methane Detector
Figure 2 is another 5V powered transducer circuit. Like
the platinum RTD example, this circuit linearizes the
transducer’s output, but it performs a more complex
mathematical operation. The circuit’s frequency output
is directly proportional to the methane concentration
detected by the transducer specified. Figure 3 shows that
the transducer output varies as:
≈
1
Concentration
The circuit linearizes this function and its frequency output
is also plotted.
The sensor’s resistance vs methane concentration is converted to a voltage by A1. The LT1034 serves as a reference. A1’s output feeds A2. The exponential relationship
between VBE and collector current in transistors is utilized
to generate a current in Q3’s collector proportional to the
square of A2’s input current. This operation compensates
the sensor’s square root term. Q3’s collector current sets
the operating point of the A3-A4 oscillator. A3, an integrator, generates a positive going linear ramp (Trace A,
Figure 4). The ramp is compared with Q3’s current at A4’s
summing point (Trace B). A4 is configured as a current
summing comparator. The feedback diode-bound network
minimizes delay due to output slew time. When the ramp
forces the summing point positive, A4’s output (Trace C)
swings negative. CMOS inverter A (Trace D) goes high,
turning on the CD4016 switch to reset the A3 integrator.
Simultaneously, inverter B goes low (Trace E), supplying
AC positive feedback to A4’s “+” input (Trace F). When the
positive feedback decays, A4’s output goes high, A3 begins
to integrate, and the entire cycle repeats. Q3’s collector
current determines how long A3’s ramp runs before A4
resets it. The ramp time is directly proportional to Q3’s
collector current, meaning that oscillation frequency is
inversely (1/X) related to the current. The overall circuit
transfer function is:
1
X2
This linearizes the sensor’s output. In practice, the sensor’s
response slightly deviates from the equation shown, actually being:
1
1.9
Concentration
The reset time constants at A4’s input introduce enough
oscillator “dead time” to partially compensate for the deviation. The dead time’s frequency retarding characteristic
effects the oscillator’s high frequency range, providing a
first order correction. The overall linearization achieved
is within the sensor’s manufacturing tolerances. The
slight “bump” in the circuit’s response curve is due to
the mismatch between the sensors term and the circuits
X2 function.
an11f
AN11-2
Application Note 11
CD4016
0.033
2.7k
390k*
–
LT1034
1.2V
A3
1/4 LT1014
–
100k*
+
5V
A
A4
1/4 LT1014
+
470pF
–5VOUT
5
8
LTC1044
10μF
10
2
10k
C
OUTPUT
Q1
+
5V
Q2
SENSOR
1000pF
5V
SENSOR = CALECTRO-GC ELECTRONICS #J4-807
OR FIGARO #813
= 1N4148
150k*
–
A1
1/4 LT1014
100k*
+
2k
–5V
Q4
Q3
HEATER
5k
1000ppm
TRIM
470pF
3
+
4
B
5V
–
2k
12k*
A2
1/4 LT1014
+
Q1-Q4 = CA3046 ARRAY
A1-A4 = LT1014 QUAD, RUN
*1% METAL FILM RESISTOR
INVERTERS = 74C04
PIN TO LTC1044—V OUTPUT
–5V
AN11 F02
Figure 2. Linearized Methane Transducer Signal Conditioner
12
DASHED LINE = IDEAL LINEARITY
1000
CIRCUIT RESPONSE
8
800
6
600
4
400
SENSOR RESPONSE
2
0
0
2000
4000
6000
8000
METHANE CONCENTRATION (ppm)
200
0
10000
AN11 F03
Figure 3. Transducer vs Circuit Response
CIRCUIT OUTPUT (Hz)
SENSOR RESISTANCE (kΩ)
10
A = 200mV/DIV
B = 5V/DIV
C = 2V/DIV
D = 20V/DIV
E = 20V/DIV
F = 10V/DIV
HORIZONTAL = 200μs/DIV
AN11 F04
Figure 4. Linearized Methane Detector Waveforms
an11f
AN11-3
Application Note 11
The dead time correction in the oscillator smooths this
error out above 4000ppm. The LTC®1044 voltage converter
generates a negative supply directly from the 5V rail. This
approach provides necessary negative circuit potentials
while maintaining compatibility with 5V supply only operation. The sensor’s heater is powered directly from the
5V rail, as specified by the manufacturer. To calibrate the
circuit, place the sensor in a 1000ppm methane environment and adjust the 5k trim for a 100Hz output. Accuracy
from 500ppm to 10,000ppm is limited by the sensor’s
10% specification.
Cold Junction Compensated Thermocouple Signal
Conditioner
Figure 5 shows a 5V powered, complete thermocouple
signal conditioner. Cold junction compensation is included,
and the circuit allows one leg of the thermocouple to be
grounded, desirable for noise considerations. The LTC1043
combines the cold junction network differential output with
the grounded thermocouple’s signal at the LTC1052. The
LTC1052 provides stable, low drift gain. To enable swing
all the way to ground, the LTC1043’s other switch section generates a small negative potential. This allows the
LTC1052 output stage to run Class A for small outputs,
permitting swing to 0V. The table gives proper values for
R1 for various thermocouples. Output scaling may be
set by RF/RI to whatever slope is desirable. Cold junction
compensation holds within ±1°C over 0°C to 60°C.
5V Powered Precision Instrumentation Amplifier
Many transducer outputs require a true differential input
“instrumentation” type amplifier. Transducers with singleended outputs do not, in theory, require a differential input
amplifier but common mode noise often exceeds the signal
of interest. For these reasons, transducer systems often
employ these type amplifiers.
No commercially manufactured instrumentation amplifier
will function from a 5V supply and achieving good precision in a design is difficult. The circuits in Figure 6 meet
these requirements. They also feature input protection,
filtering capability and a shield driving output.
In Figure 6a, A1, A2 and A3 accomplish the differential
input-to-single-ended output conversion, with RG setting
gain. The LT1014’s high open-loop gain permits accurate
circuit gain. Offset and drift performance allows use with
low level transducers such as thermocouples and strain
gauges. The 100kΩ-1μF combination filters noise and 60Hz
pickup; the amplifier is never exposed to high frequency
common mode noise. The transistor-diode clamps combine
with the 100k resistors to prevent high voltage spikes or
faults (common in industrial environments) from damaging
the amplifiers. A4 is used as a shield driver to reduce the
effects of input cable capacitance. It drives the shield at
the input common mode voltage, which is derived from
the input amplifier’s output. Performance characteristics
are summarized in the table.
Figure 6B achieves greater DC precision at the expense of
bandwidth. Here, the LTC1043 switched-capacitor building
block alternately commutates a 1μF capacitor between the
circuit input and the LTC1052’s input. This stage, accomplishing a differential-to-single-ended transition, allows
the chopper-stabilized LTC1052 to take a ground referred
measurement. The LTC1043’s other switch stage generates
a small negative potential, allowing the LTC1052 output to
swing all the way to 0V. DC precision is excellent, surpassing all monolithic ±15V instrumentation amps, although
bandwidth is limited to 10Hz. The LTC1043’s switching
action, set at about 400Hz by the 0.01μF value, forms a
lowpass filter. This permits extremely high rejection of
noise, allowing the CMRR to remain above 120dB at 20kHz.
Overall performance is summarized in the table.
an11f
AN11-4
Application Note 11
1k
5V
+
COLD JUNCTION
COMPENSATION**
100k
THERMOCOUPLE
INPUT
1N914
(2)
VT
–
R1
5V
LT1004-1.2V
1690Ω
5k†
AT 25°C
4
5V
LTC1043
7
187Ω
3
8
1820Ω
7
+
11
1μF
6
LTC1052
2
R
VOUT = VT 1+ F
RI
8
–
4
1
1μF
0.1μF
0.1μF
RF
CF*
12
13
14
6
5
RI
43k
5V
THERMOCOUPLE
TYPE
J
K
T
S
2
+
1N914
+
1μF
NEGATIVE
BIAS
GENERATOR
1μF
3
18
15
0.0047μF
16
17
≈–0.5V
10k
R1
232k
301k
301k
2.1M
†YELLOW SPRINGS INST. CO.
AN11 F05
PART #44007
*CHOOSE CF TO FILTER NOISE
**LT1025 CAN BE USED FOR COLD
JUNCTION COMPENSATION
Figure 5. Cold Junction Compensated Thermocouple Signal Conditioner
an11f
AN11-5
Application Note 11
–
TO INPUT
CABLE SHIELDS
OFFSET VOLTAGE
OFFSET VOLTAGE DRIFT
GAIN DRIFT
CMRR
PSRR
GAIN ACCURACY
COMMON MODE RANGE
A4
+
200k*
5V
–
10k*
“–”
INPUT
100k
10k*
A2
+
1μF
10k
RG
2k TYP
5V
200k*
10k
–
–
1μF
“+”
INPUT
10k*
100k
500μV
4μV/°C
RESISTOR LIMITED
100dB
100dB
RESISTOR LIMITED
0V TO 3.5V
+
A1
+
A3
10k*
OUTPUT
AN11 F06a
*1% FILM RESISTOR MATCH 10k’s 0.05%
A1-A4 = LT1014 QUAD OP AMP
5V
GAIN EQUATION
200,000 +1
RG
= 2N2222A
A=
= 1N4148
Figure 6a. Precision Instrumentation Amplifier
5V
LTC1043
7
+
3
8
7
+
6
LTC1052
2
11
INPUT
1μF
8
–
4
1μF
OUTPUT
AV = 1000
1
0.01μF
(2)
12
R2
100Ω
–
V+
5V
13
A = R1 +1
R2
R1
99.9k
14
0.22μF
43k
6
5
OFFSET VOLTAGE
OFFSET VOLTAGE DRIFT
2
10k
+
1N914
1μF
1μF
CMRR
PSRR
3
18
16
GAIN ACCURACY
15
17
GAIN DRIFT
BANDWIDTH
COMMON MODE RANGE
≈–0.5V
5μV
0.05μV/°C
SET BY R1 DRIFT
R2
120dB DC–20kHz
>120dB
SET BY R1 ACCURACY
R2
10Hz
0V TO 5V
AN11 F06b
4
5V
0.01μF
Figure 6b. Ultra-Precision Instrumentation Amplifier
an11f
AN11-6
Application Note 11
5V Powered Strain Gauge Signal Conditioner
Figure 7 shows an unusual approach to signal conditioning
the bridge output of a strain gauge pressure transducer.
The 5V circuit needs only two amplifiers and provides an
auxiliary ratio output for a monitoring A/D converter. The
design functions by converting the bridges differential
output into a ground-referred single-ended signal which
is then amplified. This approach eliminates common mode
errors by eliminating the bridge’s common mode output
component. Additionally, the number of precision resistors
required is minimized and no matching is required.
A1 biases the LTC1044 positive-to-negative converter. The
LTC1044’s output pulls the bridge’s output negative, causing A1’s input to balance at 0V. This local loop permits a
single-ended amplifier (A2) to extract the bridge’s output
signal. The 100k-0.33μF RC filter’s noise and A2’s gain is set
to provide the desired output scale factor. Because bridge
drive is derived from the LT1034 reference, A2’s output
is not affected by supply shifts. The LT1034’s output is
available for ratio operation. To calibrate this circuit, apply
or electrically simulate 0psi and trim the zero adjustment
for 0V output. Next, apply or electrically simulate 350psi
and trim gain for 3.500V out. Repeat these adjustments
until both points are fixed.
“Tachless” Motor Speed Controller
Figure 8 shows a way to servo control the speed of a DC
motor. This circuit is particularly applicable to digitally
controlled systems in robotic and X-Y positioning applications. By functioning from the 5V logic supply it eliminates
additional motor drive supplies. The “tachless” feedback
saves additional space and cost. The circuit senses the
motor’s back EMF to determine its speed. The difference
between the speed and a setpoint is used to close a
sampled loop around the motor. Because no commercially
available sample-hold circuit will run from a 5V supply,
special techniques are required.
5V
220Ω
PRESSURE
TRANSDUCER
350Ω
LT1034-1.2V
10k
ZERO TRIM
D
39k
301k
E
A
100k
–
+
C
A1
1/2 LT1013
0.1μF
5V
+
0.33μF
A2
1/2 LT1013
–
2
OUTPUT
0V TO 3.5V
0V TO 350PSI
0.047μF
8
2k
GAIN TRIM
4 LTC1044 6
3
46k*
5
+
+
100μF
1.2V REFERENCE OUTPUT
TO A/D CONVERTER
FOR RATIOMETRIC OPERATION
0.1mA MAXIMUM
100μF
*1% FILM RESISTOR
PRESSURE TRANSDUCER-BLH/DHF-350
CIRCLED LETTER IS PIN NUMBER
ZIN = 350Ω
100Ω*
AN11 F07
Figure 7. Strain Gauge Signal Conditioner
an11f
AN11-7
Application Note 11
330k
100k
0.47μF
5V
+
–
1k
82Ω
A1
1/2 LT1013
47μF
Q3
+
1M
2k
Q1
2k
1N4001
6.8M
1/4 CD4016
1N4148
Q2
0.068μF
0.068μF
–
A2
1/2 LT1013
2k
S1
3.3M
0.047μF
•
1N4001
+
MOTOR = CANON-FN30—R13NIB
NPN = 2N3904
PNP = 2N5023
EIN
0V TO 3V
AN11 F08
Figure 8. “Tachless” Motor Speed Controller
A1 generates a pulse train (Trace A, Figure 9). When A1’s
output is high, Q1 is biased and Q3 drives the motor’s
ungrounded terminal (Trace B). When A1 goes low, Q3
turns off and the motor’s back EMF appears after the
inductive flyback ceases. During this period, S1’s input
(Trace C) is turned on, and the 0.047 capacitor acquires
the back EMF’s value. A2 compares this value with the
setpoint and the amplified difference (Trace D) changes A1’s
duty cycle, controlling motor speed. A2 has the desirable
characteristic of assuming unity gain in the absence of a
feedback signal. Start-up or input overdrive cannot force
servo lock-up due to loss of sampling action. The loop is
self-restoring and will establish control when abnormal
conditions cease.
Drive to S1’s control input must be carefully controlled
for proper operation. Q2 prevents the switch from closing until the negative going flyback interval is over and
the 0.068μF capacitor slows the switches turn-on edge.
These measures ensure clean back EMF acquisition in
the 0.047μF unit. Q2’s collector line diodes compensate
the motor’s clamp diode drop, preventing destructive
negative voltages at S1. The circuit controls from 20rpm
to full speed with good transient response under all shaft
loads. The gain and roll-off terms in A2’s feedback loop are
optimal for the motor listed, and should be reestablished
for other types.
A = 5V/DIV
B = 5V/DIV
C = 5V/DIV
D = 0.5V/DIV
AC-COUPLED
1ms/DIV
AN11 F09
Figure 9. Motor Speed Controller Waveforms
an11f
AN11-8
Application Note 11
4-20mA Current Loop Transmitter
voltage step-up and the rectified and filtered secondary
current flows through the 100Ω resistor and the load.
A1’s negative input measures the voltage across the 100Ω
resistor, completing a current control loop around T1. The
0.33μF capacitor furnishes stable loop roll-off and the
100pF unit suppresses local oscillation at Q4. Within the
compliance limit, A1 maintains constant output current,
regardless of load impedance shifts or supply changes.
To calibrate this circuit, short the output, apply 0V to the
input and adjust the “4mA trim” for 0.3996V across the
100Ω resistor. Next, shift the input to 4.000V and trim the
20mA adjustment for 1.998V across the 100Ω resistor.
Repeat this procedure until both points are fixed. The gain
trim network shunting the 100Ω resistor necessitates the
odd voltage trim target values, but output current swings
between 4.000mA and 20.00mA.
Transmission of industry standard 4-20mA current loop
signals to values and other actuators is a common requirement. Resistive line losses and actuator impedances
require current transmitters to be able to force a compliance voltage of a least 20V. Because of this, 5V powered
systems usually cannot meet current loop transmitter
requirements, but Figure 10 shows a way to do this. This
5V powered circuit utilizes a servo controlled DC/DC
converter to generate the compliance voltage necessary
for loop current requirements. It will drive 4-20mA into
loads as high as 2200Ω (44V compliance) and is inherently short-circuit protected. The circuit’s input is applied
to A2, whose output biases A1’s “+” input via the offsetting
network. A1’s output goes high, biasing Q4 to turn on
Q3. Q3’s collector drives the T1-Q1-Q2 DC/DC converter,
which is clocked by the RC gate oscillator. T1 furnishes
5V
Q3
820Ω
0.33μF
Q1
T1
10μF
10μF
150V
+
0.002μF
68Ω
*1% FILM
**0.01% RESISTOR
PNP = 2N2905
NPN = 2N2222
T1 = PICO-31080
+
1N4002
Q2
10k
+
4.3k
10k*
5V
LT1004
1.2V
1k
4mA
4k* TRIM
10k*
A2
1/2 LT1013
10k
20mA
TRIM
100Ω**
80k*
+
A1
1/2 LT1013
–
100pF
= 74C04
820Ω
100k
–
2k
Q4
10k
4mA TO 20mA OUT
AN11 F10
TO LOAD
2.2kΩ MAXIMUM
0 TO 4VIN
Figure 10. 4-20mA Current Loop Transmitter
an11f
AN11-9
Application Note 11
Figure 11 details modifications which permit the circuit’s
output current to galvanically float. This is often useful
in industrial situations where the output lines may be
exposed to common mode voltages or high voltage fault
conditions. The transformer’s primary current, which
theoretically reflects current delivered by the secondary,
is sensed across a shunt and fed back via A2. In practice,
–
Figure 12’s 5V powered circuit performs a fully isolated
limit comparison on low level signals. It produces a digital
10μF
150V
100k
+
4.3k
Fully Isolated Limit Comparator
T1
A1
1/2 LT1013
TO INVERTER
DRIVE
current control precision is limited by non-ideal transformer
behavior to about 1%. Common mode voltage is limited
by T1’s 300V breakdown.
+
0.1Ω
10k*
5V
LT1004
1.2V
+
4k*
A2
1/2 LT1013
1k
4mA TRIM
68k*
–
4mA TO 20mA OUT
FULLY FLOATING
0V TO 4V
INPUT
AN11 F11
301Ω*
*1% FILM RESISTOR
1k
20mA TRIM
Figure 11. Floating Output Option for Current Loop Transmitter
A = 100
ISOLATED
VOLTAGE
INPUT
+ISOL
+
–
C1
1018B
–
33Ω
+
20k
3k
C2
1018A
Q3
2N3906
1μF
4.7k
5V
0.1μF
1M
10k
10k
OUTPUT
PULSE
200pF
Q2
2N3904
+ISOL
1M
LIMIT SET
100k
100
5V
+
100μF
+ISOL
LT1034
1.2V
Q1
2N3906
500μS
INPUT
INTERROGATION
PULSE
1k
TRW #PC-SS0-32
AN11 F12
= FLOATING COMMON
= GROUND
GATE = 74C02
Figure 12. Fully Isolated Limit Comparator with Gain of 100
an11f
AN11-10
Application Note 11
output indicating if the input is above or below a preset
limit. This circuit is ideal for process control applications
where transducers operate at high common mode voltages
or where large ground loops exist. An uncommitted gain
of 100 amplifier allow thermocouples and other low level
sources to be used with the circuit. The circuit functions
by echoing an interrogation pulse if the input is above a
preset level. If the input is below this level, no echo pulse
occurs. A transformer is used to allow a 2-way, galvanically isolated signal path and the energy contained in the
interrogation pulse powers the circuit’s floating elements.
Figure 13 shows operating waveforms for the “above limit”
case. When an input interrogation pulse is applied, Q1’s collector drives the transformer primary (Figure 13, Trace A).
Energy is transferred to the transformer’s secondary and
stored in the 100μF capacitor. The charge in the capacitor
powers the isolated circuitry (“+ISOL” potential indicated
in Figure 12). If low power dual comparator C2’s output
is low, Q3 biases and drives current into the transformer
secondary (Trace B). This is reflected in the transformer’s
primary (Trace C). Q2 and the associated gate circuitry
form a demodulator which produces an output pulse
(Trace D). If C2’s output had been high (below limit set),
the transformer would have received no secondary drive
and there would be no output pulse. The demodulator is
designed so that nothing appears on the output line unless
the circuit is above the preset limit.
Comparator C1’s output damper network allows it to function as an op amp for low level signals. This circuit easily
extracts millivolt signals buried in high common mode
voltage or ground noise and delivers its limit decision
to the output. The maximum common mode voltage is
limited by the transformer’s 500V breakdown.
A = 5V/DIV
B = 10mA/DIV
C = 5V/DIV
D = 5V/DIV
500μs/DIV
AN11 F13
Figure 13. Isolated Limit Comparator Waveforms
an11f
AN11-11
Application Note 11
Fully Isolated 10-Bit A/D Converter
predominantly digital systems. It is also useful in industrial environments, where noise and high common mode
voltages are present in transducer fed systems.
Figure 14’s 5V powered circuit is a complete 10-bit A/D
converter which is fully floating from system ground.
It is ideal for performing 10-bit A/D conversions in the
face of the high common mode noise characteristics of
50k*
+ISOL
4700pF
POLYSTYRENE
15k*
LT1034
1.2V
0 TO 3VIN
INPUT VOLTAGE
5k
FULL-SCALE
TRIM
Q1
2N4250
+
+ISOL
10k
+
100pF
+ISOL
–
1M
1M
C1A
LT1018A
Q2
2N4250
+ISOL
–
82k
100k
C1B
LT1018B
4.7k
1M
+ISOL
4μF
POLYSTYRENE
4.7k
Q3
2N3904
3k
74C906
(5 IN PARALLEL)
DATA
OUTPUT
74C906
= 74C04
10k
Q4
2N3904
+ISOL
= 1N4148
* = 1% FILM RESISTOR
47pF
5V
10k
100Ω
“CONVERT COMMAND”
(500μS)
+
100μF
= GROUND
TRW PCO-35
= FLOATING COMMON
AN11 F14
Figure 14. Fully Isolated A/D Converter
an11f
AN11-12
Application Note 11
Circuit operation is initiated by applying a pulse to the
“convert command” input (Trace A, Figure 15). This pulse
appears at the transformer secondary and charges the
100μF capacitor. This potential is used to supply power
to the floating A/D conversion circuitry. The transformer’s
secondary pulse biases the inverter-open drain buffer
combination to discharge the 4μF capacitor (Trace B). The
secondary pulse also biases a diode to stop the C1B 3kHz
oscillator output (Trace D). Concurrently, C1A goes high,
forcing the inverter in its output line low (Trace C). When
the convert command pulse ceases, the Q1-Q2 current
source charges the 4μF capacitor with a linear ramp. The
C1B oscillator now runs. When the ramp crosses the input
voltage’s value, C1A’s output switches and its output line
inverter (Trace C) goes high, cutting off the oscillator. The
number of oscillator pulses occurring during this interval
is proportional to the input voltage value. These pulses are
differentiated and fed to Q3, which drives the transformer.
The differentiation causes narrow spikes to be fed to the
transformer, easing power drain on the 100μF energy
storage capacitor. Q3’s RC base delay and inverter-buffer
combination at C1B’s output prevent Q3’s emitter pulses
from triggering a ramp reset. The waveforms appearing at
the transformer’s input do not reflect the circuit’s complex
operation, and easily interface to a digital system. Figure 16,
Trace A shows the “convert command” pulse. Trace B is the
transformer primary. The differentiated oscillator pulses
coming from the transformer secondary appear as small
amplitude spikes. In this case, a small voltage is being
converted and the number of pulses is small. Trace C, the
“data output”, is taken at Q4’s collector, and the pulses
are TTL compatible.
A = 5V/DIV
B = 2V/DIV
C = 5V/DIV
D = 10V/DIV
50ms/DIV
AN11 F15
Figure 15. A/D Waveforms—Isolated Section
A = 5V/DIV
B = 2V/DIV
C = 5V/DIV
500μs/DIV
AN11 F16
Figure 16. Detail of A/D Waveforms—Grounded Section
an11f
AN11-13
Application Note 11
Several subtile factors contribute to the 10-bit performance
of this circuit. The 4700pF and 4μF polystyrene capacitors are both –120ppm/°C gain terms. Because of this,
their temperature drift’s ratio and overall circuit gain drift
is about 25ppm/°C. The five parallel 74C906 open-drain
buffers provide an effective 0V reset for the 4μF capacitor,
minimizing reset offset errors. Parallel inverters in C1B’s
output line reduce saturation caused errors, aiding oscillator stability with shifts in supply and temperature.
Finally, the diode path at Q3’s emitter averts a ±1 count
uncertainty error by synchronizing the oscillator to the
conversion sequence. The 5k potentiometer in the current source trims calibration to equal 1024 counts out for
3.000V input. The transformer used allows the converter
to function at common mode levels up to 175V. The circuit
requires 330ms to complete a 10-bit conversion and drifts
less than 1LSB over 25°C ±25°.
High Performance Single Supply Analog Building
Blocks
Two new components provide the basic building blocks
needed for high performance 5V single supply circuits.
The LT1014 quad op amp, also available as the LT1013
dual, features DC specifications nearly as good as
the best ±15V op amps. The LT1017/LT1018 series
dual comparators combine low power and high DC
precision with speed adequate for most applications.
To ease single supply operation, both units’ common
mode range includes ground, and the op amp’s output
swings nearly to ground.
LT1014 Basic Features
LT1017/LT1018 Basic Features
EOS—70μV
EOSΔTC—2μV/°C
IBIAS—15nA
Gain—1.0 × 106
Supply I—310μA
Noise—0.1Hz-10Hz-0.55μVP-P
Common Mode Range—Ground to (V+) – 1.5V
Supply Range—3.4V to 40V
Output Swing No Load
(V+) – 0.6V
(V–) + 0.015V
600Ω Load (V+) – 1.0V
(V–) + 0.005V
EOS—500μV
EOSΔTC—5μV/°C
IBIAS—15nA
Gain—1 Million
Supply I—LT1017—30μA, LT1018—110μA
Response Time—6μs (LT1018)
Common Mode Range—V– to (V+) – 0.9V
Supply Range—1.1V to 40V
Output Current—65mA Pull-Down—60μA Pull-Up
Output Stage Can Pull Down Loads Above VS
an11f
AN11-14
Application Note 11
Linear Power Supplies—Past, Present, and Future
The amplitude of linear circuitry’s power supplies has
been determined by the available technology used for
linear functions.
Probably the first standard linear supply was ±300V, used
in analog computers. The operating characteristics of
vacuum tubes necessitated high voltages. Additionally,
because analog computers (from which operational
amplifiers descended) were mathematical machines,
bipolar supplies were desirable for computational purposes. With the arrival of solid state linear components
in the early 1960s, new supply voltages were necessary
and desirable. ±15V was adapted because of transistor
breakdown limits, as well as the availability of low voltage references (Zener and avalanche diodes). Power and
size advantages afforded by the new supply standard
were also obviously attractive. It is significant that while
the supply drop decreased available dynamic range of
operation by over 20dB, the usable signal processing
range did not decrease. This was due to the lower noise
and drift of the solid state components.
The arrival of monolithic linear circuits in the late 1960s
and subsequent design refinements expanded the available territory at the lower end of the signal processing
range.
Currently available precision linear components and
technology shifts are causing reevaluation of the power
supply issue. In particular, several trends argue for linear
functions to be able to operate from the low voltage digital
rail. The increasing digital content of systems makes 5V
compatible linear components attractive. Cost and space
considerations in these systems often make separate
linear supplies undesirable. This situation is not universal,
and never will be, but is increasingly common.
A move to lower voltages for digital circuits, which
must occur, underscores the need for low voltage, high
performance linear ICs. Drops in digital supply value will
be forced by increasing density requirements, which will
lower IC breakdown limits.
Lower power consumption in systems goes along
with supply voltage and density issues. Increasingly
complex systems are being put into smaller physical
spaces, necessitating attention to dissipation. In many
instances, portable operation is desirable, so circuitry
must be directly compatible with battery potentials, as
well as lower power. Linear components must not give
up performance to function in this low voltage, digitally
driven environment. Demands for precision remain high,
and low voltage linear circuits, despite their narrowed
dynamic operating range, must meet these requirements. This is not easy, considering that a 12-bit ±15V
system typically has a 2.5mV/LSB error budget. At 5V,
this number shrinks to 1mV. To deal with this, design
techniques developed for ±15V components are being
used in new, low voltage ICs and new approaches will
also be employed.
Using Logic Supplies for Linear Functions
The fast clocking and transient high currents characteristic of digital systems make logic supplies an unfriendly
place for linear components to operate. A key to achieving good results is considering power bus routing as an
integral part of the signal processing chain. The figure
shows that supply rail impedances will cause both DC
and AC errors at various points in a system. This is
true of any power distribution scheme, but is especially
troublesome in digitally oriented systems, where fast current spiking and clock harmonics are present. Circuitry
located at position “A” will see appreciable positive rail
noise and ground potential will be corrupted by fast,
relatively high currents returning through conductor
impedances. Supply bypassing can reduce positive
rail noise, but ground potential uncertainty can cause
unacceptable errors. Locating linear circuits as shown
in “B” reduces both positive and ground rail problems
by eliminating the digitally derived currents. The linear
devices lower operating currents allow lower errors due to
supply distribution impedances. Appropriate bypassing
techniques are also shown. LC filters can be substantially more effective than simple capacitors, especially
in cases where it is not practical to route the positive
rail directly from the supply. RC filtering forces voltage
drop across the resistor, but is often acceptable due to
the linear components low power requirements.
an11f
Information furnished by Linear Technology Corporation is believed to be accurate and reliable.
However, no responsibility is assumed for its use. Linear Technology Corporation makes no representation that the interconnection of its circuits as described herein will not infringe on existing patent rights.
AN11-15
Application Note 11
In many cases it is not possible to arrange a “clean”
supply for the linear components. In such circumstances
it may be possible to synchronize noise sensitive linear
circuit operations to occur between system clock pulses.
This approach utilizes the synchronous nature of most
digital systems and the fact that supply bus disturbances
are often minimized between clock pulses.
Probably the most effective technique for dealing with
digital supply noise is to galvanically isolate the linear
circuits from the supply. The most obvious way to do
this is provision of separate power supplies for the
linear circuits, but this is often unacceptable. Instead,
transformer and optical isolation circuit techniques allow logic rail driven, galvanically isolated circuits (see
Figures 11, 12 and 14).
SUPPLY RAIL LOSSES
DIFFERENT SUPPLY VOLTAGES—
HIGH DIGITAL CURRENTS MEAN
LARGE ERRORS
POWER
SUPPLY
TERMINALS
I
DIGITAL
CIRCUITS
I
LINEAR
CIRCUITS
“A”
TO
OTHER
CIRCUITS
DIGITAL
CIRCUITS
I
DIFFERENT GROUND POTENTIALS—
HIGH DIGITAL CURRENTS MEAN
LARGE ERRORS
SUPPLY RAIL LOSSES
BYPASSING TECHNIQUES
SMALL LINEAR
CIRCUIT CURRENTS
MEAN SUPPLY
LOSSES HAVE
RELATIVELY SMALL
EFFECTS
TYPE
LINEAR
CIRCUITS
“B”
COMMENTS
+V
SIMPLE
+V
MORE COMPLEX, BUT
GOOD HIGH FREQUENCY
REJECTION WITH LOW DC
LOSSES. INAPPROPRIATE
FOR USE WITH FAST
LINEAR CIRCUITRY
GROUND RAIL LOSSES
TO
RAIL
+V
TO
RAIL
RC FILTER EFFECTIVE
FOR HIGH FREQUENCY
REJECTION BUT HAS DC
LOSSES. LOW CURRENT
LINEAR CIRCUITS CAN
ALLOW THIS APPROACH.
DO NOT USE WITH FAST
LINEAR
AN11 F17
an11f
AN11-16
Linear Technology Corporation
GP/IM 0985 10K • PRINTED IN USA
1630 McCarthy Blvd., Milpitas, CA 95035-7417
(408) 432-1900 ● FAX: (408) 434-0507
●
www.linear.com
© LINEAR TECHNOLOGY CORPORATION 1985