24-Bit ADC in Narrow SSOP-16 Package
(SO-8 Footprint)
■
Single Supply 2.7V to 5.5V Operation
■
Low Supply Current (200µA) and Auto Shutdown
■
Fully Differential Version of LTC2400
U
APPLICATIOS
■
Direct Sensor Digitizer
■
Weight Scales
■
Direct Temperature Measurement
■
Gas Analyzers
■
Strain-Gage Transducers
■
Instrumentation
■
Data Acquisition
■
Industrial Process Control
■
6-Digit DVMs
24-Bit No Latency ∆Σ
ADC
with Differential Input and
Differential Reference
U
TM
DESCRIPTIO
The LTC®2410 is a 2.7V to 5.5V micropower 24-bit
differential ∆Σ analog to digital converter with an inte-
grated oscillator, 2ppm INL and 0.16ppm RMS noise. It
uses delta-sigma technology and provides single cycle
settling time for multiplexed applications. Through a
single pin, the LTC2410 can be configured for better than
110dB input differential mode rejection at 50Hz or 60Hz
±2%, or it can be driven by an external oscillator for a user
defined rejection frequency. The internal oscillator requires no external frequency setting components.
The converter accepts any external differential reference
voltage from 0.1V to VCC for flexible ratiometric and
remote sensing measurement configurations. The fullscale differential input range is from –0.5V
The reference common mode voltage, V
input common mode voltage, V
dently set anywhere within the GND to VCC range of the
LTC2410. The DC common mode input rejection is better
than 140dB.
The LTC2410 communicates through a flexible 3-wire
digital interface which is compatible with SPI and
MICROWIRETM protocols.
, LTC and LT are registered trademarks of Linear Technology Corporation.
No Latency ∆Σ is a trademark of Linear Technology Corporation.
MICROWIRE is a trademark of National Semiconductor Corporation.
INCM
April 2000
to 0.5V
REF
REFCM
REF
, and the
, may be indepen-
.
TYPICAL APPLICATIOS
2.7V TO 5.5V
1µF
214
V
CC
LTC2410
3
REFERENCE
VOLTAGE
0.1V TO V
CC
ANALOG INPUT RANGE
TO 0.5V
–0.5V
REF
REF
1, 7, 8, 9, 10, 15, 16
+
REF
REF
IN
IN
GND
SCK
–
+
SDO
–
4
5
6
U
V
CC
= INTERNAL OSC/50Hz REJECTION
F
O
13
12
11
CS
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.
Internal SCK Duty Cycle(Note 10)●4555%
External SCK Frequency Range(Note 9)●2000kHz
External SCK Low Period(Note 9)●250ns
External SCK High Period(Note 9)●250ns
Internal SCK 32-Bit Data Output TimeInternal Oscillator (Notes 10, 12)●1.641.671.70ms
External SCK 32-Bit Data Output Time(Note 9)●32/f
CS ↓ to SDO Low Z●0200ns
SCK ↓ to SDO Valid●220ns
SDO Hold After SCK ↓(Note 5)●15ns
SCK Set-Up Before CS ↓●50ns
SCK Hold After CS ↓●50ns
= 2.7 to 5.5V unless otherwise specified.
CC
INCM
pin tied to GND or to VCC or to external conversion clock
O
= 153600Hz unless otherwise specified.
EOSC
= (REF+ + REF–)/2;
REFCM
= (IN+ + IN–)/2.
= 25°C. (Note 3)
A
= 153600Hz ±2%
EOSC
The ● denotes specifications which apply over the full operating temperature
F
= V
O
CC
External Oscillator (Note 11)
External Oscillator (Notes 10, 11)f
External Oscillator (Notes 10, 11)
Note 8: FO = VCC (internal oscillator) or f
(external oscillator).
Note 9: The converter is in external SCK mode of operation such that
the SCK pin is used as digital input. The frequency of the clock signal
driving SCK during the data output is f
Note 10: The converter is in internal SCK mode of operation such that
the SCK pin is used as digital output. In this mode of operation the
SCK pin has a total equivalent load capacitance C
Note 11: The external oscillator is connected to the FO pin. The external
oscillator frequency, f
Note 12: The converter uses the internal oscillator.
= 0V or FO = VCC.
F
O
Note 13: The output noise includes the contribution of the internal
calibration operations.
Note 14: Guaranteed by design and test correlation.
●157.03160.23163.44ms
●20510/f
●256/f
, is expressed in kHz.
EOSC
(in kHz)ms
EOSC
/8kHz
EOSC
(in kHz)ms
EOSC
(in kHz)ms
ESCK
= 128000Hz ±2%
EOSC
and is expressed in kHz.
ESCK
= 20pF.
LOAD
5
LTC2410
UUU
PI FUCTIOS
GND (Pins 1, 7, 8, 9, 10, 15, 16): Ground. Multiple
ground pins internally connected for optimum ground
current flow and VCC decoupling. Connect each one of
these pins to a ground plane through a low impedance
connection.
VCC (Pin 2): Positive Supply Voltage. Bypass to GND
(Pin␣ 1) with a 10µF tantalum capacitor in parallel with
0.1µF ceramic capacitor as close to the part as possible.
REF+ (Pin 3), REF– (Pin 4): Differential Reference Input.
The voltage on these pins can have any value between GND
and VCC as long as the reference positive input, REF+, is
maintained more positive than the reference negative
input, REF–, by at least 0.1V.
IN+ (Pin 5), IN– (Pin 6): Differential Analog Input. The
voltage on these pins can have any value between
GND – 0.3V and VCC + 0.3V. Within these limits the
converter bipolar input range (VIN = IN+ – IN–) extends
from – 0.5 • (V
the converter produces unique overrange and underrange
output codes.
CS (Pin 11): Active LOW Digital Input. A LOW on this pin
enables the SDO digital output and wakes up the ADC.
Following each conversion the ADC automatically enters
the Sleep mode and remains in this low power state as
long as CS is HIGH. A LOW-to-HIGH transition on CS
during the Data Output transfer aborts the data transfer
and starts a new conversion.
) to 0.5 • (V
REF
). Outside this input range
REF
SDO (Pin 12): Three-State Digital Output. During the Data
Output period this pin is used as serial data output. When
the chip select CS is HIGH (CS = VCC) the SDO pin is in a
high impedance state. During the Conversion and Sleep
periods this pin is used as the conversion status output.
The conversion status can be observed by pulling CS LOW.
SCK (Pin 13): Bidirectional Digital Clock Pin. In Internal
Serial Clock Operation mode, SCK is used as digital output
for the internal serial interface clock during the Data
Output period. In External Serial Clock Operation mode,
SCK is used as digital input for the external serial interface
clock during the Data Output period. A weak internal pullup is automatically activated in Internal Serial Clock Operation mode. The Serial Clock Operation mode is determined by the logic level applied to the SCK pin at power up
or during the most recent falling edge of CS.
FO (Pin 14): Frequency Control Pin. Digital input that
controls the ADC’s notch frequencies and conversion
time. When the FO pin is connected to VCC (FO = VCC), the
converter uses its internal oscillator and the digital filter
first null is located at 50Hz. When the FO pin is connected
to GND (FO = OV), the converter uses its internal oscillator
and the digital filter first null is located at 60Hz. When F
is driven by an external clock signal with a frequency f
the converter uses this signal as its system clock and the
digital filter first null is located at a frequency f
EOSC
EOSC
/2560.
O
,
6
LTC2410
1.69k
SDO
2410 TA04
Hi-Z TO V
OL
VOH TO V
OL
VOL TO Hi-Z
C
LOAD
= 20pF
V
CC
CONVERT
SLEEP
DATA OUTPUT
2410 F01
TRUE
FALSE
CS = LOW
AND
SCK
UU
W
FUCTIOAL BLOCK DIAGRA
V
CC
GND
+
IN
–
IN
+
–
∫∫∫
∑
+
REF
–
REF
–+
DAC
TEST CIRCUITS
SDO
ADC
AUTOCALIBRATION
AND CONTROL
DECIMATING FIR
INTERNAL
OSCILLATOR
SERIAL
INTERFACE
(INT/EXT)
F
O
SDO
SCK
CS
2410 FD
1.69k
Hi-Z TO V
VOL TO V
OH
VOH TO Hi-Z
U
OH
= 20pF
C
LOAD
2410 TA03
WUU
APPLICATIOS IFORATIO
CONVERTER OPERATION
Converter Operation Cycle
The LTC2410 is a low power, delta-sigma analog-todigital converter with an easy to use 3-wire serial interface.
Its operation is made up of three states. The converter
operating cycle begins with the conversion, followed by
the low power sleep state and ends with the data output
(see Figure 1). The 3-wire interface consists of serial data
output (SDO), serial clock (SCK) and chip select (CS).
Initially, the LTC2410 performs a conversion. Once the
conversion is complete, the device enters the sleep state.
While in this sleep state, power consumption is reduced by
an order of magnitude. The part remains in the sleep state
as long as CS is HIGH. The conversion result is held
indefinitely in a static shift register while the converter is
in the sleep state.
Figure 1. LTC2410 State Transition Diagram
7
LTC2410
U
WUU
APPLICATIOS IFORATIO
Once CS is pulled LOW, the device begins outputting the
conversion result. There is no latency in the conversion
result. The data output corresponds to the conversion just
performed. This result is shifted out on the serial data out
pin (SDO) under the control of the serial clock (SCK). Data
is updated on the falling edge of SCK allowing the user to
reliably latch data on the rising edge of SCK (see Figure 3).
The data output state is concluded once 32 bits are read
out of the ADC or when CS is brought HIGH. The device
automatically initiates a new conversion and the cycle
repeats.
Through timing control of the CS and SCK pins, the
LTC2410 offers several flexible modes of operation
(internal or external SCK and free-running conversion
modes). These various modes do not require programming configuration registers; moreover, they do not disturb the cyclic operation described above. These modes of
operation are described in detail in the Serial Interface
Timing Modes section.
Conversion Clock
A major advantage the delta-sigma converter offers over
conventional type converters is an on-chip digital filter
(commonly implemented as a Sinc or Comb filter). For
high resolution, low frequency applications, this filter is
typically designed to reject line frequencies of 50 or 60Hz
plus their harmonics. The filter rejection performance is
directly related to the accuracy of the converter system
clock. The LTC2410 incorporates a highly accurate onchip oscillator. This eliminates the need for external frequency setting components such as crystals or oscillators. Clocked by the on-chip oscillator, the LTC2410
achieves a minimum of 110dB rejection at the line frequency (50Hz or 60Hz ±2%).
Ease of Use
The LTC2410 data output has no latency, filter settling
delay or redundant data associated with the conversion
cycle. There is a one-to-one correspondence between the
conversion and the output data. Therefore, multiplexing
multiple analog voltages is easy.
The LTC2410 performs offset and full-scale calibrations
every conversion cycle. This calibration is transparent to
the user and has no effect on the cyclic operation described above. The advantage of continuous calibration is
extreme stability of offset and full-scale readings with respect to time, supply voltage change and temperature drift.
Power-Up Sequence
The LTC2410 automatically enters an internal reset state
when the power supply voltage VCC drops below approximately 2.2V. This feature guarantees the integrity of the
conversion result and of the serial interface mode selection. (See the 2-wire I/O sections in the Serial Interface
Timing Modes section.)
When the VCC voltage rises above this critical threshold,
the converter creates an internal power-on-reset (POR)
signal with a duration of approximately 0.5ms. The POR
signal clears all internal registers. Following the POR
signal, the LTC2410 starts a normal conversion cycle and
follows the succession of states described above. The first
conversion result following POR is accurate within the
specifications of the device if the power supply voltage is
restored within the operating range (2.7V to 5.5V) before
the end of the POR time interval.
Reference Voltage Range
This converter accepts a truly differential external reference voltage. The absolute/common mode voltage specification for the REF+ and REF– pins covers the entire range
from GND to VCC. For correct converter operation, the
REF+ pin must always be more positive than the REF– pin.
The LTC2410 can accept a differential reference voltage
from 0.1V to VCC. The converter output noise is determined by the thermal noise of the front-end circuits, and
as such, its value in nanovolts is nearly constant with
reference voltage. A decrease in reference voltage will not
significantly improve the converter’s effective resolution.
On the other hand, a reduced reference voltage will improve the converter’s overall INL performance. A reduced
reference voltage will also improve the converter performance when operated with an external conversion clock
(external FO signal) at substantially higher output data
rates (see the Output Data Rate section).
8
LTC2410
U
WUU
APPLICATIOS IFORATIO
Input Voltage Range
The analog input is truly differential with an absolute/
common mode range for the IN+ and IN– input pins
extending from GND – 0.3V to VCC + 0.3V. Outside
these limits the ESD protection devices begin to turn on
and the errors due to input leakage current increase
rapidly. Within these limits the LTC2410 converts the
bipolar differential input signal, VIN = IN+ – IN–, from
–FS = – 0.5 • V
REF+ – REF–. Outside this range the converter indicates the
overrange or the underrange condition using distinct
output codes.
Input signals applied to IN+ and IN– pins may extend by
300mV below ground and above VCC. In order to limit any
fault current, resistors of up to 5k may be added in series
with the IN+ and IN– pins without affecting the performance of the device. In the physical layout, it is important
to maintain the parasitic capacitance of the connection
between these series resistors and the corresponding pins
as low as possible; therefore, the resistors should be
located as close as practical to the pins. The effect of the
series resistance on the converter accuracy can be evaluated from the curves presented in the Input Current/
Reference Current sections. In addition, series resistors
will introduce a temperature dependent offset error due to
the input leakage current. A 1nA input leakage current will
develop a 1ppm offset error on a 5k resistor if V
This error has a very strong temperature dependency.
Output Data Format
The LTC2410 serial output data stream is 32 bits long. The
first 3 bits represent status information indicating the sign
and conversion state. The next 24 bits are the conversion
result, MSB first. The remaining 5 bits are sub LSBs
beyond the 24-bit level that may be included in averaging
or discarded without loss of resolution. The third and
fourth bit together are also used to indicate an underrange
condition (the differential input voltage is below –FS) or an
overrange condition (the differential input voltage is above
+FS).
Bit 31 (first output bit) is the end of conversion (EOC)
indicator. This bit is available at the SDO pin during the
conversion and sleep states whenever the CS pin is LOW.
to +FS = 0.5 • V
REF
where V
REF
REF
=
REF
= 5V.
This bit is HIGH during the conversion and goes LOW
when the conversion is complete.
Bit 30 (second output bit) is a dummy bit (DMY) and is
always LOW.
Bit 29 (third output bit) is the conversion result sign indicator (SIG). If VIN is >0, this bit is HIGH. If VIN is <0, this
bit is LOW.
Bit 28 (fourth output bit) is the most significant bit (MSB)
of the result. This bit in conjunction with Bit 29 also
provides the underrange or overrange indication. If both
Bit 29 and Bit 28 are HIGH, the differential input voltage is
above +FS. If both Bit 29 and Bit 28 are LOW, the
differential input voltage is below –FS.
The function of these bits is summarized in Table 1.
Table 1. LTC2410 Status Bits
Bit 31 Bit 30 Bit 29 Bit 28
Input RangeEOCDMYSIGMSB
VIN ≥ 0.5 • V
0V ≤ VIN < 0.5 • V
–0.5 • V
VIN < –0.5 • V
REF
≤ VIN < 0V0001
REF
REF
REF
0011
0010
0000
Bits 28-5 are the 24-bit conversion result MSB first.
Bit 5 is the least significant bit (LSB).
Bits 4-0 are sub LSBs below the 24-bit level. Bits 4-0 may
be included in averaging or discarded without loss of
resolution.
Data is shifted out of the SDO pin under control of the serial
clock (SCK), see Figure 3. Whenever CS is HIGH, SDO
remains high impedance and any externally generated
SCK clock pulses are ignored by the internal data out shift
register.
In order to shift the conversion result out of the device, CS
must first be driven LOW. EOC is seen at the SDO pin of the
device once CS is pulled LOW. EOC changes real time from
HIGH to LOW at the completion of a conversion. This
signal may be used as an interrupt for an external
microcontroller. Bit 31 (EOC) can be captured on the first
rising edge of SCK. Bit 30 is shifted out of the device on the
first falling edge of SCK. The final data bit (Bit 0) is shifted
out on the falling edge of the 31st SCK and may be latched
9
LTC2410
U
WUU
APPLICATIOS IFORATIO
on the rising edge of the 32nd SCK pulse. On the falling
edge of the 32nd SCK pulse, SDO goes HIGH indicating the
initiation of a new conversion cycle. This bit serves as EOC
(Bit 31) for the next conversion cycle. Table 2 summarizes
the output data format.
As long as the voltage on the IN+ and IN– pins is maintained
within the –0.3V to (VCC + 0.3V) absolute maximum
operating range, a conversion result is generated for any
differential input voltage VIN from –FS = –0.5 • V
+FS = 0.5 • V
. For differential input voltages greater than
REF
+FS, the conversion result is clamped to the value corresponding to the +FS + 1LSB. For differential input voltages
below –FS, the conversion result is clamped to the value
corresponding to –FS – 1LSB.
CS
REF
to
Frequency Rejection Selection (FO)
The LTC2410 internal oscillator provides better than 110dB
normal mode rejection at the line frequency and all its
harmonics for 50Hz ±2% or 60Hz ±2%. For 60Hz rejection, FO should be connected to GND while for 50Hz
rejection the FO pin should be connected to VCC.
The selection of 50Hz or 60Hz rejection can also be made
by driving FO to an appropriate logic level. A selection
change during the sleep or data output states will not
disturb the converter operation. If the selection is made
during the conversion state, the result of the conversion in
progress may be outside specifications but the following
conversions will not be affected.
When a fundamental rejection frequency different from
50Hz or 60Hz is required or when the converter must be
0.25 • V
000100 0 0…0
–1LSB00011 1 1…1
–0.25 • V
–0.25 • V
–0.5 • V
VIN* < –0.5 • V
*The differential input voltage VIN = IN+ – IN–.
**The differential reference voltage V
REF
** – 1LSB0 0101 1 1…1
REF
**00101 0 0…0
REF
** – 1LSB00100 1 1…1
REF
**00011 0 0…0
REF
** – 1LSB00010 1 1…1
REF
**00010 0 0…0
REF
REF
EOC
12345 262732
**00110 0 0…0
**00001 1 1…1
= REF+ – REF–.
REF
BIT 28BIT 29BIT 30
MSBSIG“0”
Figure 3. Output Data Timing
LSB
BIT 0BIT 27BIT 5
24
2410 F03
10
LTC2410
U
WUU
APPLICATIOS IFORATIO
synchronized with an outside source, the LTC2410 can
operate with an external conversion clock. The converter
automatically detects the presence of an external clock
signal at the FO pin and turns off the internal oscillator. The
frequency f
2560Hz (1Hz notch frequency) to be detected. The external clock signal duty cycle is not significant as long as the
minimum and maximum specifications for the high and
low periods t
While operating with an external conversion clock of a
frequency f
normal mode rejection in a frequency range f
±4% and its harmonics. The normal mode rejection as a
function of the input frequency deviation from f
is shown in Figure 4.
Whenever an external clock is not present at the FO pin the
converter automatically activates its internal oscillator and
enters the Internal Conversion Clock mode. The LTC2410
operation will not be disturbed if the change of conversion
clock source occurs during the sleep state or during the
data output state while the converter uses an external
serial clock. If the change occurs during the conversion
state, the result of the conversion in progress may be
outside specifications but the following conversions will
not be affected. If the change occurs during the data output
state and the converter is in the Internal SCK mode, the
serial clock duty cycle may be affected but the serial data
stream will remain valid.
of the external signal must be at least
EOSC
and t
HEO
, the LTC2410 provides better than 110dB
EOSC
are observed.
LEO
EOSC
EOSC
/2560
/2560
–80
–85
–90
–95
–100
–105
–110
–115
–120
–125
NORMAL MODE REJECTION (dB)
–130
–135
–140
–12–8–404812
DIFFERENTIAL INPUT SIGNAL FREQUENCY
DEVIATION FROM NOTCH FREQUENCY f
Figure 4. LTC2410 Normal Mode Rejection When
Using an External Oscillator of Frequency f
EOSC
/2560(%)
2410 F04
EOSC
Table 3 summarizes the duration of each state and the
achievable output data rate as a function of FO.
SERIAL INTERFACE PINS
The LTC2410 transmits the conversion results and receives the start of conversion command through a synchronous 3-wire interface. During the conversion and
sleep states, this interface can be used to assess the
converter status and during the data output state it is used
to read the conversion result.
Table 3. LTC2410 State Duration
StateOperating ModeDuration
CONVERTInternal OscillatorFO = LOW133ms, Output Data Rate ≤ 7.5 Readings/s
SLEEPAs Long As CS = HIGH Until CS = LOW and SCK
DATA OUTPUTInternal Serial ClockFO = LOW/HIGHAs Long As CS = LOW But Not Longer Than 1.67ms
(Internal Oscillator)(32 SCK cycles)
FO = External Oscillator withAs Long As CS = LOW But Not Longer Than 256/f
Frequency f
External Serial Clock withAs Long As CS = LOW But Not Longer Than 32/f
Frequency f
kHz(32 SCK cycles)
SCK
EOSC
kHz
EOSC
kHz(32 SCK cycles)
s, Output Data Rate ≤ f
EOSC
/20510 Readings/s
EOSC
SCK
EOSC
ms
ms
11
LTC2410
U
WUU
APPLICATIOS IFORATIO
Serial Clock Input/Output (SCK)
The serial clock signal present on SCK (Pin 13) is used to
synchronize the data transfer. Each bit of data is shifted out
the SDO pin on the falling edge of the serial clock.
In the Internal SCK mode of operation, the SCK pin is an
output and the LTC2410 creates its own serial clock by
dividing the internal conversion clock by 8. In the External
SCK mode of operation, the SCK pin is used as input. The
internal or external SCK mode is selected on power-up and
then reselected every time a HIGH-to-LOW transition is
detected at the CS pin. If SCK is HIGH or floating at powerup or during this transition, the converter enters the internal SCK mode. If SCK is LOW at power-up or during this
transition, the converter enters the external SCK mode.
Serial Data Output (SDO)
The serial data output pin, SDO (Pin 12), provides the
result of the last conversion as a serial bit stream (MSB
first) during the data output state. In addition the SDO pin
is used as an end of conversion indicator during the
conversion and sleep states.
When CS (Pin 11) is HIGH, the SDO driver is switched to
a high impedance state. This allows sharing the serial
interface with other devices. If CS is LOW during the
convert or sleep state, SDO will output EOC. If CS is LOW
during the conversion phase, the EOC bit appears HIGH on
the SDO pin. Once the conversion is complete, EOC goes
LOW. The device remains in the sleep state until the first
rising edge of SCK occurs while CS = LOW.
Chip Select Input (CS)
The active LOW chip select, CS (Pin 11), is used to test the
conversion status and to enable the data output transfer as
described in the previous sections.
In addition, the CS signal can be used to trigger a new
conversion cycle before the entire serial data transfer has
been completed. The LTC2410 will abort any serial data
transfer in progress and start a new conversion cycle
anytime a LOW-to-HIGH transition is detected at the CS
pin after the converter has entered the data output state
(i.e., after the first rising edge of SCK occurs with
CS␣=␣LOW).
Finally, CS can be used to control the free-running modes
of operation, see Serial Interface Timing Modes section.
Grounding CS will force the ADC to continuously convert
at the maximum output rate selected by FO. Tying a
capacitor to CS will reduce the output rate and power
dissipation by a factor proportional to the capacitor’s
value, see Figures 12 to 14.
SERIAL INTERFACE TIMING MODES
The LTC2410’s 3-wire interface is SPI and MICROWIRE
compatible. This interface offers several flexible modes of
operation. These include internal/external serial clock,
2- or 3-wire I/O, single cycle conversion and autostart. The
following sections describe each of these serial interface
timing modes in detail. In all these cases, the converter
can use the internal oscillator (FO = LOW or FO = HIGH) or
an external oscillator connected to the FO pin. Refer to
Table␣ 4 for a summary.
External Serial Clock, Single Cycle Operation
(SPI/MICROWIRE Compatible)
This timing mode uses an external serial clock to shift out
the conversion result and a CS signal to monitor and
control the state of the conversion cycle, see Figure 5.
Table 4. LTC2410 Interface Timing Modes
ConversionDataConnection
SCKCycleOutputand
ConfigurationSourceControlControlWaveforms
External SCK, Single Cycle ConversionExternalCS and SCKCS and SCKFigures 5, 6
External SCK, 2-Wire I/OExternalSCKSCKFigure 7
Internal SCK, Single Cycle ConversionInternalCS ↓CS ↓Figures 8, 9
Internal SCK, 2-Wire I/O, Continuous ConversionInternalContinuousInternalFigure 10
Internal SCK, Autostart ConversionInternalC
EXT
InternalFigure 11
12
LTC2410
U
WUU
APPLICATIOS IFORATIO
The serial clock mode is selected on the falling edge of CS.
To select the external serial clock mode, the serial clock pin
(SCK) must be LOW during each CS falling edge.
The serial data output pin (SDO) is Hi-Z as long as CS is
HIGH. At any time during the conversion cycle, CS may be
pulled LOW in order to monitor the state of the converter.
While CS is pulled LOW, EOC is output to the SDO pin.
EOC␣ =␣ 1 while a conversion is in progress and EOC = 0 if
the device is in the sleep state. Independent of CS, the
device automatically enters the low power sleep state once
the conversion is complete.
When the device is in the sleep state (EOC = 0), its
conversion result is held in an internal static shift register. The device remains in the sleep state until the first
rising edge of SCK is seen while CS is LOW. Data is
out the SDO pin on each falling edge of SCK. This enables
external circuitry to latch the output on the rising edge of
SCK. EOC can be latched on the first rising edge of SCK
and the last bit of the conversion result can be latched on
the 32nd rising edge of SCK. On the 32nd falling edge of
SCK, the device begins a new conversion. SDO goes HIGH
(EOC = 1) indicating a conversion is in progress.
At the conclusion of the data cycle, CS may remain LOW
and EOC monitored as an end-of-conversion interrupt.
Alternatively, CS may be driven HIGH setting SDO to Hi-Z.
shifted
As described above, CS may be pulled LOW at any time in
order to monitor the conversion status.
Typically, CS remains LOW during the data output state.
However, the data output state may be aborted by pulling
CS HIGH anytime between the first rising edge and the
32nd falling edge of SCK, see Figure 6. On the rising edge
of CS, the device aborts the data output state and immediately initiates a new conversion. This is useful for systems not requiring all 32 bits of output data, aborting an
invalid conversion cycle or synchronizing the start of a
conversion.
External Serial Clock, 2-Wire I/O
This timing mode utilizes a 2-wire serial I/O interface. The
conversion result is shifted out of the device by an externally generated serial clock (SCK) signal, see Figure 7. CS
may be permanently tied to ground, simplifying the user
interface or isolation barrier.
The external serial clock mode is selected at the end of the
power-on reset (POR) cycle. The POR cycle is concluded
approximately 0.5ms after VCC exceeds 2.2V. The level
applied to SCK at this time determines if SCK is internal or
external. SCK must be driven LOW prior to the end of POR
in order to enter the external serial clock timing mode.
Figure 6. External Serial Clock, Reduced Data Output Length
Since CS is tied LOW, the end-of-conversion (EOC) can be
continuously monitored at the SDO pin during the convert
and sleep states. EOC may be used as an interrupt to an
external controller indicating the conversion result is
ready. EOC = 1 while the conversion is in progress and
EOC␣ =␣ 0 once the conversion enters the low power sleep
state. On the falling edge of EOC, the conversion result is
loaded into an internal static shift register. The device
remains in the sleep state until the first rising edge of SCK.
Data is shifted out the SDO pin on each falling edge of SCK
enabling external circuitry to latch data on the rising edge
of SCK. EOC can be latched on the first rising edge of SCK.
On the 32nd falling edge of SCK, SDO goes HIGH (EOC␣ =␣ 1)
indicating a new conversion has begun.
Internal Serial Clock, Single Cycle Operation
This timing mode uses an internal serial clock to shift out
the conversion result and a CS signal to monitor and
control the state of the conversion cycle, see Figure 8.
In order to select the internal serial clock timing mode, the
serial clock pin (SCK) must be floating (Hi-Z) or pulled
HIGH prior to the falling edge of CS. The device will not
enter the internal serial clock mode if SCK is driven LOW
on the falling edge of CS. An internal weak pull-up resistor
is active on the SCK pin during the falling edge of CS;
therefore, the internal serial clock timing mode is automatically selected if SCK is not externally driven.
The serial data output pin (SDO) is Hi-Z as long as CS is
HIGH. At any time during the conversion cycle, CS may be
pulled LOW in order to monitor the state of the converter.
Once CS is pulled LOW, SCK goes LOW and EOC is output
to the SDO pin. EOC = 1 while a conversion is in progress
and EOC = 0 if the device is in the sleep state.
When testing EOC, if the conversion is complete (EOC = 0),
the device will exit the sleep state and enter the data output
state if CS remains LOW. In order to prevent the device
from exiting the low power sleep state, CS must be pulled
HIGH before the first rising edge of SCK. In the internal
SCK timing mode, SCK goes HIGH and the device begins
outputting data at time t
(if EOC = 0) or t
after EOC goes LOW (if CS is LOW
EOCtest
during the falling edge of EOC). The value of t
after the falling edge of CS
EOCtest
EOCtest
is 23µs
if the device is using its internal oscillator (F0 = logic LOW
or HIGH). If FO is driven by an external oscillator of
Figure 8. Internal Serial Clock, Single Cycle Operation
15
LTC2410
U
WUU
APPLICATIOS IFORATIO
frequency f
HIGH before time t
state. The conversion result is held in the internal static
shift register.
If CS remains LOW longer than t
edge of SCK will occur and the conversion result is serially
shifted out of the SDO pin. The data output cycle begins on
this first rising edge of SCK and concludes after the 32nd
rising edge. Data is shifted out the SDO pin on each falling
edge of SCK. The internally generated serial clock is output
to the SCK pin. This signal may be used to shift the
conversion result into external circuitry. EOC can be
latched on the first rising edge of SCK and the last bit of the
conversion result on the 32nd rising edge of SCK. After the
32nd rising edge, SDO goes HIGH (EOC = 1), SCK stays
HIGH and a new conversion starts.
Typically, CS remains LOW during the data output state.
However, the data output state may be aborted by pulling
CS HIGH anytime between the first and 32nd rising edge
of SCK, see Figure 9. On the rising edge of CS, the device
aborts the data output state and immediately initiates a
EOSC
, then t
EOCtest
EOCtest
is 3.6/f
. If CS is pulled
EOSC
, the device remains in the sleep
, the first rising
EOCtest
new conversion. This is useful for systems not requiring
all 32 bits of output data, aborting an invalid conversion
cycle, or synchronizing the start of a conversion. If CS is
pulled HIGH while the converter is driving SCK LOW, the
internal pull-up is not available to restore SCK to a logic
HIGH state. This will cause the device to exit the internal
serial clock mode on the next falling edge of CS. This can
be avoided by adding an external 10k pull-up resistor to
the SCK pin or by never pulling CS HIGH when SCK is LOW.
Whenever SCK is LOW, the LTC2410’s internal pull-up at
pin SCK is disabled. Normally, SCK is not externally driven
if the device is in the internal SCK timing mode. However,
certain applications may require an external driver on SCK.
If this driver goes Hi-Z after outputting a LOW signal, the
LTC2410’s internal pull-up remains disabled. Hence, SCK
remains LOW. On the next falling edge of CS, the device is
switched to the external SCK timing mode. By adding an
external 10k pull-up resistor to SCK, this pin goes HIGH
once the external driver goes Hi-Z. On the next CS falling
edge, the device will remain in the internal SCK timing
mode.
Figure 9. Internal Serial Clock, Reduced Data Output Length
LTC2410
U
WUU
APPLICATIOS IFORATIO
A similar situation may occur during the sleep state when
CS is pulsed HIGH-LOW-HIGH in order to test the conversion status. If the device is in the sleep state (EOC = 0), SCK
will go LOW. Once CS goes HIGH (within the time period
defined above as t
For a heavy capacitive load on the SCK pin, the internal
pull-up may not be adequate to return SCK to a HIGH level
before CS goes low again. This is not a concern under
normal conditions where CS remains LOW after detecting
EOC = 0. This situation is easily overcome by adding an
external 10k pull-up resistor to the SCK pin.
Internal Serial Clock, 2-Wire I/O,
Continuous Conversion
This timing mode uses a 2-wire, all output (SCK and SDO)
interface. The conversion result is shifted out of the device
by an internally generated serial clock (SCK) signal, see
Figure 10. CS may be permanently tied to ground, simplifying the user interface or isolation barrier.
The internal serial clock mode is selected at the end of the
power-on reset (POR) cycle. The POR cycle is concluded
approximately 0.5ms after VCC exceeds 2.2V. An internal
), the internal pull-up is activated.
EOCtest
weak pull-up is active during the POR cycle; therefore, the
internal serial clock timing mode is automatically selected
if SCK is not externally driven LOW (if SCK is loaded such
that the internal pull-up cannot pull the pin HIGH, the
external SCK mode will be selected).
During the conversion, the SCK and the serial data output
pin (SDO) are HIGH (EOC = 1). Once the conversion is
complete, SCK and SDO go LOW (EOC = 0) indicating the
conversion has finished and the device has entered the
low power sleep state. The part remains in the sleep state
a minimum amount of time (1/2 the internal SCK period)
then immediately begins outputting data. The data output
cycle begins on the first rising edge of SCK and ends after
the 32nd rising edge. Data is shifted out the SDO pin on
each falling edge of SCK. The internally generated serial
clock is output to the SCK pin. This signal may be used
to shift the conversion result into external circuitry. EOC
can be latched on the first rising edge of SCK and the last
bit of the conversion result can be latched on the 32nd
rising edge of SCK. After the 32nd rising edge, SDO goes
HIGH (EOC = 1) indicating a new conversion is in progress.
SCK remains HIGH during the conversion.
Figure 10. Internal Serial Clock, Continuous Operation
17
LTC2410
U
WUU
APPLICATIOS IFORATIO
Internal Serial Clock, Autostart Conversion
This timing mode is identical to the internal serial clock,
2-wire I/O described above with one additional feature.
Instead of grounding CS, an external timing capacitor is
tied to CS.
While the conversion is in progress, the CS pin is held
HIGH by an internal weak pull-up. Once the conversion is
complete, the device enters the low power sleep state and
an internal 25nA current source begins discharging the
capacitor tied to CS, see Figure 11. The time the converter
spends in the sleep state is determined by the value of the
external timing capacitor, see Figures 12 and 13. Once the
voltage at CS falls below an internal threshold (≈1.4V), the
device automatically begins outputting data. The data
output cycle begins on the first rising edge of SCK and
ends on the 32nd rising edge. Data is shifted out the SDO
pin on each falling edge of SCK. The internally generated
serial clock is output to the SCK pin. This signal may be
used to shift the conversion result into external circuitry.
After the 32nd rising edge, CS is pulled HIGH and a new
conversion is immediately started. This is useful in applications requiring periodic monitoring and ultralow power.
Figure 14 shows the average supply current as a function
of capacitance on CS.
It should be noticed that the external capacitor discharge
current is kept very small in order to decrease the converter power dissipation in the sleep state. In the autostart
mode the analog voltage on the CS pin cannot be observed
without disturbing the converter operation using a regular
oscilloscope probe. When using this configuration, it is
important to minimize the external leakage current at the
CS pin by using a low leakage external capacitor and
properly cleaning the PCB surface.
The internal serial clock mode is selected every time the
voltage on the CS pin crosses an internal threshold voltage. An internal weak pull-up at the SCK pin is active while
Figure 11. Internal Serial Clock, Autostart Operation
LTC2410
U
WUU
APPLICATIOS IFORATIO
7
6
5
4
(SEC)
3
SAMPLE
t
2
1
0
10100100000
1
CAPACITANCE ON CS (pF)
Figure 12. CS Capacitance vs t
8
7
6
5
4
3
SAMPLE RATE (Hz)
2
1
0
0
VCC = 5V
1010010000
CAPACITANCE ON CS (pF)
Figure 13. CS Capacitance vs Output Rate
300
)
RMS
250
200
150
VCC = 5V
VCC = 3V
VCC = 5V
100010000
V
= 3V
CC
1000
V
= 3V
CC
2400 F12
SAMPLE
100000
2400 F13
CS is discharging; therefore, the internal serial clock
timing mode is automatically selected if SCK is floating. It
is important to ensure there are no external drivers pulling
SCK LOW while CS is discharging.
PRESERVING THE CONVERTER ACCURACY
The LTC2410 is designed to reduce as much as possible
the conversion result sensitivity to device decoupling,
PCB layout, antialiasing circuits, line frequency perturbations and so on. Nevertheless, in order to preserve the
extreme accuracy capability of this part, some simple
precautions are desirable.
Digital Signal Levels
The LTC2410’s digital interface is easy to use. Its digital
inputs (FO, CS and SCK in External SCK mode of operation)
accept standard TTL/CMOS logic levels and the internal
hysteresis receivers can tolerate edge rates as slow as
100µs. However, some considerations are required to take
advantage of the exceptional accuracy and low supply
current of this converter.
The digital output signals (SDO and SCK in Internal SCK
mode of operation) are less of a concern because they are
not generally active during the conversion state.
While a digital input signal is in the range 0.5V to
(VCC␣ –␣ 0.5V), the CMOS input receiver draws additional
current from the power supply. It should be noted that,
when any one of the digital input signals (FO, CS and SCK
in External SCK mode of operation) is within this range, the
LTC2410 power supply current may increase even if the
signal in question is at a valid logic level. For micropower
operation, it is recommended to drive all digital input
signals to full CMOS levels [VIL < 0.4V and VOH >
(VCC – 0.4V)].
100
SUPPLY CURRENT (µA
50
0
1
10100100010000
CAPACITANCE ON CS (pF)
100000
2400 F14
Figure 14. CS Capacitance vs Supply Current
During the conversion period, the undershoot and/or
overshoot of a fast digital signal connected to the LTC2410
pins may severely disturb the analog to digital conversion
process. Undershoot and overshoot can occur because of
the impedance mismatch at the converter pin when the
transition time of an external control signal is less than
twice the propagation delay from the driver to LTC2410.
For reference, on a regular FR-4 board, signal propagation
19
LTC2410
U
WUU
APPLICATIOS IFORATIO
velocity is approximately 183ps/inch for internal traces
and 170ps/inch for surface traces. Thus, a driver generating a control signal with a minimum transition time of
1ns must be connected to the converter pin through a
trace shorter than 2.5 inches. This problem becomes
particularly difficult when shared control lines are used
and multiple reflections may occur. The solution is to
carefully terminate all transmission lines close to their
characteristic impedance.
Parallel termination near the LTC2410 pin will eliminate
this problem but will increase the driver power dissipation.
A series resistor between 27Ω and 56Ω placed near the
driver or near the LTC2410 pin will also eliminate this
problem without additional power dissipation. The actual
resistor value depends upon the trace impedance and
connection topology.
An alternate solution is to reduce the edge rate of the
control signals. It should be noted that using very slow
edges will increase the converter power supply current
during the transition time. The multiple ground pins used
in this package configuration, as well as the differential
input and reference architecture, reduce substantially the
converter’s sensitivity to ground currents.
Particular attention must be given to the connection of the
FO signal when the LTC2410 is used with an external
conversion clock. This clock is active during the conversion time and the normal mode rejection provided by the
internal digital filter is not very high at this frequency. A
normal mode signal of this frequency at the converter
reference terminals may result into DC gain and INL
errors. A normal mode signal of this frequency at the
converter input terminals may result into a DC offset error.
Such perturbations may occur due to asymmetric capacitive coupling between the FO signal trace and the converter
input and/or reference connection traces. An immediate
solution is to maintain maximum possible separation
between the FO signal trace and the input/reference signals. When the FO signal is parallel terminated near the
converter, substantial AC current is flowing in the loop
formed by the FO connection trace, the termination and the
ground return path. Thus, perturbation signals may be
inductively coupled into the converter input and/or reference. In this situation, the user must reduce to a minimum
the loop area for the FO signal as well as the loop area for
the differential input and reference connections.
Driving the Input and Reference
The input and reference pins of the LTC2410 converter are
directly connected to a network of sampling capacitors.
Depending upon the relation between the differential input
voltage and the differential reference voltage, these capacitors are switching between these four pins transfering
small amounts of charge in the process. A simplified
equivalent circuit is shown in Figure 15.
For a simple approximation, the source impedance R
driving an analog input pin (IN+, IN–, REF+ or REF–) can be
considered to form, together with RSW and CEQ (see
Figure␣ 15), a first order passive network with a time
constant τ = (RS + RSW) • CEQ. The converter is able to
sample the input signal with better than 1ppm accuracy if
the sampling period is at least 14 times greater than the
input circuit time constant τ. The sampling process on the
four input analog pins is quasi-independent so each time
constant should be considered by itself and, under worstcase circumstances, the errors may add.
When using the internal oscillator (FO = LOW or HIGH), the
LTC2410’s front-end switched-capacitor network is clocked
at 76800Hz corresponding to a 13µs sampling period.
Thus, for settling errors of less than 1ppm, the driving
source impedance should be chosen such that τ ≤ 13µs/14
= 920ns. When an external oscillator of frequency f
used, the sampling period is 2/f
error of less than 1ppm, τ ≤ 0.14/f
Input Current
If complete settling occurs on the input, conversion results will be unaffected by the dynamic input current. An
incomplete settling of the input signal sampling process
may result in gain and offset errors, but it will not degrade
the INL performance of the converter. Figure 15 shows the
mathematical expressions for the average bias currents
flowing through the IN+ and IN– pins as a result of the
sampling charge transfers when integrated over a substantial time period (longer than 64 internal clock cycles).
and, for a settling
EOSC
.
EOSC
EOSC
S
is
20
LTC2410
R
SOURCE
(Ω)
1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05
+FS ERROR (ppm OF V
REF
)
2410 F18
0
–10
–20
–30
–40
–50
VCC = 5V
REF
+
= 5V
REF
–
= GND
IN
+
= GND
IN
–
= 2.5V
F
O
= GND
T
A
= 25°C
CIN = 0.01µF
CIN = 0.001µF
CIN = 100pF
CIN = 0pF
U
WUU
APPLICATIOS IFORATIO
V
CC
I
+
REF
V
+
REF
IIN+
VIN+
IIN–
VIN–
I
–
REF
V
–
REF
SWITCHING FREQUENCY
= 76800Hz INTERNAL OSCILLATOR (FO = LOW OR HIGH)
f
SW
= 0.5 • f
f
SW
I
LEAK
I
LEAK
V
CC
V
CC
I
LEAK
I
LEAK
V
CC
EXTERNAL OSCILLATOR
EOSC
I
LEAK
I
LEAK
I
LEAK
I
LEAK
RSW (TYP)
20k
RSW (TYP)
20k
RSW (TYP)
20k
RSW (TYP)
20k
2410 F15
Figure 15. LTC2410 Equivalent Analog Input Circuit
C
EQ
18pF
(TYP)
VV V
+−
IN INCMREFCM
+
IIN
=
()
AVG
−
IIN
()
AVG
+
IREF
()
AVG
−
IREF
()
AVG
::
where
=−
VREFREF
REF
=
V
REFCM
+−
=−
VININ
IN
=
V
INCM
==
RM INTERNAL OSCILLATOR Hz Notch FLOW
36160
.
EQO
RM INTERNAL OSCILLATOR Hz Notch F HIGH
==
43250
.
EQO
RfEXTERNAL OSCILLATOR
=•
0 555 10
()
EQEOSC
.
R
•
05
−+ −
=
=
=
+−
REFREF
+−
IN IN
Ω
Ω
./
EQ
VV V
IN INCMREFCM
.
•
R
05
EQ
•− +
.
VV V
15
REF INCMREFCM
•
.
R
05
EQ
.
VV V
−• − +
15
REF INCMREFCM
.
R
•
05
+−
+
2
−
2
12
2
V
IN
−
•
VR
REF EQ
2
V
IN
+
VR
EQ
•
REF EQ
()
()
The effect of this input dynamic current can be analyzed
using the test circuit of Figure 16. The C
includes the LTC2410 pin capacitance (5pF typical) plus
the capacitance of the test fixture used to obtain the results
shown in Figures 17 and 18. A careful implementation can
bring the total input capacitance (CIN + C
thus achieving better performance than the one predicted
by Figures 17 and 18. For simplicity two distinct situations
can be considered.
50
40
)
REF
30
20
+FS ERROR (ppm OF V
10
0
1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05
CIN = 0.001µF
VCC = 5V
+
= 5V
REF
–
= GND
REF
+
= 5V
IN
–
= 2.5V
IN
= GND
F
O
= 25°C
T
A
CIN = 0.01µF
CIN = 100pF
CIN = 0pF
R
(Ω)
SOURCE
at IN+ or IN– (Small CIN)
SOURCE
capacitor
PAR
) closer to 5pF
PAR
2410 F17
R
SOURCE
V
+ 0.5V
V
INCM
INCM
– 0.5V
IN
R
SOURCE
IN
Figure 16. An RC Network at IN+ and IN
Figure 18. –FS Error vs R
C
IN
C
IN
SOURCE
+
IN
C
PAR
≅20pF
C
≅20pF
PAR
LTC2410
–
IN
2410 F16
–
at IN+ or IN– (Small CIN)Figure 17. +FS Error vs R
21
LTC2410
U
WUU
APPLICATIOS IFORATIO
For relatively small values of input capacitance (CIN <
0.01µF), the voltage on the sampling capacitor settles
almost completely and relatively large values for the
source impedance result in only small errors. Such values
for CIN will deteriorate the converter offset and gain
performance without significant benefits of signal filtering
and the user is advised to avoid them. Nevertheless, when
small values of CIN are unavoidably present as parasitics
of input multiplexers, wires, connectors or sensors, the
LTC2410 can maintain its exceptional accuracy while
operating with relative large values of source resistance as
shown in Figures 17 and 18. These measured results may
be slightly different from the first order approximation
suggested earlier because they include the effect of the
actual second order input network together with the nonlinear settling process of the input amplifiers. For small C
values, the settling on IN+ and IN– occurs almost independently and there is little benefit in trying to match the
source impedance for the two pins.
Larger values of input capacitors (CIN > 0.01µF) may be
required in certain configurations for antialiasing or general input signal filtering. Such capacitors will average the
input sampling charge and the external source resistance
will see a quasi constant input differential impedance.
When FO = LOW (internal oscillator and 60Hz notch), the
typical differential input resistance is 1.8MΩ which will
generate a gain error of approximately 0.28ppm for each
ohm of source resistance driving IN+ or IN–. When FO =
HIGH (internal oscillator and 50Hz notch), the typical
differential input resistance is 2.16MΩ which will generate
a gain error of approximately 0.23ppm for each ohm of
source resistance driving IN+ or IN–. When FO is driven by
an external oscillator with a frequency f
conversion clock operation), the typical differential input
resistance is 0.28 • 1012/f
source resistance driving IN+ or IN– will result in
1.78 • 10–6 • f
resistance on the two input pins is additive with respect to
this gain error. The typical +FS and –FS errors as a function
of the sum of the source resistance seen by IN+ and IN– for
large values of CIN are shown in Figures 19 and 20.
In addition to this gain error, an offset error term may also
appear. The offset error is proportional with the mismatch
between the source impedance driving the two input pins
ppm gain error. The effect of the source
EOSC
Ω and each ohm of
EOSC
EOSC
(external
IN
IN+ and IN– and with the difference between the input and
reference common mode voltages. While the input drive
circuit nonzero source impedance combined with the
converter average input current will not degrade the INL
performance, indirect distortion may result from the modulation of the offset error by the common mode component
of the input signal. Thus, when using large CIN capacitor
values, it is advisable to carefully match the source impedance seen by the IN+ and IN– pins. When FO = LOW
(internal oscillator and 60Hz notch), every 1Ω mismatch
in source impedance transforms a full-scale common
mode input signal into a differential mode input signal of
0.28ppm. When FO = HIGH (internal oscillator and 50Hz
notch), every 1Ω mismatch in source impedance transforms a full-scale common mode input signal into a
differential mode input signal of 0.23ppm. When FO is
driven by an external oscillator with a frequency f
every 1Ω mismatch in source impedance transforms a
full-scale common mode input signal into a differential
mode input signal of 1.78 • 10–6 • f
shows the typical offset error due to input common mode
voltage for various values of source resistance imbalance
between the IN+ and IN– pins when large CIN values are
used.
If possible, it is desirable to operate with the input signal
common mode voltage very close to the reference signal
common mode voltage as is the case in the ratiometric
measurement of a symmetric bridge. This configuration
eliminates the offset error caused by mismatched source
impedances.
The magnitude of the dynamic input current depends upon
the size of the very stable internal sampling capacitors and
upon the accuracy of the converter sampling clock. The
accuracy of the internal clock over the entire temperature
and power supply range is typical better than 0.5%. Such
a specification can also be easily achieved by an external
clock. When relatively stable resistors (50ppm/°C) are
used for the external source impedance seen by IN+ and
IN–, the expected drift of the dynamic current, offset and
gain errors will be insignificant (about 1% of their respective values over the entire temperature and voltage range).
Even for the most stringent applications a one-time calibration operation may be sufficient.
ppm. Figure 21
EOSC
EOSC
,
22
LTC2410
U
WUU
APPLICATIOS IFORATIO
300
VCC = 5V
+
= 5V
REF
–
= GND
REF
240
)
REF
+FS ERROR (ppm OF V
+
= 3.75V
IN
–
= 1.25V
IN
F
= GND
O
180
= 25°C
T
A
120
60
0
100 200 300 400 500 600 700 800 900 1000
0
Figure 19. +FS Error vs R
0
–60
)
REF
–120
VCC = 5V
–180
–FS ERROR (ppm OF V
–240
–300
+
= 5V
REF
–
= GND
REF
+
= 1.25V
IN
–
= 3.75V
IN
= GND
F
O
= 25°C
T
A
100 200 300 400 500 600 700 800 900 1000
0
Figure 20. –FS Error vs R
120
100
)
REF
–20
–40
–60
–FS ERROR (ppm OF V
–80
–100
–120
A
80
60
B
40
C
20
D
0
E
F
G
0
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
A: ∆RIN = +400Ω
= +200Ω
B: ∆R
IN
= +100Ω
C: ∆R
IN
= 0Ω
D: ∆R
IN
Figure 21. Offset Error vs Common Mode Voltage
(V
= IN+ = IN–) and Input Source Resistance Imbalance
INCM
(∆R
IN
= R
SOURCEIN
+ – R
SOURCEIN
CIN = 1µF, 10µF
CIN = 0.1µF
CIN = 0.01µF
R
(Ω)
SOURCE
at IN+ or IN– (Large CIN)
SOURCE
CIN = 0.01µF
CIN = 0.1µF
CIN = 1µF, 10µF
R
(Ω)
SOURCE
at IN+ or IN– (Large CIN)
SOURCE
VCC = 5V
+
= 5V
REF
–
= GND
REF
+
= IN– = V
IN
FO = GND
T
A
R
C
= 25°C
SOURCEIN
= 10µF
IN
V
INCM
INCM
– = 500Ω
(V)
E: ∆RIN = –100Ω
F: ∆R
G: ∆R
= –200Ω
IN
= –400Ω
IN
2410 F19
2410 F20
2410 F21
–) for Large CIN Values (C
≥ 1µF)
IN
In addition to the input sampling charge, the input ESD
protection diodes have a temperature dependent leakage
current. This current, nominally 1nA (±10nA max), results
in a small offset shift. A 100Ω source resistance will create
a 0.1µV typical and 1µV maximum offset voltage.
Reference Current
In a similar fashion, the LTC2410 samples the differential
reference pins REF+ and REF– transfering small amount of
charge to and from the external driving circuits thus
producing a dynamic reference current. This current does
not change the converter offset but it may degrade the gain
and INL performance. The effect of this current can be
analyzed in the same two distinct situations.
For relatively small values of the external reference capacitors (C
< 0.01µF), the voltage on the sampling capacitor
REF
settles almost completely and relatively large values for
the source impedance result in only small errors. Such
values for C
will deteriorate the converter offset and
REF
gain performance without significant benefits of reference
filtering and the user is advised to avoid them.
Larger values of reference capacitors (C
> 0.01µF) may
REF
be required as reference filters in certain configurations.
Such capacitors will average the reference sampling charge
and the external source resistance will see a quasi constant reference differential impedance. When FO = LOW
(internal oscillator and 60Hz notch), the typical differential
reference resistance is 1.3MΩ which will generate a gain
error of approximately 0.38ppm for each ohm of source
resistance driving REF+ or REF–. When FO = HIGH (internal
oscillator and 50Hz notch), the typical differential reference resistance is 1.56MΩ which will generate a gain error
of approximately 0.32ppm for each ohm of source resistance driving REF+ or REF–. When FO is driven by an
external oscillator with a frequency f
(external conver-
EOSC
sion clock operation), the typical differential reference
resistance is 0.20 • 1012/f
Ω and each ohm of source
EOSC
resistance drving REF+ or REF– will result in
2.47 • 10–6 • f
ppm gain error. The effect of the source
EOSC
resistance on the two reference pins is additive with
respect to this gain error. The typical +FS and –FS errors
for various combinations of source resistance seen by the
23
LTC2410
R
SOURCE
(Ω)
1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05
–FS ERROR (ppm OF V
REF
)
2410 F23
50
40
30
20
10
0
VCC = 5V
REF
+
= 5V
REF
–
= GND
IN
+
= GND
IN
–
= 2.5V
F
O
= GND
T
A
= 25°C
C
REF
= 0.01µF
C
REF
= 0.001µF
C
REF
= 100pF
C
REF
= 0pF
R
SOURCE
(Ω)
0
100 200 300 400 500 600 700 800 900 1000
+FS ERROR (ppm OF V
REF
)
2410 F25
450
360
270
180
90
0
VCC = 5V
REF
+
= 5V
REF
–
= GND
IN
+
= 1.25V
IN
–
= 3.75V
F
O
= GND
T
A
= 25°C
C
REF
= 0.01µF
C
REF
= 0.1µF
C
REF
= 1µF, 10µF
U
WUU
APPLICATIOS IFORATIO
REF+ and REF– pins and external capacitance C
nected to these pins are shown in Figures 22, 23, 24
and␣ 25.
In addition to this gain error, the converter INL performance is degraded by the reference source impedance.
When FO = LOW (internal oscillator and 60Hz notch), every
100Ω of source resistance driving REF+ or REF– translates
into about 1.34ppm additional INL error. When FO = HIGH
(internal oscillator and 50Hz notch), every 100Ω of source
resistance driving REF+ or REF– translates into about
1.1ppm additional INL error. When FO is driven by an
0
VCC = 5V
+
= 5V
REF
–
= GND
REF
–10
)
REF
–20
IN
IN
F
O
T
A
+
= 5V
–
= 2.5V
= GND
= 25°C
REF
con-
external oscillator with a frequency f
, every 100Ω of
EOSC
source resistance driving REF+ or REF– translates into
about 8.73 • 10–6 • f
ppm additional INL error.
EOSC
Figure␣ 26 shows the typical INL error due to the source
resistance driving the REF+ or REF– pins when large C
REF
values are used. The effect of the source resistance on the
two reference pins is additive with respect to this INL error.
In general, matching of source impedance for the REF
+
and REF– pins does not help the gain or the INL error. The
user is thus advised to minimize the combined source
impedance driving the REF+ and REF– pins rather than to
try to match it.
–30
+FS ERROR (ppm OF V
–40
–50
1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05
C
REF
C
REF
C
Figure 22. +FS Error vs R
0
–90
)
REF
–180
VCC = 5V
+
= 5V
REF
–
= GND
REF
+
= 3.75V
IN
–
= 1.25V
IN
= GND
F
O
= 25°C
T
A
100 200 300 400 500 600 700 800 900 1000
0
+FS ERROR (ppm OF V
–270
–360
–450
Figure 24. +FS Error vs R
24
= 0.01µF
= 0.001µF
= 100pF
REF
C
= 0pF
REF
R
(Ω)
SOURCE
at REF+ or REF– (Small CIN)
SOURCE
C
= 0.01µF
REF
C
= 0.1µF
REF
C
= 1µF, 10µF
REF
R
(Ω)
SOURCE
at REF+ and REF– (Large C
SOURCE
2410 F22
2410 F24
Figure 23. –FS Error vs R
)Figure 25. – FS Error vs R
REF
at REF+ or REF– (Small CIN)
SOURCE
at REF+ and REF– (Large C
SOURCE
REF
)
LTC2410
U
WUU
APPLICATIOS IFORATIO
15
R
12
9
)
6
REF
3
0
–3
R
= 100Ω
INL (ppm OF V
–12
–15
Figure 26. INL vs Differential Input Voltage (VIN = IN+ – IN–)
and Reference Source Resistance (R
Large C
Values (C
REF
SOURCE
–6
–9
–0.5–0.4–0.3–0.2–0.1 0 0.1 0.2 0.3 0.4 0.5
VCC = 5V
REF+ = 5V
REF– = GND
= 0.5 • (IN+ + IN–) = 2.5V
V
INCM
≥ 1µF)
REF
The magnitude of the dynamic reference current depends
upon the size of the very stable internal sampling capacitors and upon the accuracy of the converter sampling
clock. The accuracy of the internal clock over the entire
temperature and power supply range is typical better than
0.5%. Such a specification can also be easily achieved by
an external clock. When relatively stable resistors
(50ppm/°C) are used for the external source impedance
seen by REF+ and REF–, the expected drift of the dynamic
current gain error will be insignificant (about 1% of its
value over the entire temperature and voltage range). Even
for the most stringent applications a one-time calibration
operation may be sufficient.
In addition to the reference sampling charge, the reference
pins ESD protection diodes have a temperature dependent
leakage current. This leakage current, nominally 1nA
(±10nA max), results in a small gain error. A 100Ω source
resistance will create a 0.05µV typical and 0.5µV maxi-
mum full-scale error.
Output Data Rate
When using its internal oscillator, the LTC2410 can produce up to 7.5 readings per second with a notch frequency
of 60Hz (FO = LOW) and 6.25 readings per second with a
notch frequency of 50Hz (FO = HIGH). The actual output
SOURCE
R
SOURCE
V
INDIF/VREFDIF
= 1000Ω
= 500Ω
SOURCE
FO = GND
= 10µF
C
REF
= 25°C
T
A
2410 F26
at REF+ and REF– for
data rate will depend upon the length of the sleep and data
output phases which are controlled by the user and which
can be made insignificantly short. When operated with an
external conversion clock (FO connected to an external
oscillator), the LTC2410 output data rate can be increased
as desired. The duration of the conversion phase is 20510/
f
EOSC
. If f
= 153600Hz, the converter behaves as if the
EOSC
internal oscillator is used and the notch is set at 60Hz.
There is no significant difference in the LTC2410 performance between these two operation modes.
An increase in f
over the nominal 153600Hz will
EOSC
translate into a proportional increase in the maximum
output data rate. This substantial advantage is nevertheless accompanied by three potential effects, which must
be carefully considered.
First, a change in f
will result in a proportional change
EOSC
in the internal notch position and in a reduction of the
converter differential mode rejection at the power line
frequency. In many applications, the subsequent performance degradation can be substantially reduced by relying upon the LTC2410’s exceptional common mode rejection and by carefully eliminating common mode to differential mode conversion sources in the input circuit. The
user should avoid single-ended input filters and should
maintain a very high degree of matching and symmetry in
the circuits driving the IN+ and IN– pins.
Second, the increase in clock frequency will increase
proportionally the amount of sampling charge transferred
through the input and the reference pins. If large external
input and/or reference capacitors (CIN, C
) are used, the
REF
previous section provides formulae for evaluating the
effect of the source resistance upon the converter performance for any value of f
or reference capacitors (CIN, C
. If small external input and/
EOSC
) are used, the effect of
REF
the external source resistance upon the LTC2410 typical
performance can be inferred from Figures 17, 18, 22 and
23 in which the horizontal axis is scaled by 153600/f
EOSC
.
Third, an increase in the frequency of the external oscillator above 460800Hz (a more than 3× increase in the output
data rate) will start to decrease the effectiveness of the
internal autocalibration circuits. This will result in a progressive degradation in the converter accuracy and linear-
25
LTC2410
OUTPUT DATA RATE (READINGS/SEC)
0 102030405060708090100
OFFSET ERROR (ppm OF V
REF
)
2410 F27
500
450
400
350
300
250
200
150
100
50
0
TA = 85°C
VCC = 5V
REF
+
= 5V
REF
–
= GND
V
INCM
= 2.5V
V
IN
= 0V
F
O
= EXTERNAL OSCILLATOR
TA = 25°C
APPLICATIOS IFORATIO
ity. Typical measured performance curves for output data
rates up to 100 readings per second are shown in Figures␣ 27, 28, 29, 30, 31, 32, 33 and 34. In order to obtain
the highest possible level of accuracy from this converter
at output data rates above 20 readings per second, the
user is advised to maximize the power supply voltage used
and to limit the maximum ambient operating temperature.
In certain circumstances, a reduction of the differential
reference voltage may be beneficial.
Input Bandwidth
The combined effect of the internal Sinc4 digital filter and
of the analog and digital autocalibration circuits determines the LTC2410 input bandwidth. When the internal
oscillator is used with the notch set at 60Hz (FO = LOW),
the 3dB input bandwidth is 3.63Hz. When the internal
oscillator is used with the notch set at 50Hz (FO = HIGH),
the 3dB input bandwidth is 3.02Hz. If an external conversion clock generator of frequency f
FO pin, the 3dB input bandwidth is 0.236 • 10–6 • f
Due to the complex filtering and calibration algorithms
utilized, the converter input bandwidth is not modeled very
accurately by a first order filter with the pole located at the
3dB frequency. When the internal oscillator is used, the
shape of the LTC2410 input bandwidth is shown in Figure␣ 35 for FO = LOW and FO = HIGH. When an external
oscillator of frequency f
LTC2410 input bandwidth can be derived from Figure␣ 35,
FO = LOW curve in which the horizontal axis is scaled by
f
/153600.
EOSC
The conversion noise (800nV
can be modeled by a white noise source connected to a
noise free converter. The noise spectral density is
62.75nV√Hz for an infinite bandwidth source and
76.8nV√Hz for a single 0.5MHz pole source. From these
numbers, it is clear that particular attention must be given
to the design of external amplification circuits. Such
circuits face the simultaneous requirements of very low
bandwidth (just a few Hz) in order to reduce the output
referred noise and relatively high bandwidth (at least
500kHz) necessary to drive the input switched-capacitor
network. A possible solution is a high gain, low bandwidth
amplifier stage followed by a high bandwidth unity-gain
buffer.
26
U
is used, the shape of the
EOSC
RMS
WUU
is connected to the
EOSC
typical for V
REF
.
EOSC
= 5V)
Figure 27. Offset Error vs Output Data Rate and Temperature
7000
VCC = 5V
6000
)
REF
5000
4000
3000
2000
+FS ERROR (ppm OF V
1000
+
= 5V
REF
–
= GND
REF
+
= 3.75V
IN
–
= 1.25V
IN
= EXTERNAL OSCILLATOR
F
O
TA = 85°C
TA = 25°C
0
0 102030405060708090100
OUTPUT DATA RATE (READINGS/SEC)
2410 F28
Figure 28. +FS Error vs Output Data Rate and Temperature
0
–1000
)
REF
–2000
–3000
–4000
–5000
–FS ERROR (ppm OF V
–6000
–7000
VCC = 5V
+
= 5V
REF
–
= GND
REF
+
= 1.25V
IN
–
= 3.75V
IN
= EXTERNAL OSCILLATOR
F
O
0 102030405060708090100
OUTPUT DATA RATE (READINGS/SEC)
TA = 85°C
TA = 25°C
2410 F29
Figure 29. –FS Error vs Output Data Rate and Temperature
LTC2410
OUTPUT DATA RATE (READINGS/SEC)
0 102030405060708090100
RESOLUTION (BITS)
2410 F31
22
20
18
16
14
12
10
8
TA = 85°C
VCC = 5V
REF
+
= 5V
REF
–
= GND
V
INCM
= 2.5V
–2.5V < V
IN
< 2.5V
F
O
= EXTERNAL OSCILLATOR
RESOLUTION = LOG2(V
REF
/INL
MAX
)
TA = 25°C
OUTPUT DATA RATE (READINGS/SEC)
0 102030405060708090100
RESOLUTION (BITS)
2410 F33
24
23
22
21
20
19
18
17
16
15
14
13
12
V
REF
= 5V
VCC = 5V
REF
–
= GND
V
INCM
= 2.5V
V
IN
= 0V
F
O
= EXTERNAL OSCILLATOR
T
A
= 25°C
RESOLUTION = LOG
2(VREF
/NOISE
RMS
)
V
REF
= 2.5V
U
WUU
APPLICATIOS IFORATIO
24
23
22
21
20
19
18
VCC = 5V
17
16
RESOLUTION (BITS)
15
14
13
12
+
= 5V
REF
–
= GND
REF
= 2.5V
V
INCM
= 0V
V
IN
= EXTERNAL OSCILLATOR
F
O
RESOLUTION = LOG
0 102030405060708090100
OUTPUT DATA RATE (READINGS/SEC)
Figure 30. Resolution (Noise
vs Output Data Rate and Temperature
250
VCC = 5V
225
)
200
REF
175
150
125
100
75
50
OFFSET ERROR (ppm OF V
25
+
= GND
REF
= 2.5V
V
INCM
= 0V
V
IN
= EXTERNAL OSCILLATOR
F
O
= 25°C
T
A
0
0 102030405060708090100
OUTPUT DATA RATE (READINGS/SEC)
TA = 85°C
2(VREF
V
= 2.5V
REF
TA = 25°C
/NOISE
RMS
V
= 5V
REF
)
RMS
2410 F30
≤ 1LSB)
2410 F32
Figure 31. Resolution (INL
RMS
≤ 1LSB)
vs Output Data Rate and Temperature
Figure 32. Offset Error vs Output
Data Rate and Reference Voltage
Figure 34. Resolution (INL
22
RESOLUTION =
/INL
20
18
16
V
= 2.5VV
14
RESOLUTION (BITS)
12
10
8
REF
TA = 25°C
= 5V
V
CC
–
= GND
REF
= 0.5 • REF
V
INCM
–0.5V • V
REF
FO = EXTERNAL OSCILLATOR
0 102030405060708090100
OUTPUT DATA RATE (READINGS/SEC)
≤ 1LSB) vs Output Data Rate and Reference Voltage
MAX
LOG
REF
+
< VIN < 0.5 • V
2(VREF
= 5V
REF
Figure 33. Resolution (Noise
≤ 1LSB) vs
RMS
Output Data Rate and Reference Voltage
)
MAX
2410 F34
27
LTC2410
U
WUU
APPLICATIOS IFORATIO
Normal Mode Rejection and Antialiasing
One of the advantages delta-sigma ADCs offer over conventional ADCs is on-chip digital filtering. Combined with
a large oversampling ratio, the LTC2410 significantly
simplifies antialiasing filter requirements.
The Sinc4 digital filter provides greater than 120dB normal
mode rejection at all frequencies except DC and integer
multiples of the modulator sampling frequency (fS). The
LTC2410’s autocalibration circuits further simplify the
antialiasing requirements by additional normal mode signal filtering both in the analog and digital domain. Independent of the operating mode, fS = 256 • fN = 2048 •
f
OUTMAX
the maximum output data rate. In the internal oscillator
mode with a 50Hz notch setting, fS = 12800Hz and with a
60Hz notch setting fS = 15360Hz. In the external oscillator
mode, fS = f
The combined normal mode rejection performance is
shown in Figure␣ 36 for the internal oscillator with 50Hz
notch setting (FO = HIGH) and in Figure␣ 37 for the internal
oscillator with 60Hz notch setting (FO = LOW) and for the
external oscillator mode. The regions of low rejection
occurring at integer multiples of fS have a very narrow
bandwidth. Magnified details of the normal mode rejection
curves are shown in Figure␣ 38 (rejection near DC) and
Figure␣ 39 (rejection at fS = 256fN) where fN represents the
notch frequency. These curves have been derived for the
external oscillator mode but they can be used in all
operating modes by appropriately selecting the fN value.
The user can expect to achieve in practice this level of
performance using the internal oscillator as it is demonstrated by Figures 40 and 41. Typical measured values of
the normal mode rejection of the LTC2410 operating with
an internal oscillator and a 60Hz notch setting are shown
in Figure 40 superimposed over the theoretical calculated
curve. Similarly, typical measured values of the normal
mode rejection of the LTC2410 operating with an internal
oscillator and a 50Hz notch setting are shown in Figure 41
superimposed over the theoretical calculated curve.
As a result of these remarkable normal mode specifications, minimal (if any) antialias filtering is required in front
of the LTC2410. If passive RC components are placed in
Figure 41. Input Normal Mode Rejection vs Input Frequency
29
LTC2410
U
WUU
APPLICATIOS IFORATIO
front of the LTC2410, the input dynamic current should be
considered (see Input Current section). In cases where
large effective RC time constants are used, an external
buffer amplifier may be required to minimize the effects of
dynamic input current.
Traditional high order delta-sigma modulators, while providing very good linearity and resolution, suffer from potential instabilities at large input signal levels. The proprietary architecture used for the LTC2410 third order modulator resolves this problem and guarantees a predictable
stable behavior at input signal levels of up to 150% of full
scale. In many industrial applications, it is not uncommon
to have to measure microvolt level signals superimposed
over volt level perturbations and LTC2410 is eminently
suited for such tasks. When the perturbation is differential,
the specification of interest is the normal mode rejection
0
–20
–40
–60
for large input signal levels. With a reference voltage
V
␣ =␣ 5V, the LTC2410 has a full-scale differential input
REF
range of 5V peak-to-peak. Figures 42 and 43 show measurement results for the LTC2410 normal mode rejection
ratio with a 7.5V peak-to-peak (150% of full scale) input
signal superimposed over the more traditional normal mode
rejection ratio results obtained with a 5V peak-to-peak (full
scale) input signal. In Figure 42, the LTC2410 uses the
internal oscillator with the notch set at 60Hz (FO = LOW)
and in Figure 43 it uses the internal oscillator with the
notch set at 50Hz (FO = HIGH). It is clear that the LTC2410
rejection performance is maintained with no compromises
in this extreme situation. When operating with large input
signal levels, the user must observe that such signals do
not violate the device absolute maximum ratings.
Figure 43. Measured Input Normal Mode Rejection vs Input Frequency
30
LTC2410
U
WUU
APPLICATIOS IFORATIO
SYNCHRONIZATION OF MULTIPLE LTC2410s
Since the LTC2410’s absolute accuracy (total unadjusted
error) is 5ppm, applications utilizing multiple synchronized ADCs are possible.
Simultaneous Sampling with Two LTC2410s
One such application is synchronizing multiple LTC2410s,
see Figure 44. The start of conversion is synchronized to
the rising edge of CS. In order to synchronize multiple
LTC2410s, CS is a common input to all the ADCs.
To prevent the converters from autostarting a new conversion at the end of data output read, 31 or fewer SCK
clock signals are applied to the LTC2410 instead of 32 (the
32nd falling edge would start a conversion). The exact
timing and frequency for the SCK signal is not critical
since it is only shifting out the data. In this case, two
LTC2410’s simultaneously start and end their conversion
cycles under the external control of CS.
Increasing the Output Rate Using Mulitple LTC2410s
A second application uses multiple LTC2410s to increase
the effective output rate by 4×, see Figure 45. In this case,
four LTC2410s are interleaved under the control of separate CS signals. This increases the effective output rate
from 7.5Hz to 30Hz (up to a maximum of 60Hz). Additionally, the one-shot output spectrum is unfolded allowing
further digital signal processing of the conversion results.
SCK and SDO may be common to all four LTC2410s. The
four CS rising edges equally divide one LTC2410 conversion cycle (7.5Hz for 60Hz notch frequency). In order to
synchronize the start of conversion to CS, 31 or less SCK
clock pulses must be applied to each ADC.
Both the synchronous and 4× output rate applications use
the external serial clock and single cycle operation with
reduced data output length (see Serial Interface Timing
Modes section and Figure 6). An external oscillator clock
is applied commonly to the FO pin of each LTC2410 in
order to synchronize the sampling times. Both circuits
may be extended to include more LTC2410s.
SCK1
SCK2
SDO1
SDO2
SCK2
SCK1
LTC2410
#1
V
F
CC
O
+
REF
SCK
–
SDO
µCONTROLLER
CS
SDO1
SDO2
CS
31 OR LESS CLOCK CYCLES
REF
IN
IN
GND
+
CS
–
31 OR LESS CLOCK CYCLES
V
REF
REF
IN
IN
GND
CC
+
LTC2410
#2
+
–
–
SCK
SDO
F
O
CS
EXTERNAL OSCILLATOR
(153,600HZ)
V
+
REF
V
–
REF
2410 F44
Figure 44. Synchronous Conversion—Extendable
31
LTC2410
U
WUU
APPLICATIOS IFORATIO
LTC2410
#1
V
F
CC
O
+
REF
SCK
–
SDO
REF
+
CS
IN
–
IN
µCONTROLLER
SCK
SDO
CS1
CS2
CS3
CS4
CS1
CS2
CS3
GND
LTC2410
V
CC
REF
REF
+
IN
–
IN
GND
+
V
REF
V
–
REF
EXTERNAL OSCILLATOR
(153,600HZ)
V
REF
REF
IN
IN
GND
CC
LTC2410
#3
+
–
+
–
SCK
SDO
F
O
CS
#2
F
O
+
SCK
–
SDO
CS
V
REF
REF
IN
IN
GND
CC
+
–
LTC2410
#4
+
–
SCK
SDO
F
O
CS
CS4
31 OR LESS
SCK
SDO
CLOCK PULSES
Figure 45. Actual Frequency Rate LTC2410 System
BRIDGE APPLICATIONS
Typical strain gauge based bridges deliver only 2mV/Volt
of excitation. As the maximum reference voltage of the
LTC2410 is 5V, remote sensing of applied excitation
without additional circuitry requires that excitation be
limited to 5V. This gives only 10mV full scale, which can
be resolved to 1 part in 10000 without averaging. For many
solid state sensors, this is still better than the sensor. For
example, averaging 64 samples however reduces the
noise level by a factor of eight, bringing the resolving
power to 1 part in 80000, comparable to better weighing
systems. Hysteresis and creep effects in the load cells are
typically much greater than this. Most applications that
require strain measurements to this level of accuracy are
measuring slowly changing phenomena, hence the time
required to average a large number of readings is usually
2410 F45
not an issue. For those systems that require accurate
measurement of a small incremental change on a significant tare weight, the lack of history effects in the LTC2400
family is of great benefit.
For those applications that cannot be fulfilled by the
LTC2410 alone, compensating for error in external amplification can be done effectively due to the “no latency”
feature of the LTC2410. No latency operation allows
samples of the amplifier offset and gain to be interleaved
with weighing measurements. The use of correlated double
sampling allows suppression of 1/f noise, offset and
thermocouple effects within the bridge. Correlated double
sampling involves alternating the polarity of excitation and
dealing with the reversal of input polarity mathematically.
Alternatively, bridge excitation can be increased to as
much as ±10V, if one of several precision attenuation
32
LTC2410
U
WUU
APPLICATIOS IFORATIO
techniques is used to produce a precision divide operation
on the reference signal. Another option is the use of a
reference within the 5V input range of the LTC2410 and
developing excitation via fixed gain, or LTC1043 based
voltage multiplication, along with remote feedback in the
excitation amplifiers, as shown in Figures 34 and 35.
Figure 46 shows an example of a simple bridge connection. Note that it is suitable for any bridge application
where measurement speed is not of the utmost importance. For many applications where large vessels are
weighed, the average weight over an extended period of
time is of concern and short term weight is not readily
determined due to movement of contents, or mechanical
resonance. Often, large weighing applications involve load
cells located at each load bearing point, the output of
which can be summed passively prior to the signal processing circuitry, actively with amplification prior to the
ADC, or can be digitized via multiple ADC channels and
summed mathematically. The mathematical summation
of the output of multiple LTC2410’s provides the benefit of
a root square reduction in noise. The low power consumption of the LTC2410 makes it attractive for multidrop
communication schemes where the ADC is located within
the load-cell housing.
A direct connection to a load cell is perhaps best incorporated into the load-cell body, as minimizing the distance to
the sensor largely eliminates the need for protection
LT1019
+
R1
3
4350Ω
BRIDGE
5
6
R2
R1 AND R2 CAN BE USED TO INCREASE TOLERABLE AC COMPONENT ON REF SIGNALS
Figure 46. Simple Bridge Connection
REF
REF
IN
IN
+
–
2
V
REF
+
–
LTC2410
GND
1, 7, 8, 9,
10, 15, 16
SDO
SCK
12
13
11
CS
14
F
O
2410 F46
devices, RFI suppression and wiring. The LTC2410 exhibits extremely low temperature dependent drift. As a result,
exposure to external ambient temperature ranges does
not compromise performance. The incorporation of any
amplification considerably complicates thermal stability,
as input offset voltages and currents, temperature coefficient of gain settling resistors all become factors.
The circuit in Figure 47 shows an example of a simple
amplification scheme. This example produces a differential output with a common mode voltage of 2.5V, as
determined by the bridge. The use of a true three amplifier
instrumentation amplifier is not necessary, as the LTC2410
has common mode rejection far beyond that of most
amplifiers. The LTC1051 is a dual autozero amplifier that
can be used to produce a gain of 15 before its input
referred noise dominates the LTC2410 noise. This example shows a gain of 34, that is determined by a feedback
network built using a resistor array containing 8 individual
resistors. The resistors are organized to optimize temperature tracking in the presence of thermal gradients. The
second LTC1051 buffers the low noise input stage from
the transient load steps produced during conversion.
The gain stability and accuracy of this approach is very
good, due to a statistical improvement in resistor matching due to individual error contribution being reduced. A
gain of 34 may seem low, when compared to common
practice in earlier generations of load-cell interfaces, however the accuracy of the LTC2410 changes the rationale.
Achieving high gain accuracy and linearity at higher gains
may prove difficult, while providing little benefit in terms
of noise reduction.
At a gain of 100, the gain error that could result from
typical open-loop gain of 160dB is –1ppm, however,
worst-case is at the minimum gain of 116dB, giving a gain
error of –158ppm. Worst-case gain error at a gain of 34,
is –54ppm. The use of the LTC1051A reduces the worstcase gain error to –33ppm. The advantage of gain higher
than 34, then becomes dubious, as the input referred
noise sees little improvement1 and gain accuracy is potentially compromised.
Note that this 4-amplifier topology has advantages over
the typical integrated 3-amplifier instrumentation amplifier in that it does not have the high noise level common in
33
LTC2410
U
WUU
APPLICATIOS IFORATIO
the output stage that usually dominates when an instrumentation amplifier is used at low gain. If this amplifier is
used at a gain of 10, the gain error is only 10ppm and input
referred noise is reduced to 0.1µV
can also be configured to provide gain of up to 50 with high
gain stability and linearity.
Figure 48 shows an example of a single amplifier used to
produce single-ended gain. This topology is best used in
applications where the gain setting resistor can be made
to match the temperature coefficient of the strain gauges.
If the bridge is composed of precision resistors, with only
one or two variable elements, the reference arm of the
bridge can be made to act in conjunction with the feedback
resistor to determine the gain. If the feedback resistor is
incorporated into the design of the load cell, using resistors which match the temperature coefficient of the loadcell elements, good results can be achieved without the
need for resistors with a high degree of absolute accuracy.
The common mode voltage in this case, is again a function
of the bridge output. Differential gain as used with a 350Ω
bridge is AV = 1+ R2/(R1+175Ω). Common mode gain is
half the differential gain. The maximum differential signal
that can be used is 1/4 V
, as opposed to 1/2 V
REF
2-amplifier topology above.
. The buffer stages
RMS
REF
in the
Remote Half Bridge Interface
As opposed to full bridge applications, typical half bridge
applications must contend with nonlinearity in the bridge
output, as signal swing is often much greater. Applications
include RTD’s, thermistors and other resistive elements
that undergo significant changes over their span. For
single variable element bridges, the nonlinearity of the half
bridge output can be eliminated completely; if the reference arm of the bridge is used as the reference to the ADC,
as shown in Figure 49. The LTC2410 can accept inputs up
to 1/2 V
. Hence, the reference resistor R1 must be at
REF
least 2x the highest value of the variable resistor.
In the case of 100Ω platinum RTD’s, this would suggest a
value of 800Ω for R1. Such a low value for R1 is not
advisable due to self-heating effects. A value of 25.5k is
shown for R1, reducing self-heating effects to acceptable
levels for most sensors.
The basic circuit shown in Figure 49 shows connections
for a full 4-wire connection to the sensor, which may be
located remotely. The differential input connections will
reject induced or coupled 60Hz interference, however, the
1
Input referred noise for AV = 34 for approximately 0.05µV
Figure 47. Using Autozero Amplifiers to Reduce Input Referred Noise
5V
REF
0.1µF
2
V
CC
312
+
4
5
6
2410 F47
REF
REF
IN
IN
–
+
LTC2410
–
GND
SDO
SCK
CS
F
O
1, 7, 8, 9,
10, 15, 16
13
11
14
34
LTC2410
U
WUU
APPLICATIOS IFORATIO
reference inputs do not have the same rejection. If 60Hz or
other noise is present on the reference input, a low pass
filter is recommended as shown in Figure 50. Note that you
cannot place a large capacitor directly at the junction of R1
and R2, as it will store charge from the sampling process.
A better approach is to produce a low pass filter decoupled
from the input lines with a high value resistor (R3).
The use of a third resistor in the half bridge, between the
variable and fixed elements gives essentially the same
result as the two resistor version, but has a few benefits.
If, for example, a 25k reference resistor is used to set the
excitation current with a 100Ω RTD, the negative reference input is sampling the same external node as the
positive input, but may result in errors if used with a long
cable. For short cable applications, the errors may be
acceptalby low. If instead the single 25k resistor is replaced with a 10k 5% and a 10k 0.1% negative reference
resistor, the noise level introduced at the reference, at
least at higher frequencies, will be reduced. A filter can be
introduced into the network, in the form of one or more
capacitors, or ferrite beads, as long as the sampling pulses
are not translated into an error. The reference voltage is
also reduced, but this is not undesirable, as it will decrease
the value of the LSB, although, not the input referred noise
level.
The circuit shown in Figure 50 shows a more rigorous
example of Figure 49, with increased noise suppression
and more protection for remote applications.
Figure 51 shows an example of gain in the excitation circuit
and remote feedback from the bridge. The LTC1043’s
provide voltage multiplication, providing ±10V from a 5V
reference with only 1ppm error. The amplifiers are used at
unity-gain and, hence, introduce a very little error due to
gain error or due to offset voltages. A 1µV/°C offset voltage
drift translates into 0.05ppm/°C gain error. Simpler alternatives, with the amplifiers providing gain using resistor
arrays for feedback, can produce results that are similar to
bridge sensing schemes via attenuators. Note that the
amplifiers must have high open-loop gain or gain error will
be a source of error. The fact that input offset voltage has
relatively little effect on overall error may lead one to use
low performance amplifiers for this application. Note that
the gain of a device such as an LF156, (25V/mV over
temperature) will produce a worst-case error of –180ppm
at a noise gain of 3, such as would be encountered in an
inverting gain of 2, to produce –10V from a 5V reference.
350Ω
BRIDGE
+
5V
0.1µV
7
3
+
LTC1050S8
2
–
1µF
AV = 9.98 1 +
R1
4.98k
()
4.99k + 175Ω
Figure 48. Bridge Amplification Using a Single Amplifier
4
46.4k
R2
46.4k
175Ω
6
+
1µF
20k
20k
10µF
3
4
5
6
+
REF
REF
IN
IN
V
CC
+
–
+
LTC2410
–
GND
5V
0.1µF
2
1, 7, 8, 9,
10, 15, 16
2410 F48
35
LTC2410
U
WUU
APPLICATIOS IFORATIO
The error associated with the 10V excitation would be
–80ppm. Hence, overall reference error could be as high
as 130ppm, the average of the two.
Figure 52 shows a similar scheme to provide excitation
using resistor arrays to produce precise gain. The circuit
is configured to provide 10V and –5V excitation to the
bridge, producing a common mode voltage at the input to
the LTC2410 of 2.5V, maximizing the AC input range for
applications where induced 60Hz could reach amplitudes
up to 2V
The last two example circuits could be used where multiple bridge circuits are involved and bridge output can be
multiplexed onto a single LTC2410, via an inexpensive
multiplexer such as the 74HC4052.
RMS
.
R1
25.5k
0.1%
Figure 53 shows the use of an LTC2410 with a differential
multiplexer. This is an inexpensive multiplexer that will
contribute some error due to leakage if used directly with
the output from the bridge, or if resistors are inserted as
a protection mechanism from overvoltage. Although the
bridge output may be within the input range of the A/D and
multiplexer in normal operation, some thought should be
given to fault conditions that could result in full excitation
voltage at the inputs to the multiplexer or ADC. The use of
amplification prior to the multiplexer will largely eliminate
errors associated with channel leakage developing error
voltages in the source impedance.
V
S
2.7V TO 5.5V
2
V
CC
3
+
REF
4
–
REF
LTC2410
PLATINUM
100Ω
RTD
5
+
PLATINUM
100Ω
RTD
IN
6
–
IN
GND
Figure 49. Remote Half Bridge Interface
5V
+
LTC1050
–
0.1%
R1
10k, 5%
10k
R2
R3
1µF
10k
5%
1, 7, 8, 9,
10, 15, 16
2410 F49
560Ω
10k
10k
5V
2
V
CC
3
+
REF
4
–
REF
LTC2410
5
+
IN
6
–
IN
GND
1, 7, 8, 9,
10, 15, 16
2410 F50
36
Figure 50. Remote Half Bridge Sensing with Noise Suppression on Reference
LTC2410
U
WUU
APPLICATIOS IFORATIO
15V15V
7
38
LTC1150
4
–15V
15V
7
LTC1150
4
–15V
+
2
–
3
+
2
–
350Ω
BRIDGE
2N3904
33Ω
Q1
–15V
20Ω
33Ω
Q2
2N3904
20Ω
6
0.1µF
1k
6
15V
U1
LTC1043
200Ω
10V5V
1µF
11
12
14
5
15
4
7
*
13
17
U2
LTC1043
6
*
2
3
18
LT1236-5
++
47µF0.1µF
+
10µF
0.1µF
3
4
5
6
REF
REF
IN
IN
+
–
2
V
CC
LTC2410
+
–
GND
1, 7, 8, 9,
10, 15, 16
10V
5V
0.1µF
1k
5V
4
7
*
11
12
13
17
–10V
1µF
FILM
–10V
200Ω
U2
LTC1043
8
14
*FLYING CAPACITORS ARE
1µF FILM (MKP OR EQUIVALENT)
SEE LTC1043 DATA SHEET FOR
DETAILS ON UNUSED HALF OF U1
Figure 51. LTC1043 Provides Precise 3X Reference for Excitation Voltages
2410 F51
37
LTC2410
U
WUU
APPLICATIOS IFORATIO
15V
–5V
20Ω
RN1
10k
RN1
10k
20Ω
350Ω BRIDGE
TWO ELEMENTS
VARYING
2N3904
22Ω
10V
33Ω
×2
Q2, Q3
2N3904
Q1
×2
–15V
3
+
C1
0.1µF
C2
0.1µF
1/2
LT1112
15V
8
1/2
LT1112
4
–15V
2
–
3
RN1
10k
4
6
–
5
+
1
21
65
7
5V
LT1236-5
+
C3
47µF
2
V
CC
LTC2410
3
+
REF
4
–
REF
5
+
IN
6
–
8
RN1
10k
7
RN1 IS CADDOCK T914 10K-010-02
IN
GND
1, 7, 8, 9,
10, 15, 16
C1
0.1µF
2410 F52
5V
Figure 52. Use Resistor Arrays to Provide Precise Matching in Excitation Amplifier
TO OTHER
DEVICES
5V
12
14
15
11
16
74HC4052
1
5
2
4
98
+
47µF
13
3
6
10
REF
REF
IN
IN
+
–
5V
V
CC
+
–
LTC2410
GND
1, 7, 8
2410 F53
A0
A1
Figure 53. Use a Differential Multiplexer to Expand Channel Capability
38
U
TYPICAL APPLICATIOS
LTC2410
Sample Driver for LTC2410 SPI Interface
The LTC2410 has a very simple serial interface that makes
interfacing to microprocessors and microcontrollers very
easy.
The listing in Figure 55 is a simple assembler routine for
the 68HC11 microcontroller. It uses PORT D, configuring
it for SPI data transfer between the controller and the
LTC2410. Figure 54 shows the simple 3-wire SPI
connection.
The code begins by declaring variables and allocating four
memory locations to store the 32-bit conversion result.
This is followed by initializing PORT D’s SPI configuration.
The program then enters the main sequence. It activates
the LTC2410’s serial interface by setting the SS output
low, sending a logic low to CS. It next waits in a loop for
a logic low on the data line, signifying end-of-conversion.
After the loop is satisfied, four SPI transfers are completed, retrieving the conversion. The main sequence ends
by setting SS high. This places the LTC2410’s serial
interface in a high impedance state and initiates another
conversion.
The performance of the LTC2410 can be verified using the
demonstration board DC291A, see Figure 56 for the
schematic. This circuit uses the computer’s serial port to
generate power and the SPI digital signals necessary for
starting a conversion and reading the result. It includes a
Labview application software program (see Figure 57)
which graphically captures the conversion results. It can
be used to determine noise performance, stability and
with an external source, linearity. As exemplified in the
schematic, the LTC2410 is extremely easy to use. This
demonstration board and associated software is available
by contacting Linear Technology.
13
SCK
LTC2410
Figure 54. Connecting the LTC2410 to a 68HC11 MCU Using the SPI Serial Interface
SDO
12
11
CS
68HC11
SCK (PD4)
MISO (PD2)
SS (PD5)
2410 F54
39
LTC2410
U
TYPICAL APPLICATIOS
*****************************************************
* This example program transfers the LTC2410's 32-bit output *
* conversion result into four consecutive 8-bit memory locations. *
*****************************************************
*68HC11 register definition
PORTDEQU$1008Port D data register
*" – , – , SS* ,CSK ;MOSI,MISO,TxD ,RxD"
DDRDEQU$1009Port D data direction register
SPSREQU$1028SPI control register
*"SPIE,SPE ,DWOM,MSTR;SPOL,CPHA,SPR1,SPR0"
SPSREQU$1029SPI status register
*"SPIF,WCOL, – ,MODF; – , – , – , – "
SPDREQU$102ASPI data register; Read-Buffer; Write-Shifter
*
* RAM variables to hold the LTC2410's 32 conversion result
*
DIN1EQU$00This memory location holds the LTC2410's bits 31 - 24
DIN2EQU$01This memory location holds the LTC2410's bits 23 - 16
DIN3EQU$02This memory location holds the LTC2410's bits 15 - 08
DIN4EQU$03This memory location holds the LTC2410's bits 07 - 00
*
**********************
* Start GETDATA Routine *
**********************
*
ORG$C000Program start location
INIT1LDS#$CFFFTop of C page RAM, beginning location of stack
LDAA#$2F–,–,1,0;1,1,1,1
*–, –, SS*-Hi, SCK-Lo, MOSI-Hi, MISO-Hi, X, X
STAAPORTDKeeps SS* a logic high when DDRD, bit 5 is set
LDAA#$38–,–,1,1;1,0,0,0
STAADDRDSS*, SCK, MOSI are configured as Outputs
*MISO, TxD, RxD are configured as Inputs
*DDRD's bit 5 is a 1 so that port D's SS* pin is a general output
LDAA#$50
STAASPCRThe SPI is configured as Master, CPHA = 0, CPOL = 0
*and the clock rate is E/2
*(This assumes an E-Clock frequency of 4MHz. For higher E*Clock frequencies, change the above value of $50 to a value
*that ensures the SCK frequency is 2MHz or less.)
GETDATA PSHX
PSHY
PSHA
LDX#$0The X register is used as a pointer to the memory locations
*that hold the conversion data
LDY#$1000
BCLRPORTD, Y %00100000This sets the SS* output bit to a logic
*low, selecting the LTC2410
*
40
U
TYPICAL APPLICATIOS
**********************************
* The next short loop waits for the *
* LTC2410's conversion to finish before *
* starting the SPI data transfer *
**********************************
*
CONVEND LDAAPORTDRetrieve the contents of port D
ANDA#%00000100Look at bit 2
*Bit 2 = Hi; the LTC2410's conversion is not
*complete
*Bit 2 = Lo; the LTC2410's conversion is complete
BNECONVENDBranch to the loop's beginning while bit 2 remains
high
*
*
********************
* The SPI data transfer *
********************
*
TRFLP1 LDAA#$0Load accumulator A with a null byte for SPI transfer
STAASPDRThis writes the byte in the SPI data register and starts
*the transfer
WAIT1LDAASPSRThis loop waits for the SPI to complete a serial
transfer/exchange by reading the SPI Status Register
BPLWAIT1The SPIF (SPI transfer complete flag) bit is the SPSR's MSB
*and is set to one at the end of an SPI transfer. The branch
*will occur while SPIF is a zero.
LDAASPDRLoad accumulator A with the current byte of LTC2410 data
that was just received
STAA0,XTransfer the LTC2410's data to memory
INXIncrement the pointer
CPX#DIN4+1 Has the last byte been transferred/exchanged?
BNETRFLP1If the last byte has not been reached, then proceed to the
*next byte for transfer/exchange
BSETPORTD,Y %00100000 This sets the SS* output bit to a logic high,
*de-selecting the LTC2410
PULARestore the A register
PULYRestore the Y register
PULXRestore the X register
RTS
LTC2410
Figure 55. This is an Example of 68HC11 Code That Captures the LTC2410’s
Conversion Results Over the SPI Serial Interface Shown in Figure 54
41
LTC2410
TYPICAL APPLICATIOS
V
CC
BANANA JACK
1
J4
V
EXT
BANANA JACK
1
J6
+
REF
BANANA JACK
1
J7
–
REF
BANANA JACK
1
J8
V
+
IN
BANANA JACK
1
J9
V
–
IN
BANANA JACK
1
J10
GND
JP1
JUMPER
2
31
+
R2
3Ω
JP3
JUMPER
31
2
J3
V
CC
J5
GND
1
JP5
JUMPER
2
U1
LT1460ACN8-2.5
62
V
OUTVIN
GND
C1
10µF
35V
4
1
1
NOTES:
INSTALL JUMBER JP1 AT PIN 1 AND PIN 2
INSTALL JUMBER JP2 AT PIN 1 AND PIN 2
INSTALL JUMBER JP3 AT PIN 1 AND PIN 2
U
C6
0.1µF
V
CC
+
C2
22µF
25V
V
CC
+
C5
10µF
35V
+
REF
–
REF
+SDO
V
IN
–GND
V
IN
LTC2410CGN
CC
U4
JP2
JUMPER
GNDGND GND GND
21
CSV
SCK
GND
GND
112
F
O
9871
+
JP4
JUMPER
2
143
134
125
166
15
10
U2
LT1236ACN8-5
62
V
OUTVIN
31
C3
10µF
35V
GND
U3E
74HC14
U3B
74HC14
U3C
74HC14
+
4
1110
34
65
R7
22k
R8
51k
BYPASS CAP
9
C4
100µF
16V
U3F
74HC14
U3A
74HC14
U3D
74HC14
FOR U3
R1
10Ω
1312
12
8
3
V
CC
D1
BAV74LT1
R3
51k
R4
51k
R5
49.9Ω
1
C7
0.1µF
21
1
3
2
J1
V
1
J2
GND
1
6
2
7
3
8
4
9
5
R6
3k
Q1
MMBT3904LT1
EXT
P1
DB9
2410 F56
Figure 56. 24-Bit A/D Demo Board Schematic
Figure 57. Display Graphic
42
PACKAGE DESCRIPTIO
0.015
± 0.004
(0.38 ± 0.10)
0.007 – 0.0098
(0.178 – 0.249)
0.016 – 0.050
(0.406 – 1.270)
* DIMENSION DOES NOT INCLUDE MOLD FLASH. MOLD FLASH
SHALL NOT EXCEED 0.006" (0.152mm) PER SIDE
** DIMENSION DOES NOT INCLUDE INTERLEAD FLASH. INTERLEAD
FLASH SHALL NOT EXCEED 0.010" (0.254mm) PER SIDE
0° – 8° TYP
UW
PCB LAYOUT A D FIL
U
Dimensions in inches (millimeters) unless otherwise noted.