Quad 16-Bit Voltage-/Current-Output DACs Save Space, Cost, and Power in...

Quad 16-Bit Voltage-/CurrentOutput DACs Save Space, Cost,
and Power in Multichannel PLCs
By Colm Slattery
Overview
Programmable logic controllers (PLCs) use fast, deterministic
functions, such as logic, sequencing, timing, counting, and
arithmetic algorithms, to control machines and processes. They
use analog and digital signals to communicate with end nodes
(reading sensors and controlling actuators, for example). Typical
methods of communication include current/voltage loops,
Fieldbus,1 and industrial Ethernet2 protocols.
The industry has a continuing tendency to increase the number
of sensor and control nodes in the remote area, causing a
corresponding increase in the number of I/O module nodes in the
controller—and some distributed control systems (DCS) can handle
thousands of nodes. This concentration of nodes brings increased
temperature-related challenges, especially for systems that
implement the 4-mA to 20-mA loop communications standard.
Perhaps the biggest and most relevant challenge to the system
designer is the need for greater efficiency and reduced power
consumption, as the inefficiency of existing systems results in
wasted power and increased operating costs. This article explains
the challenges of designing such systems for greater efficiency and
introduces the AD5755, a versatile, 4-channel, 16-bit digital-toanalog converter (DAC) as a more integrated solution to help
resolve these issues.
Some of the advantages of PROFINET are
•Increased speed, up from 9.6 kbps with RS-232 to 1 Gbps.
•Improved overall performance.
•Increased distance.
•Ability to use standard access points, routers, switches, hubs,
cables, and optical fiber—which are immensely cheaper than
the equivalent serial-port devices.
•Ability to have more than two nodes on link. This was possible
with RS-485, but not with RS-232.
At the field level, field bus protocols, used to interconnect industrial
drives, motors, actuators, and controllers to the PLC/DCS I/O
systems, are numerous, including DeviceNet,™ CAN,6 and InterBus,®7
as well as the above-mentioned PROFIBUS and Fieldbus.
An input-output (I/O) controller connects to sensors and control
actuators in a factory- or process environment; it communicates
with multiple end nodes by analog and digital means, as noted
above. Intrinsically safe systems connect via 4-mA to 20-mA
current loops, and some use isolation. The control processor is
typically an 8-bit to 32-bit processor with performance of up to
100 DMIPS (Dhrystone millions of instructions per second).
Factory automation equipment is ruggedly constructed for fanless
operation in a harsh industrial environment.
Examples of 8-channel analog I/O modules are featured in
Figure 2. Because of their small size, they have limited powerdissipation capability, some even less than 5 W.
System
The levels of communication in a typical industrial control system
are shown in Figure 1. Until recently, the distributed input/output
(remote I/O and PLC) would typically be connected using such open
or proprietary protocols as Modbus,3 PROFIBUS4 (process field
bus), or Fieldbus. Nowadays, there is an increasing interest in using
PROFINET,5 a form of industrial Ethernet protocol that is designed
for the fast exchange of data between Ethernet-based devices.
ENGINEERING
STATION
OPERATIONS MANAGEMENT LEVEL
CONTROL LEVEL
Figure 2. I/O modules.
Analog 4-mA to 20-mA current loops are commonly used for
signaling in industrial process control, with 4 mA representing the
low end of the range and 20 mA the high end. The key advantages
of the current loop are that accuracy of the signal is not affected by
voltage drops in the interconnecting wiring and the loop can supply
up to 4 mA for powering the device. Even if the line has significant
electrical resistance, the current loop transmitter will maintain the
proper current, up to its maximum voltage capability.
PROFINET
INDUSTRIAL ETHERNET
HMI
PROFIBUS, MODBUS,
INTERBUS, ETC.
CONTROLLER
FIELD LEVEL
DRIVES
DISTRIBUTED
I/O
TO SENSOR/ACTUATOR
4mA TO 20mA, ð10V, RS-485
Figure 1. Hierarchy in a control system.
Analog Dialogue 44-07, July (2010)
The live-zero represented by 4 mA allows the receiving instrument
to detect some failures of the loop (for example, 0 mA indicates an
open loop, or 3 mA could indicate a fault condition on the sensor)
and also allows 2-wire-transmitter devices to be powered by the
loop current. Such instruments are used to measure pressure,
temperature, flow, pH, and other process variables and to control a
valve positioner or other output actuator. The current in an analog
current loop can be converted to a voltage input at any point in
the loop with a series precision resistor. Since input terminals
of instruments may have one side of the current loop tied to the
chassis ground (earth), analog isolators may be required when
connecting several instruments in series.
www.analog.com/analogdialogue
1
Power Dissipation Concerns
Figure 3 shows a system in which one channel is configured for
4-mA to 20-mA communications (in this case to drive an actuator
load from a DAC). The termination resistance of the actuator
determines the maximum compliance voltage needed across the
loop. For example, a 100-Ω resistance would require at least 2 V
at 20 mA. It is very common that today’s systems must be able to
drive loads of up to (and sometimes exceeding) 1 kΩ. With this
load resistance, and a full-scale current of 20 mA, the supply would
need to furnish at least 20 V. The power generated would be
P = V × I = 20 V × 0.02 A = 0.4 W.
If the load resistance was changed to 100 Ω, using the same supply
(a valid condition), the power dissipated would still be 0.4 W, even
though only 0.04 W is needed. In this case there is a 90% loss of
efficiency in the system, with 360 mW being wasted.
20V
DAC
20V
20mA
DAC
1k±
POWER GENERATED = 0.4W
POWER NEEDED
= 0.4W
POWER WASTED
= 0W
20mA
100±
POWER GENERATED = 0.4W
POWER NEEDED
= 0.04W
POWER WASTED
= 0.36W
Figure 3. Power is wasted when full-scale output is much
less than the power-supply voltage.
With an 8-channel module, the total power dissipation with a
20 V supply would be 3.2 W, of which as much as 2.88 W would
be wasted in the module (if all loads are 100 Ω). In such cases,
self heating, as well as the effect of the increased power budget,
starts to become a consideration. Increased temperatures within
the module can lead to increased system errors—the drift specs
of the individual components need to be factored into the overall
system error budget.
Designers may consider various ways to solve these problems:
•Increasing the module size allows more power dissipation, but the
added cost makes this solution less competitive.
•Heat sinking and/or fan control can be used—an expensive
solution that also increases space. Indeed, in some safetycritical applications, such temperature control devices are
not allowed.
•The maximum load resistance can be reduced to limit the overall
power dissipation in the circuit. This is a performance-limiting
factor in some applications and is noncompetitive from a system
marketing point of view.
In any event, the trend to provide an increased number of channels
in a smaller space will cause further thermal power problems for
many system designers.
One way to help solve this problem is to start with a 5-V supply.
Monitor the output load voltage, then efficiently boost and
regulate the output voltage as needed. The 5-V supply and an
efficient dc-to-dc boost converter use feedback control to provide
the appropriate output voltage, minimizing the on-chip power
dissipation (Figure 4).
5V
DYNAMIC SUPPLY
CONTROL
DC-DC
DAC
OUTPUT
SENSE
20mA
100±
Figure 4. Dynamic supply control principle.
2
This kind of closed-loop dynamic power capability can be found
in the AD5755 family of 4-channel, 16-bit, serial-input, voltageand current-output DACs (see Appendix—Figure A). Because
each of its four channels can individually furnish either current
or voltage with 16-bit resolution, with output powered by an
individual dc-to-dc converter under dynamic power control, the
device provides the equivalent of four low-dissipation nodes in a
very compact 9-mm × 9-mm × 0.8-mm package.
The simplified circuit of Figure 5 shows how the dynamic power
control works, using an inductive boost circuit. Each channel is
capable of providing a boosted output voltage greater than 30 V.
The dynamic power control mechanism uses feedback to regulate
the output voltage, which is divided down by a resistive voltage
divider and compared to the reference voltage in an internal
error amplifier to create an error current. At the beginning of
the switching cycle, the MOSFET switch is turned on and the
inductor current ramps up. The MOSFET current, converted to
a voltage, is measured. When the current-sense voltage is greater
than the error voltage, the MOSFET is turned off and the inductor
current ramps down until the internal clock initiates the next
switching cycle. A similar scheme is used to regulate the output
compliance voltage in current mode. In this case a feedback error
current is used.
5V
10¿H
4.7¿H
16-BIT
DAC
CONTROL
4mA
TO 20mA
RLOAD
OUTPUT SENSE
Figure 5. Voltage boost with power control.
The user has the option to switch the frequency and phase of each
channel’s dc-to-dc converter switching signals to allow for circuit
and component optimization.
Programmable Switching
Frequency:
Programmable DC-to-DC
Clock Phase:
Offers the ability to
change dc-to-dc switching
frequencies, allowing for
system optimization and
more flexible choices of
external components
Offers the ability to change
the phase of the clock edges
of individual dc-to-dc
blocks, allowing for system
optimization
•333 kHz
•400 kHz
•500 kHz
•667 kHz
•All four channels clock on
same edge
•ChanA and ChanB clock
on one edge, ChanC and
ChanD clock on opposite
edge
•ChanA and ChanC clock
on one edge, ChanB and
ChanD on opposite edge
•ChanA, ChanB, ChanC,
and ChanD clock 90° out
of phase from each other
(0°, 90°, 180°, and 270°)
The dynamic power control on the output driver is designed to
minimize package power dissipation. Typical ICs can operate at
internal junction temperatures (T JMAX) up to 125°C. Assume the
ambient temperature, TA, in the system is 85°C. The thermal
impedance, θJA , for the LFCSP package is typically 28°C/W.
Analog Dialogue 44-07, July (2010)
To calculate the allowable on-chip dissipation we can use the
following analysis.
MCU
ALERT
CLEAR
Without dynamic power control, assuming a 24-V supply, the
worst-case power dissipation (per channel) can be calculated
to be
Power Dissipation = Supply Voltage × Max Current
= 24 V × 20 mA
= 0.48 W
Four channels would dissipate nearly 2 W under similar
conditions; this would cause problems for both the module and
the semiconductor circuitry. By enabling the dynamic power
feature, the AD5755 regulates the supply to minimize the
on-chip power dissipation. Figure 6 shows a comparison of the
power dissipated per channel with dynamic power enabled and
disabled (fixed supply).
450
400
POWER (mW)
350
DPC DISABLED
300
4mA
TO 20mA
AD5755
Figure 7. Watchdog timer flags loss of control signal and
returns DAC to clear setting.
Even with the MCU operating normally, communications
sig nals can become cor r upted in noisy industr ial
environments. For dealing with this possibility, the AD5755
has a n opt iona l packet er ror - checking ( PEC) f u nc t ion,
which implements a CRC8 polynomial routine. This can
be enabled or disabled through software to ensure that the
output is never incorrectly updated.
Miswiring on the output can often lead to open- or short-circuit
connections, potentially damaging the system. (Even if no
damage occurs, the problem can often be difficult to diagnose.
The AD5755 has open- and short-circuit detection, immediately
setting a fault flag to alert a technician of the problem). In
addition, short-circuit protection limits the output current in
the event of a short circuit. All faults can be communicated via
the SPI interface or through a hardware fault pin, allowing the
user to take immediate action.
Flexible Output-Range Programmability
250
200
150
100
DPC ENABLED
50
0
WATCHDOG
TIMER
0
5
10
15
20
25
30
PROGRAMMED CURRENT (4mA TO 24mA)
Figure 6. Dissipation comparison with and without dynamic
power control.
When the dynamic power capability is enabled, the on-chip
power dissipation is about 50 mW with output current of
24 mA vs. 400 mW with no regulation. This ability to control the
on-chip power dissipation is of great value to the system designer
because the number of channels in the system can be increased
while minimizing module dissipation. It thus eliminates the
need to consider extensive (and expensive) methods to control
system temperatures.
System Error Checking and Diagnostics Under Fault Conditions:
For industrial applications, it is important to be able to monitor
and report system-level faults and critical to have as much control
as possible over the system under a fault condition. The AD5755
includes many on-chip diagnostic features that provide the user
with system-level error checking.
One serious consideration is where the MCU/DSP that controls
the DAC goes when a fault condition occurs. With no ability to
control the output, the user would lose complete control of the
system. The AD5755 has a watchdog timer (with programmable
timeouts) that sets an alert flag (active high) if it has not received
a command over the SPI interface within the timeout period. If
desired, this alert pin can be directly connected to the CLEAR
pin (also active high) to set the outputs into a known safe
condition (Figure 7). Each channel on the AD5755 has a 16-bit
programmable clear code register, giving the user flexibility to
clear the output to any code.
Analog Dialogue 44-07, July (2010)
To deal with the required variety of voltages and currents, the
AD5755 has many programmable ranges available for each
channel, including: 4 mA to 20 mA, 0 mA to 24 mA, 0 mA to
20 mA, 0 V to 5 V, 0 V to 10 V, ±5 V, ±10 V, and ±12 V. It
is also possible for the user to digitally program the gain and
offset of each range on individual channels. These gain- and
offset registers have 16-bit resolution. For example, to set a
0-V to 10.5-V output range (as in Figure 8), first select the 0-V
to 12-V range, then program the gain code to trim the gain to 10.5 V.
Once the gain trim is complete, the output range will be 0 V to
10.5 V, with 16-bit resolution. The offset can be programmed
in a similar manner.
SELECT INITIAL RANGE
12V
FFFF
PROGRAM GAIN CODE
10.5V
16 BITS OF
RESOLUTION OVER
0V TO 10.5V RANGE
NEW
OUTPUT
RANGE
OUTPUT
RANGE
0V
0000
FFFF
0V
0000
Figure 8. Arbitrary range scaling.
Communicating Additional Information Over the 4-mA to 20-mA
Current Loop
The disadvantage of a pure 4-mA to 20-mA current loop is
its unidirectional communication of a single process variable,
an annoying limitation in modern industrial control systems.
The development of the highway-addressable remote transducer
(HART) standard opened up new possibilities for 4-mA to 20-mA
communication lines.
HART provides for a digital two-way communication scheme
that is compatible with 4-mA to 20 mA current loops. A 1-mA
peak-to-peak frequency-shift-keyed (FSK) signal is superimposed
on the 4-mA to 20-mA analog current signal. The two frequencies
used are 1200 Hz (Logic 1) and 2200 Hz (Logic 0), based on the
BELL 202 communications standard (Figure 9).
3
AD5755 Complete Solution
20mA
Figure 11 shows a typical setup using the AD5755. (One HART
modem channel is shown in the diagram, but four HART inputs
are available—one per channel). When enabled, the dynamic
power control feature requires four external components per
channel: an inductor with a saturation current of the order
of 1 A, a switching diode, and two capacitors having low
equivalent series resistance (ESR). With a minimal number of
external components, the AD5755 provides an integrated
high-performance system capability on a single chip. The total
unadjusted error (TUE), including all gain and offset errors at
25°C, is typically 0.01%.
DIGITAL SIGNAL
“1”
4mA
“0”
“0”
“1”
“1”
“0”
“1”
“0”
“1”
ANALOG SIGNAL
NOTE: NOT TO SCALE
TIME
Figure 9. HART signal riding on an increasing loop current.
The AD5755 can be configured to transmit a HART signal
with only two external components. The output of the HART
modem is attenuated and ac-coupled at the CHART pin of the
AD5755; this results in the modem output being modulated on
top of the 4-mA to 20-mA analog current without affecting the
“dc” level of the current. The circuit in Figure 10 shows how
the AD5755 can interface to a HART modem to embody this
dual form of communication.
500mV p-p
HART
MODEM
150mV p-p
C1
AD5755
4mA
TO 20mA
CHART
C2
Figure 10. The AD5755 in HART communication.
The HART specification requires that the maximum rate of change
of analog current not interfere with HART communications.
Obviously, step changes in the current output can disrupt HART
signaling. Fortunately, the AD5755 has controllable slew rate,
which, when enabled, allows the user to digitally limit the slew
rate of the current output.
+15V
SDI
SYNC
SDO
WATCHDOG
TIMER
PROGRAMMABLE
RANGES
(ON EVERY PIN)
DC-DC
CONTROL
HIGH
VOLTAGE
CONDITIONING
5V 16-BIT
DAC
US ERR OFFSET
TRIM
ALERT
As a 4-channel device in a 9-mm × 9-mm CSP package, the
AD5755 dramatically helps reduce board area while increasing
channel density. With dynamic power control, the on-chip
power dissipation is regulated and module power dissipation
is minimized. The addition of on-chip diagnostics, including
watchdog timers, PEC error checking, and open-/shortcircuit detection and protection gives the end user higher
confidence that the robust design is capable of working in
harsh industrial environments. The AD5755 is a true systemon-a-chip solution.
+5V
USER GAIN
TRIM
DIGITAL
CONTROL
As both the required number of channels and the density of
channels per module increase, a number of problems present
themselves to systems designers: How can I increase the number
of channels while keeping the form factor of the module small?
How can I increase the number of channels and design an energyefficient system, while minimizing self-heating effects and drift
errors within the system? How can I offer the most flexibility to
my customer in terms of programmability of the outputs? What
safety features and diagnostics can I provide to ensure robust
systems in which problems can be easily tracked?
–15V
AD5755
SCLK
Conclusion
0V TO 5V
IOUT/
VOUT
0V TO 10V
ð5V
CHANNEL A
ð10V
0V TO 6V
SDO
CHANNEL B
AD1
AD0
IOUT/
VOUT
0V TO 12V
ð6V
DEVICE ADDRESS PINS
CHANNEL C
REFERENCE
CHANNEL D
IOUT/
VOUT
ð12V
4mA TO 20mA
0mA TO 20mA
IOUT/
VOUT
0mA TO 24mA
HART
MODEM
Figure 11. AD5755 setup.
4
Analog Dialogue 44-07, July (2010)
References
Appendix
More About the AD5755 Quad DAC
1
The AD57558 quad voltage- and current-output DACs operate
with a –26-V to +33-V power supply. On-chip dynamic power
control minimizes package power dissipation in current mode
by regulating the voltage on the output driver between 7 V
and 30 V.
The AD5755 uses a versatile 3-wire serial interface that operates
at clock rates up to 30 MHz and is compatible with standard
SPI,® QSPI,™ MICROWIRE,™ DSP, and microcontroller interface
standards. The interface also features optional CRC-8 packet error
checking, as well as a watchdog timer that monitors activity on
the interface.
The AD5755 features 16-bit resolution and monotonicity,
voltage or current output on the same pin, user-programmable
offset and gain, on-chip diagnostics, an on-chip 5 ppm/°C
max voltage reference, and a – 40°C to +105°C operating
temperature range. Available current-output ranges are 0 mA to
20 mA, 4 mA to 20 mA, and 0 mA to 24 mA ± 0.05%; available
voltage ranges are 0 V to 5 V, 0 V to 10 V, ±5 V, ±10 V, ±6 V,
and ±12 V ± 0.05%.
www.fieldbus.org.
www.cisco.com/web/strategy/docs/manufacturing/industrial_
ethernet.pdf.
3
www.lammertbies.nl/comm/info/modbus.html.
4
www.profibus.com.
5
www.anybus.com/technologies/profinet.shtml.
6
www.interfacebus.com/Design_Connector_CAN.html.
7
www.interfacebus.com/INTERBUS_Field_Bus_Description.html.
8
www.analog.com/pr/AD5755.
2
Author
Colm Slattery [[email protected]]
graduated in 1995 from the University of Limerick,
Ireland, with a bachelor’s degree in electronic
engineering. After working in test-development
engineering at Microsemi, he joined ADI in 1998.
He spent three years in an applications role in
Shanghai and is currently working as a system
applications engineer for the Industrial and
Instrumentation segment.
AVCC
5.0V
AVSS
AVDD
–15V AGND +15V
SW A
AD5755
–5V
DVDD
DGND
LDAC
CLEAR
SCLK
SDI
SYNC
SDO
POWER
CONTROL
INPUT
SHIFT
REGISTER
AND
CONTROL
16
INPUT
REG A
DAC 16
REG A
DC-DC
0V TO 24V
VREG
VSEN1
R2
DAC A
VBOOST A
VSEN2
R3
GAIN REG A
OFFSET REG A
IOUT A
FAULT
STATUS
REGISTER
ALERT
WATCHDOG
TIMER
(SPI ACTIVITY)
REFOUT
STATUS
REGISTER
REFIN
REF
BUFFERS
RSET A
R1
+VSENSE A
VOUT
RANGE
SCALING
VOUT A
DAC CHANNEL A
–VSENSE A
IOUT B, C, D
DAC CHANNEL B
RSET B, C, D
+VSENSE B, C, D
DAC CHANNEL C
AD1
AD0
VOUT B, C, D
–VSENSE B, C, D
DAC CHANNEL D
SW B, C, D
BOOST B, C, D
Figure A. Functional block diagram of the AD5755 quad DAC. All four channels are identical.
Analog Dialogue 44-07, July (2010)
5