power developer: july 2014
DESCRIPTION
Linear Technology's Products Drive Industry Interview with Lothar Maier, CEO of Linear TechnologyTRANSCRIPT
Linear Technology's
Products Drive Industry
Interview with Lothar Maier CEO of Linear Technology
Ultralow-Power Embedded Applications
Safe Power Supply in Medical Apps Worldwide
July 2014
POWERDEVELOPER Read Power Developer, the monthly newsletter for Engineers:
http://www.embeddeddeveloper.com/news_letter/
All the forces in the world are not as powerful as an idea whose time has come.—Victor Hugo, 1800
Power Developer contains new ideas that come every month.—Power Developer Editors, 2013
Power Developer
Modern input filter components in surface
mount technology have better performance
than through-hole parts. However, this
improvement is outpaced by the increase in
operating switching frequencies of switching
regulators. Higher efficiency, low minimum on-
off times result in higher harmonic content due
to the faster switch transitions.
For every doubling in switching frequency,
the EMI becomes 6dB worse when all other
parameters, such as switch capacity and
transition times, remain constant. The wideband
EMI behaves like a first-order high pass with
20dB higher emissions if the switching frequency
increases by 10x.
Savvy printed circuit board (PCB) designers will
make the hot loops small and use shielding
ground (GND) layers as close to the active layer
as possible; nevertheless pinout, package
construction, thermal design requirements,
and package sizes needed for adequate energy
storage in decoupling components dictates a
certain minimum hot-loop size.
To make layout even more challenging, on a
typical planar printed circuit board the magnetic
or transformer-style coupling between traces
above 30MHz will diminish all filter efforts since
the higher the harmonic frequencies are, the more
effective unwanted magnetic coupling becomes.
The tried and true solution is to use a shielding
box for the complete circuit. Of course, this adds
costs, increases required board space, makes
thermal management and testing more difficult,
and introduces additional assembly costs.
Another frequently used method is to slow down
the switching edges. This has the undesired
effect of reducing the efficiency, increasing
minimum on-off times as well as the required
dead times, and compromising the potential
current-control loop speed.
With Linear’s new LT8614 Silent Switcher™
regulator, you have the effect of a shielded box
without using a shield and also eliminate the
above-mentioned drawbacks (figure 1).
Figure 1. The LT8614 Silent Switcher minimizes EMI/EMC emissions
while delivering high efficiency at frequencies up to 3MHz.
Figure 2. Blue trace is the noise floor; red trace is the LT8614 board
at CISPR25 radiated measurement in an anechoic chamber.
Figure 3. Blue trace (ch. 2) is the LT8614; purple trace (ch. 1) is the
LT8610; both 13.5Vin, 3.3V out at 2.2A load.
“Higher efficiency,
low minimum on-off
times result in higher
harmonic content due
to the faster switch
transitions.” 4
12
20
26
Linear Technology is known to have the highest operating margin in the industry. How did the company get to this point?
There is no single factor that leads to the margins that we see here at Linear. However, the most important element, from a product standpoint, is that we make new, unique products. We do not try to copy anything from our competitors or second-source another competitor’s products. We come up with products that are not only technically interesting, but also compelling in terms of bringing value to the customer. This may sound easy, but is a rather tall order because it typically takes two to three years for us to develop a product. Then it takes another few years for the customer to take our product and design it into their product. We are always making decisions to develop products that customers are going to buy and find valuable five years in the future—this is tough to do, but we do it.
Another key factor that contributes to our high operating margin is product pricing. We price each product based on the value it brings to the customers, not the transistor count. Also, we recognize that we not only have to develop superior products, but we have to actually deliver them. Linear has a great industry reputation for providing products on time and with short lead times. The combination of all of these aspects results in Linear having optimal operating margins.
Which markets do you feel are the most important to target?
The best markets, historically speaking, have changed. Fifteen years ago, Linear was very concentrated in the communications market and then came the “dot-com bubble” which made that market less important. We used to have a very strong presence in consumer products with high-end analog components that were sold into digital cameras and high-functioning cellular phones. This was a significant part of our business. Around 2007, it became obvious that this consumer business, which we thought was high-performance analog, had deteriorated to just plain consumer grade. That is when we made the decision to move away from those markets and find new ones. At that time, we refocused the company on the automotive, communications, and industrial markets. That brings us to where we are today. Seven years ago, it would have seemed unwise to become so involved in the automotive and industrial markets, but these markets today have become some of the fastest-growing segments of the analog market. We believe we are in the right markets, at the right time.
“We are always making decisions to develop products that customers are going to buy and find valuable five years in the future—this is tough to do, but we do it.”
Safe Power Supply
By Thomas RechlinSenior FAE for Europe RECOM Engineering, Gmunden, Austria
Only about 100 years ago, doctors had to rely on simple equipment such as microscopes and stethoscopes to make a diagnosis. Surgical procedures always posed a huge risk
since the equipment was somewhat rudimentary. The life and well-being of patients therefore primarily depended on the skills and experience of the surgeon. Today, hospitals are full of high-tech equipment and computerised devices that allow for much more detailed and early diagnoses, as well as keyhole surgeries. These new technologies help minimize the risk to patients and ensure fast recovery. The technical progress made in the field of medicine, however, poses great challenges, not the least of which is the power supply.
DC/DC Converters in Medical Applications
for Patients and Medical Staff
The following power modes are
supported by most other devices from
other manufacturers too:
• Active Mode
• Sleep Mode
• Deep-Sleep Mode
• Hibernate Mode
• Stop Mode
Let us see generally what these
power modes offer.
1. Active
In this power mode, the CPU, as
well as all other resources on the
chip, are up and running. This mode
contributes towards the major
portion of a system’s overall power
consumption. Various peripherals
available on the chip can be powered
down individually in this mode when
they are not in use.
2. Sleep
This is another common power mode
for controllers. This power mode
primarily relates to the CPU. When
the CPU goes to sleep, its clock is
removed. The only contribution
by the CPU towards the total
power consumption is static power
consumption, as there is no switching
due to the clock and hence no
dynamic power consumption. Other
peripherals like analog-to-digital
conversions (ADCs) and comparators
are available in this mode.
3. Deep Sleep
In this power mode, even the system
clock is disabled. Therefore, none
of the high-frequency resources
are available in this mode. However,
the current state of these resources
remains intact (i.e., CPU registers,
SRAM). As the high-frequency clock
is disabled, power that would have
been consumed due to switching
is saved. Generally, deep-sleep
modes have an option to keep a
low-frequency clock running which
can be used to drive low-frequency
resources such as a timer. Also, this
mode allows developers to use
communication protocol blocks, such
as a I2C slave, which do not need a
clock to be generated by the device
itself. This can be achieved given that
a primary way to enter this mode is to
disable the main clock to the system.
However, power is kept on for the
blocks. This mode mainly contributes
to power consumption through the
static power consumption from all
the blocks on the chip.
4. Hibernate
In this mode, all clocks are switched
off, including the low-speed oscillator,
and power is removed from all
resources on the chip with the
exception of those resources used
as a wake-up source when triggered
by an external event. As almost
everything is powered down in this
mode, this mode yields the least
power consumption as both static
and dynamic power consumption
components have been reduced.
5. Stop
Stop mode, as the name suggests,
powers down all peripherals and does
not even retain RAM and CPU register
contents. In devices like PSoC 4, only
IO pins’ states are retained. Wake
from this mode leads to chip restart.
When analyzing power consumption for
an application, It is important to check
power consumption across all power
modes used.
It’s important to determine which wake-
up sources are available to come out of
a particular power mode. For example,
to come out of sleep mode, any interrupt
may do the job. In hibernate mode; it may
be just the I2C address-match interrupt
that can wake up the device. It is worth
understanding which resources are
working in each mode and which wake up
resources are available. For example, it is
useful to have a comparator interrupt as a
wake-up source in a system to be able to
use an analog input to wake the system
when it exceeds a set threshold. In the
case of the application shown in figure 1,
a wake-up using general-purpose input-
output (GPIO) interrupt would be needed
or even a hard reset would do as the RTC
will be running all the time anyway, and
the controller does not need to retain its
previous state.
“It’s important
to determine
which wake-
up sources are
available to
come out of
a particular
power mode.”
Display active
Controller
sleep Syst
em p
ower
dow
n
Syst
em p
ower
dow
n
Syst
em c
urre
nt
Con
trolle
r act
ive
10s
1ms
20nA
Time
1nA
300nA
Figure 1. Current drawn in various states.
TECH ARTICLESafe Power Supply for Patients and Medical StaffDC/DC Converters in Medical Applications
COVER INTERVIEWLinear Technology’s Products Drive IndiustryInterview with Lothar Maier, CEO of Linear Technology
TECH COLUMNDesigning Embedded Applications for Ultralow-Power, Part 2
TECH ARTICLEReduce Interference & Improve Efficiency with Silent Switcher Designs
CONTENTS
3
44
Power Developer
with Silent Switcher DesignsSwitching regulators replace linear regulators in areas
where low heat dissipation and efficiency are valued. The switching regulator is typically the first active
component on the input-power bus line, and therefore has a significant impact on the electromagnetic interference (EMI) performance of the complete converter circuit.
By Christian Kueck, Strategic Marketing Manager Power Management ProductsLinear Technology Corporation
Reduce InteRfeRence & ImpRove effIcIency
5
TECH ARTICLE
5
with Silent Switcher DesignsSwitching regulators replace linear regulators in areas
where low heat dissipation and efficiency are valued. The switching regulator is typically the first active
component on the input-power bus line, and therefore has a significant impact on the electromagnetic interference (EMI) performance of the complete converter circuit.
By Christian Kueck, Strategic Marketing Manager Power Management ProductsLinear Technology Corporation
Reduce InteRfeRence & ImpRove effIcIency
66
Power Developer
Modern input filter components in surface mount technology have better performance than through-hole parts. However, this improvement is outpaced by the increase in operating switching frequencies of switching regulators. Higher efficiency, low minimum on-off times result in higher harmonic content due to the faster switch transitions.
For every doubling in switching frequency, the EMI becomes 6dB worse when all other parameters, such as switch capacity and transition times, remain constant. The wideband EMI behaves like a first-order high pass with 20dB higher emissions if the switching frequency increases by 10x.
Savvy printed circuit board (PCB) designers will make the hot loops small and use shielding ground (GND) layers as close to the active layer as possible; nevertheless pinout, package construction, thermal design requirements, and package sizes needed for adequate energy storage in decoupling components dictates a
certain minimum hot-loop size.To make layout even more challenging, on a typical planar printed circuit board the magnetic or transformer-style coupling between traces above 30MHz will diminish all filter efforts since the higher the harmonic frequencies are, the more effective unwanted magnetic coupling becomes.
The tried and true solution is to use a shielding box for the complete circuit. Of course, this adds costs, increases required board space, makes thermal management and testing more difficult, and introduces additional assembly costs. Another frequently used method is to slow down the switching edges. This has the undesired effect of reducing the efficiency, increasing minimum on-off times as well as the required dead times, and compromising the potential current-control loop speed.
With Linear’s new LT8614 Silent Switcher™ regulator, you have the effect of a shielded box without using a shield and also eliminate the above-mentioned drawbacks (figure 1).
Figure 1. The LT8614 Silent Switcher minimizes EMI/EMC emissions while delivering high efficiency at frequencies up to 3MHz.
Figure 2. Blue trace is the noise floor; red trace is the LT8614 board at CISPR25 radiated measurement in an anechoic chamber.
Figure 3. Blue trace (ch. 2) is the LT8614; purple trace (ch. 1) is the LT8610; both 13.5Vin, 3.3V out at 2.2A load.
“Higher efficiency,
low minimum on-off
times result in higher
harmonic content due
to the faster switch
transitions.”
7
TECH ARTICLE
7
Modern input filter components in surface mount technology have better performance than through-hole parts. However, this improvement is outpaced by the increase in operating switching frequencies of switching regulators. Higher efficiency, low minimum on-off times result in higher harmonic content due to the faster switch transitions.
For every doubling in switching frequency, the EMI becomes 6dB worse when all other parameters, such as switch capacity and transition times, remain constant. The wideband EMI behaves like a first-order high pass with 20dB higher emissions if the switching frequency increases by 10x.
Savvy printed circuit board (PCB) designers will make the hot loops small and use shielding ground (GND) layers as close to the active layer as possible; nevertheless pinout, package construction, thermal design requirements, and package sizes needed for adequate energy storage in decoupling components dictates a
certain minimum hot-loop size.To make layout even more challenging, on a typical planar printed circuit board the magnetic or transformer-style coupling between traces above 30MHz will diminish all filter efforts since the higher the harmonic frequencies are, the more effective unwanted magnetic coupling becomes.
The tried and true solution is to use a shielding box for the complete circuit. Of course, this adds costs, increases required board space, makes thermal management and testing more difficult, and introduces additional assembly costs. Another frequently used method is to slow down the switching edges. This has the undesired effect of reducing the efficiency, increasing minimum on-off times as well as the required dead times, and compromising the potential current-control loop speed.
With Linear’s new LT8614 Silent Switcher™ regulator, you have the effect of a shielded box without using a shield and also eliminate the above-mentioned drawbacks (figure 1).
Figure 1. The LT8614 Silent Switcher minimizes EMI/EMC emissions while delivering high efficiency at frequencies up to 3MHz.
Figure 2. Blue trace is the noise floor; red trace is the LT8614 board at CISPR25 radiated measurement in an anechoic chamber.
Figure 3. Blue trace (ch. 2) is the LT8614; purple trace (ch. 1) is the LT8610; both 13.5Vin, 3.3V out at 2.2A load.
“Higher efficiency,
low minimum on-off
times result in higher
harmonic content due
to the faster switch
transitions.”
88
Power Developer
Figure 4. Ch 1: LT8610, ch 2: LT8614 switch node rising edge both at 8.4V in, 3.3Vout at 2.2A.
Figure 5. Ch 1: LT8610, ch 2: LT8614, both at 13.2V in, 3.3V 2.2A out.
fixed switching frequency of 700KHz.To compare the LT8614 Silent Switcher technology against a current state-of-the-art switching regulator, the part was measured against an LT8610. The test was performed in a gigahertz transverse electromagnetic (GTEM) cell using the same load, input voltage, and inductor on the standard demo boards for both parts.
One can see that up to a 20dB improvement is made using the LT8614 Silent Switcher technology compared to the already very good EMI performance of the LT8610, especially in the more-difficult-to-manage higher frequency area. This enables simpler and more compact designs where the LT8614 switching power supply needs less filtering and distance compared to other sensitive systems in the overall design.
In the time domain, the LT8614 shows very benign behavior on the switch node edges, as shown in figure 4. Even at 4ns/div, the LT8614 Silent Switcher regulator shows very low ringing (ch. 2 in figure 3). The LT8610 has a good damped ringing (ch. 1, figure 3) but one can see the higher energy stored in the hot loop compared to the LT8614 (ch. 2).
Figure 5 shows the switch node at 13.2V in. One can see the extremely low deviation from the ideal square wave of the LT8614 (ch. 2). All-time domain measurements in figures 3 to 5 are done with 500MHz Tektronix P6139A probes with close probe-tip shield connection to the PCB GND plane, both on the standard demo boards.
“With Linear’s
new LT8614 Silent
Switcher™ regulator,
you have the effect
of a shielded box
without using a
shield.”
The LT8614 has a world-class low quiescent current (IQ) of the LT861x series of only 2.5µA operating current. This is the total supply current consumed by the device, in regulation, with no load. It features the same ultralow dropout of this family, which is only limited by the internal top switch. Unlike alternative solutions, the LT8614’s resistance between drain and source on (RDS(on)) is not limited by maximum duty cycle and minimum off times. The part skips its switch-off cycles in dropout and performs only the minimum required off cycles to keep the internal top switch boost-stage voltage sustained, as shown in figure 6.
At the same time, the minimum operating input voltage is 2.9V typical (3.4V max.) and the device can supply a 3.3V rail with the part in dropout. The LT8614 is higher efficiency than the LT8610/11 at high currents since its total switch resistance is lower. It can also be synchronised to an external frequency operating from 200kHz to 3MHz.
The AC switch losses are low, so it can be operated at high switching frequencies without much efficiency loss. In EMI-sensitive applications such as automotive environment, a good balance can be attained, and the LT8614 can run either below the AM band for even lower EMI, or above the AM band. In a setup with 700KHz operating switching frequency, the standard LT8614 demo board does not exceed the noise floor in a CISPR25 measurement.The figure 2 measurements were taken in an anechoic chamber 12V in 3.3V out at 2A with a
9
TECH ARTICLE
9
Figure 4. Ch 1: LT8610, ch 2: LT8614 switch node rising edge both at 8.4V in, 3.3Vout at 2.2A.
Figure 5. Ch 1: LT8610, ch 2: LT8614, both at 13.2V in, 3.3V 2.2A out.
fixed switching frequency of 700KHz.To compare the LT8614 Silent Switcher technology against a current state-of-the-art switching regulator, the part was measured against an LT8610. The test was performed in a gigahertz transverse electromagnetic (GTEM) cell using the same load, input voltage, and inductor on the standard demo boards for both parts.
One can see that up to a 20dB improvement is made using the LT8614 Silent Switcher technology compared to the already very good EMI performance of the LT8610, especially in the more-difficult-to-manage higher frequency area. This enables simpler and more compact designs where the LT8614 switching power supply needs less filtering and distance compared to other sensitive systems in the overall design.
In the time domain, the LT8614 shows very benign behavior on the switch node edges, as shown in figure 4. Even at 4ns/div, the LT8614 Silent Switcher regulator shows very low ringing (ch. 2 in figure 3). The LT8610 has a good damped ringing (ch. 1, figure 3) but one can see the higher energy stored in the hot loop compared to the LT8614 (ch. 2).
Figure 5 shows the switch node at 13.2V in. One can see the extremely low deviation from the ideal square wave of the LT8614 (ch. 2). All-time domain measurements in figures 3 to 5 are done with 500MHz Tektronix P6139A probes with close probe-tip shield connection to the PCB GND plane, both on the standard demo boards.
“With Linear’s
new LT8614 Silent
Switcher™ regulator,
you have the effect
of a shielded box
without using a
shield.”
The LT8614 has a world-class low quiescent current (IQ) of the LT861x series of only 2.5µA operating current. This is the total supply current consumed by the device, in regulation, with no load. It features the same ultralow dropout of this family, which is only limited by the internal top switch. Unlike alternative solutions, the LT8614’s resistance between drain and source on (RDS(on)) is not limited by maximum duty cycle and minimum off times. The part skips its switch-off cycles in dropout and performs only the minimum required off cycles to keep the internal top switch boost-stage voltage sustained, as shown in figure 6.
At the same time, the minimum operating input voltage is 2.9V typical (3.4V max.) and the device can supply a 3.3V rail with the part in dropout. The LT8614 is higher efficiency than the LT8610/11 at high currents since its total switch resistance is lower. It can also be synchronised to an external frequency operating from 200kHz to 3MHz.
The AC switch losses are low, so it can be operated at high switching frequencies without much efficiency loss. In EMI-sensitive applications such as automotive environment, a good balance can be attained, and the LT8614 can run either below the AM band for even lower EMI, or above the AM band. In a setup with 700KHz operating switching frequency, the standard LT8614 demo board does not exceed the noise floor in a CISPR25 measurement.The figure 2 measurements were taken in an anechoic chamber 12V in 3.3V out at 2A with a
1010
Power Developer
Figure 6. Ch 1: LT8610, ch 2: LT8614 switch node dropout behavior.
Figure 7. LT8614 dropout behavior.
Besides their 42V absolute maximum input voltage rating in automotive environments, the dropout behavior is also very important. Often critical 3.3V logic supplies need to be supported through cold-crank situations. The LT8614 Silent Switcher regulator maintains the close to ideal behavior of the LT861x family in this case. Instead of higher undervoltage lockout voltages and maximum-duty cycle clamps of alternative parts, the LT8610/11/14 devices operate down to 3.4V and start skipping off cycles as soon as necessary, as shown in figure 6. This results in the ideal dropout behavior, as shown in figure 7. The LT8614’s low minimum on-time of 30ns enables large step-down ratios even at high switching frequencies. As a result, it can supply logic-core voltages with a single step-down from inputs up to 42V.
In conclusion, the LT8614 Silent Switcher regulator reduces EMI from current state-of-the-art switching regulators by more than 20dB, while increasing conversion efficiencies with no drawbacks. A 10x improvement of EMI in the frequency range above 30MHz is attained without compromising minimum on-off times or efficiency in the same board area. This is accomplished with no special components or shielding, representing a significant breakthrough in switching regulator design. This level of performance in a single integrated circuit (IC) has not been possible until now. This is just the sort of breakthrough product that allows end-system designers to take their products to the next level.
The increased amount of data and video
being transmitted via the cloud has placed
huge bandwidth and power demands on the
infrastructure market. Designers a need a power
partner with the expertise to improve system
efficiency and simplify the design process.
Power Modules
Digital Power
Integrated FET Regulators
Switching Regulators
FPGA Power
Hot Plug
Voltage Monitors
Power Sequencers
See Intersil’s leadership
power technologies at
intersil.com/infrastructure
Go to Intersil.com/power-management
WERINGPINFRASTRUCTURE
Download DatasheetRead App NoteOrder Demo BoardWatch Video
intersil.com
• Integrated high-side, low-side FETs
• Input voltage range 4.5V to 18V
• Output voltage adjustable from 0.8V, ±1%
• Efficiency up to 95%
The ISL85003/3A are the industry’s most efficient 3A synchronous buck regulators.
Intersil’s new technology delivers the
industry’s lowest RDS(ON) performance.
1212
Power Developer
Safe Power Supply
By Thomas RechlinSenior FAE for Europe RECOM Engineering, Gmunden, Austria
Only about 100 years ago, doctors had to rely on simple equipment such as microscopes and stethoscopes to make a diagnosis. Surgical procedures always posed a huge risk
since the equipment was somewhat rudimentary. The life and well-being of patients therefore primarily depended on the skills and experience of the surgeon. Today, hospitals are full of high-tech equipment and computerised devices that allow for much more detailed and early diagnoses, as well as keyhole surgeries. These new technologies help minimize the risk to patients and ensure fast recovery. The technical progress made in the field of medicine, however, poses great challenges, not the least of which is the power supply.
DC/DC Converters in Medical Applications
for Patients and Medical Staff
13
TECH ARTICLE
13
Safe Power Supply
By Thomas RechlinSenior FAE for Europe RECOM Engineering, Gmunden, Austria
Only about 100 years ago, doctors had to rely on simple equipment such as microscopes and stethoscopes to make a diagnosis. Surgical procedures always posed a huge risk
since the equipment was somewhat rudimentary. The life and well-being of patients therefore primarily depended on the skills and experience of the surgeon. Today, hospitals are full of high-tech equipment and computerised devices that allow for much more detailed and early diagnoses, as well as keyhole surgeries. These new technologies help minimize the risk to patients and ensure fast recovery. The technical progress made in the field of medicine, however, poses great challenges, not the least of which is the power supply.
DC/DC Converters in Medical Applications
for Patients and Medical Staff
1414
Power Developer
“Functional insulation is the most straightforward and reliable type
of insulation between the input and output ends of a device.”
INSULATION IS KEY
Insulation is one of the key issues. This might not be very obvious since the relevant standard prescribes an insulation value of 3kVDC/1s. But what exactly does that mean? To fully appreciate the complexity of the issue, let’s have a closer look at what insulation actually means. The two main factors are the clearance and the creepage distance. The parameters define the distance that must be kept between the primary and the secondary circuit in a power supply. While the permissible values vary from application to application, they must always conform to binding standards.
Clearance and Creepage Distance
The air clearance is defined as the shortest distance through the air between two conductive elements (lower left diagram). The clearance must be sufficient to prevent arc-over.
The creepage distance is the shortest distance on the surface of an insulating material between two conductive elements (lower right diagram). It is crucial that creepage currents are kept to a minimum.
Please note that the air clearance and creepage distances in resin-encapsulated converters are typically measured between the pins, since the resin material is generally acting as an insulator.
However, data sheets of DC/DC converters should still contain reliable information regarding the internal air clearance and creepage distances both inside the transformer and the PCB.
INSULATION LEVEL
Closely related to the clearance and the creepage distance is the insulation level, which determines the insulation voltage. The insulation level determines the voltage that a DC/DC converter or power supply can safely withstand over a defined period of time. However, the relevant specifications are not always referring to the same thing since they differ with the actual voltage kVDC or kVAC and obviously with the specified time period (per second, minute, or permanent). In addition, one needs to take into account that the test voltage is often only applied for one second in the course of a hipot test. The values for longer periods of time are normally extrapolated and labelled in data sheets with “rated.”
For medical technical equipment, the main standards prescribe an insulation of a minimum of 3kVDC/1s. As this value is not always disclosed in the data sheets of the various manufacturers, it can be difficult to compare devices. That is why RECOM has devised an online isolation calculator (figure 1) that can be accessed at www.Recom-international.com.
INSULATION TYPE
There are three insulation types. Functional insulation is the most straightforward and reliabletype of insulation between the input and output ends of a device. It normally consists of a varnish applied to the winding wires of the transformer. With this type of insulation, the wires are wound in layers around a shared core. With this method, it is possible to achieve reliable insulation of up to 4kVDC/1s.
Much more effective and safe is double or basic insulation. With these methods, the primary and secondary windings are separated by an additional insulation barrier. In ring-core transformers, this is achieved by placing a bridge at the centre of the ring core to physically separate the windings (figure 2). With this insulation, it is, however, not possible to wind the wires one over the other, so that the electromagnetic properties might be impaired, leading to a lower rate of efficiency.
Figure 1. Online tool for calculating the correct insulation.
Effective insulation can also be achieved with what is known as a “potted core.” With this method of construction, the core and primary winding are placed in a plastic pot which is filled with epoxy. The secondary winding is then wound around the pot (figure 2). While this solution is more expensive, it is the insulation type of choice where high efficiency is required. Basic insulation caters to insulation values of up to 6.4kVDC/1s.
The best insulation type, however, is reinforced insulation. Here, the primary and the secondary windings are separated by a minimum of two separate insulation barriers. This is normally done by using special winding technology and by placing special foils between the windings (figure 3). In addition, certain specifications as regards the air clearance and creepage distances within the transformer as well as on the PCB must be met. Reinforced insulation provides effective insulation of up to 10kVDC/1s.
Figure 2. Pot core transformer (left) and transformer with bridge at the center acting as an insulation barrier (right).
15
TECH ARTICLE
15
“Functional insulation is the most straightforward and reliable type
of insulation between the input and output ends of a device.”
INSULATION IS KEY
Insulation is one of the key issues. This might not be very obvious since the relevant standard prescribes an insulation value of 3kVDC/1s. But what exactly does that mean? To fully appreciate the complexity of the issue, let’s have a closer look at what insulation actually means. The two main factors are the clearance and the creepage distance. The parameters define the distance that must be kept between the primary and the secondary circuit in a power supply. While the permissible values vary from application to application, they must always conform to binding standards.
Clearance and Creepage Distance
The air clearance is defined as the shortest distance through the air between two conductive elements (lower left diagram). The clearance must be sufficient to prevent arc-over.
The creepage distance is the shortest distance on the surface of an insulating material between two conductive elements (lower right diagram). It is crucial that creepage currents are kept to a minimum.
Please note that the air clearance and creepage distances in resin-encapsulated converters are typically measured between the pins, since the resin material is generally acting as an insulator.
However, data sheets of DC/DC converters should still contain reliable information regarding the internal air clearance and creepage distances both inside the transformer and the PCB.
INSULATION LEVEL
Closely related to the clearance and the creepage distance is the insulation level, which determines the insulation voltage. The insulation level determines the voltage that a DC/DC converter or power supply can safely withstand over a defined period of time. However, the relevant specifications are not always referring to the same thing since they differ with the actual voltage kVDC or kVAC and obviously with the specified time period (per second, minute, or permanent). In addition, one needs to take into account that the test voltage is often only applied for one second in the course of a hipot test. The values for longer periods of time are normally extrapolated and labelled in data sheets with “rated.”
For medical technical equipment, the main standards prescribe an insulation of a minimum of 3kVDC/1s. As this value is not always disclosed in the data sheets of the various manufacturers, it can be difficult to compare devices. That is why RECOM has devised an online isolation calculator (figure 1) that can be accessed at www.Recom-international.com.
INSULATION TYPE
There are three insulation types. Functional insulation is the most straightforward and reliabletype of insulation between the input and output ends of a device. It normally consists of a varnish applied to the winding wires of the transformer. With this type of insulation, the wires are wound in layers around a shared core. With this method, it is possible to achieve reliable insulation of up to 4kVDC/1s.
Much more effective and safe is double or basic insulation. With these methods, the primary and secondary windings are separated by an additional insulation barrier. In ring-core transformers, this is achieved by placing a bridge at the centre of the ring core to physically separate the windings (figure 2). With this insulation, it is, however, not possible to wind the wires one over the other, so that the electromagnetic properties might be impaired, leading to a lower rate of efficiency.
Figure 1. Online tool for calculating the correct insulation.
Effective insulation can also be achieved with what is known as a “potted core.” With this method of construction, the core and primary winding are placed in a plastic pot which is filled with epoxy. The secondary winding is then wound around the pot (figure 2). While this solution is more expensive, it is the insulation type of choice where high efficiency is required. Basic insulation caters to insulation values of up to 6.4kVDC/1s.
The best insulation type, however, is reinforced insulation. Here, the primary and the secondary windings are separated by a minimum of two separate insulation barriers. This is normally done by using special winding technology and by placing special foils between the windings (figure 3). In addition, certain specifications as regards the air clearance and creepage distances within the transformer as well as on the PCB must be met. Reinforced insulation provides effective insulation of up to 10kVDC/1s.
Figure 2. Pot core transformer (left) and transformer with bridge at the center acting as an insulation barrier (right).
1616
Power Developer
CERTIFICATION FOR MEDICAL APPLICATIONS
In June of 2012, EN 60601-1 (3rd edition) for medical technical equipment and systems, came into force in EU member nations. It would, however, be wrong to assume that the equivalent international standards based on the International Electrotechnical Commission (IEC) standard were introduced on the same day. In the U.S., for instance, the UL 60601-1 standard (3rd edition) came into force in July 2013 and the Canadian equivalent (CSA C22.2 no. 601.1) in April 2013. In many other countries, including Japan
and Australia, the new standards are still being drawn up, and other nations such as China have not even started the ratification process.
Another obstacle for manufacturers is the fact that the requirements regarding certification laid down in the existing standards vary greatly from country to country. In Europe, all devices, new and existing, must be certified according to the 3rd edition. In the U.S. and Canada, the new requirements only apply to new designs. For an overview of the new rules, see figure 4.
The main difference between the 2nd and the 3rd editions is the new distinction between patient and operator protection. The safety requirements for means of operator protection (MOOP) are significantly lower than those for means of patient protection (MOPP) and generally correspond to those laid down in EN 60950-1. The requirements for MOPP are much more stringent than before, especially with regard to insulation. Table 1 lists the insulation requirements for the two categories. It is important to note that all requirements for both means of protection must be met. This ensures that the patient or operator are still fully
protected even in the event of failure of a safety device or construction.
Another important change is the increase of the maximum permissible earth leakage current by factor 10. This is a consequence of the new MOOP and MOPP concept. The total patient leakage current is classified based on patient’s environment—the type of the device, with which the patient is in contact. The closer the contact between the device and the patient, the lower the permissible leakage current. Table 2 shows the applicable limits for normal conditions (NC) and single fault conditions (SFC).
Figure 4. Introduction dates and rules of IEC 60601-1, 3rd edition.
Table 1. Insulation requirements for class up to 250VAC and class up to 43VDC or 30VAC (grey boxes).
Table 2: Leakage current limit by device type.
“Based on a risk index matrix, all risks that might potentially arise from the power supply
must be analyzed and weighted.”
“When choosing a converter, it is advisable to obtain the respective risk-management reports from the manufacturer.”
17
TECH ARTICLE
17
CERTIFICATION FOR MEDICAL APPLICATIONS
In June of 2012, EN 60601-1 (3rd edition) for medical technical equipment and systems, came into force in EU member nations. It would, however, be wrong to assume that the equivalent international standards based on the International Electrotechnical Commission (IEC) standard were introduced on the same day. In the U.S., for instance, the UL 60601-1 standard (3rd edition) came into force in July 2013 and the Canadian equivalent (CSA C22.2 no. 601.1) in April 2013. In many other countries, including Japan
and Australia, the new standards are still being drawn up, and other nations such as China have not even started the ratification process.
Another obstacle for manufacturers is the fact that the requirements regarding certification laid down in the existing standards vary greatly from country to country. In Europe, all devices, new and existing, must be certified according to the 3rd edition. In the U.S. and Canada, the new requirements only apply to new designs. For an overview of the new rules, see figure 4.
The main difference between the 2nd and the 3rd editions is the new distinction between patient and operator protection. The safety requirements for means of operator protection (MOOP) are significantly lower than those for means of patient protection (MOPP) and generally correspond to those laid down in EN 60950-1. The requirements for MOPP are much more stringent than before, especially with regard to insulation. Table 1 lists the insulation requirements for the two categories. It is important to note that all requirements for both means of protection must be met. This ensures that the patient or operator are still fully
protected even in the event of failure of a safety device or construction.
Another important change is the increase of the maximum permissible earth leakage current by factor 10. This is a consequence of the new MOOP and MOPP concept. The total patient leakage current is classified based on patient’s environment—the type of the device, with which the patient is in contact. The closer the contact between the device and the patient, the lower the permissible leakage current. Table 2 shows the applicable limits for normal conditions (NC) and single fault conditions (SFC).
Figure 4. Introduction dates and rules of IEC 60601-1, 3rd edition.
Table 1. Insulation requirements for class up to 250VAC and class up to 43VDC or 30VAC (grey boxes).
Table 2: Leakage current limit by device type.
“Based on a risk index matrix, all risks that might potentially arise from the power supply
must be analyzed and weighted.”
“When choosing a converter, it is advisable to obtain the respective risk-management reports from the manufacturer.”
1818
Power Developer
Apart from technical changes, the new standard now demands a formal risk analysis according to ISO 14971, which might pose a challenge to certain power supply manufacturers. Based on a risk index matrix, all risks that might potentially arise from the power supply must be analysed and weighted. The matrix takes into account the occurrence, probability (unlikely to frequent), and impact (insignificant to catastrophic) of the potential risk, based on a rating system from 1 to 5 in each category. If the risk index is ≤6 (probability x impact), the risk is deemed acceptably low. Risks with a higher index must be completely eliminated.
These requirements are difficult to assess, especially for DC/DC converter manufactures, since the end device, which obviously has a major influence on the risk level, is often not known. When choosing a converter, it is advisable to obtain the respective risk-management reports from the manufacturer. Only if these documents are in place, is it possible to deal with the power supply as a “black box,” which speeds up the certification process for the actual medical technical device.
CONVERTERS FOR MEDICAL TECHNICAL EQUIPMENT
In order to meet the above requirements, and in particular the limits for insulation and leakage current, a combination of high-quality AC/DC for medical application and DC/DC converters is often the most efficient solution. This approach makes it easier to meet the stringent requirements of double patient protection.
RECOM offers a wide range of medical-grade DC/DC converters from 0.25W to 15W. These converters are equipped with basic or reinforced insulation for ratings from 3kV to 10kVDC/1s. They are certified according to EN/UL 60601-1 3rd edition, and EN/UL 60950-1, and do not contain any hazardous substances according to the RoHS2 and REACH directives. As usual for RECOM devices, these products come with a three-year warranty.
Thomas RechlinSenior FAE for Europe RECOM Engineering Gmunden, Austria
t.rechlin@ recom-power.com
+43(0)7612/9003-3102
AUTHOR
Classification of Patient Environment
Type B (Body):
No physical contact with patient.
Type BF (Body Float):
No physical contact with patient, or no risk to patient from device failure.
Type CF (Cardiac Float):
Direct contact to patient’s heart and risk of injury or death in the event of device failure.
19
TECH ARTICLE
19
Your Circuit Starts Here.Sign up to design, share, and collaborate
on your next project—big or small.
Click Here to Sign Up
20
Power Developer
Drive Industry
Linear Technology’s
ProductsLothar Maier CEO of Linear Technology
Linear Technology Corporation designs, manufactures, and markets a broad line of high-performance analog integrated circuits.
The company’s products provide an essential bridge between the analog world and digital electronics in communications, networking, industrial, automotive, computer, medical, instrumentation, consumer, military, and aerospace systems. Linear Technology has managed to negotiate the vagary of these tech industry markets while staying committed to high-quality products.
EEWeb spoke with Lothar Maier, CEO of Linear Technology, about the company’s impressive operating margin and about the evolution of automobile electronics. Maier also discussed the popularity of LTspice®.
COVER INTERVIEW
21
Drive Industry
Linear Technology’s
ProductsLothar Maier CEO of Linear Technology
Linear Technology Corporation designs, manufactures, and markets a broad line of high-performance analog integrated circuits.
The company’s products provide an essential bridge between the analog world and digital electronics in communications, networking, industrial, automotive, computer, medical, instrumentation, consumer, military, and aerospace systems. Linear Technology has managed to negotiate the vagary of these tech industry markets while staying committed to high-quality products.
EEWeb spoke with Lothar Maier, CEO of Linear Technology, about the company’s impressive operating margin and about the evolution of automobile electronics. Maier also discussed the popularity of LTspice®.
22
Power Developer
Linear Technology is known to have the highest operating margin in the industry. How did the company get to this point?
There is no single factor that leads to the margins that we see here at Linear. However, the most important element, from a product standpoint, is that we make new, unique products. We do not try to copy anything from our competitors or second-source another competitor’s products. We come up with products that are not only technically interesting, but also compelling in terms of bringing value to the customer. This may sound easy, but is a rather tall order because it typically takes two to three years for us to develop a product. Then it takes another few years for the customer to take our product and design it into their product. We are always making decisions to develop products that customers are going to buy and find valuable five years in the future—this is tough to do, but we do it.
Another key factor that contributes to our high operating margin is product pricing. We price each product based on the value it brings to the customers, not the transistor count. Also, we recognize that we not only have to develop superior products, but we have to actually deliver them. Linear has a great industry reputation for providing products on time and with short lead times. The combination of all of these aspects results in Linear having optimal operating margins.
Which markets do you feel are the most important to target?
The best markets, historically speaking, have changed. Fifteen years ago, Linear was very concentrated in the communications market and then came the “dot-com bubble” which made that market less important. We used to have a very strong presence in consumer products with high-end analog components that were sold into digital cameras and high-functioning cellular phones. This was a significant part of our business. Around 2007, it became obvious that this consumer business, which we thought was high-performance analog, had deteriorated to just plain consumer grade. That is when we made the decision to move away from those markets and find new ones. At that time, we refocused the company on the automotive, communications, and industrial markets. That brings us to where we are today. Seven years ago, it would have seemed unwise to become so involved in the automotive and industrial markets, but these markets today have become some of the fastest-growing segments of the analog market. We believe we are in the right markets, at the right time.
“We are always making decisions to develop products that customers are going to buy and find valuable five years in the future—this is tough to do, but we do it.”
COVER INTERVIEW
23
Linear Technology is known to have the highest operating margin in the industry. How did the company get to this point?
There is no single factor that leads to the margins that we see here at Linear. However, the most important element, from a product standpoint, is that we make new, unique products. We do not try to copy anything from our competitors or second-source another competitor’s products. We come up with products that are not only technically interesting, but also compelling in terms of bringing value to the customer. This may sound easy, but is a rather tall order because it typically takes two to three years for us to develop a product. Then it takes another few years for the customer to take our product and design it into their product. We are always making decisions to develop products that customers are going to buy and find valuable five years in the future—this is tough to do, but we do it.
Another key factor that contributes to our high operating margin is product pricing. We price each product based on the value it brings to the customers, not the transistor count. Also, we recognize that we not only have to develop superior products, but we have to actually deliver them. Linear has a great industry reputation for providing products on time and with short lead times. The combination of all of these aspects results in Linear having optimal operating margins.
Which markets do you feel are the most important to target?
The best markets, historically speaking, have changed. Fifteen years ago, Linear was very concentrated in the communications market and then came the “dot-com bubble” which made that market less important. We used to have a very strong presence in consumer products with high-end analog components that were sold into digital cameras and high-functioning cellular phones. This was a significant part of our business. Around 2007, it became obvious that this consumer business, which we thought was high-performance analog, had deteriorated to just plain consumer grade. That is when we made the decision to move away from those markets and find new ones. At that time, we refocused the company on the automotive, communications, and industrial markets. That brings us to where we are today. Seven years ago, it would have seemed unwise to become so involved in the automotive and industrial markets, but these markets today have become some of the fastest-growing segments of the analog market. We believe we are in the right markets, at the right time.
“We are always making decisions to develop products that customers are going to buy and find valuable five years in the future—this is tough to do, but we do it.”
24
Power Developer
In the industrial market, a product can be designed into an application for 10 to 20 years. The engineering invested into those products can be leveraged over many years, whereas in the consumer space, the product sales rise and decline quickly, and then there are no legacy sales to pay for that research and development investment.
What do you see as being the path forward for Linear, aside from automotive and industrial?
The plan for Linear over the next five years is to stay the course. We are in all the right markets, we have a huge head start, and the amount of electronics in automobiles cannot be stopped. Everything that is mechanical in a car is at some point going to become electric because this saves weight, which will ultimately improve fuel economy. Everything in a car engine that is attached to the belts will eventually become electrical. When that happens, it will become a whole new game in terms of applications because these modifications will put more demand on the internal system in the car. The command center in a car is not going to be adequate as is, so there will be more electrical components to do all of the housekeeping.
The challenge with designing for this market is that the sale of products in the automotive industry revolves around providing a part that solves a specific problem. However, as the car becomes more electrical, this becomes a suboptimal way of addressing problems. Automobile electronics are becoming
“It is one thing to provide good products, but you have to provide good service as well. LTspice® is just one of those products that hits both those marks.”
more of a system, not just a collection of mechanical components bolted together. Also, in the past, we have dealt with the suppliers to the automotive companies, and now we are dealing more closely with the car companies themselves.
Switching gears a little bit, could you talk about the philosophy behind LTspice®—why do you give it away for free, and why do you feel it is the best simulation program with integrated-circuit emphasis (SPICE) tool on the market?
The application goes back 15 years. Over the years, the original developer of our CAD tool developed the first version of LTspice®. Obviously, it is tailored to run Linear’s products, but we also designed it to run some competitors’ products as well. As I said before, it is one thing to provide good products, but you have to provide good service as well. LTspice® is just one of those products that hits both those marks. From an ease-of-use standpoint, it is an intuitive system, and we get much praise from our customers on the overall product.
“Seven years ago, it would have seemed unwise to become so involved in the automotive and industrial markets, but these markets today have become some of the fastest-growing segments of the analog market.”
What are the differences in engaging with the automotive and industrial markets as opposed to the consumer market?
The design-in and design-out cycles are much different. In consumer markets, you can get designed in relatively quickly, so you need a treadmill of new products ready to go at any time. A few years ago, we were doing well at keeping up with the high demand of consumer products. We had great margins, and we were being designed into first-generation and second-generation devices, but it became
obvious with the third-generation products that oftentimes, cost became more important than performance. We could make products that would extend the battery life of a consumer product from 10 hours to 20 hours, but nobody was willing to pay more for that—those types of advancements were not highly valued. With the automotive market, it takes much longer to develop a product that gets designed into an automobile. We are creating designs in vehicles right now that will not show up until model-year 2018. But once the product is designed in, it’s there for a long time.
COVER INTERVIEW
25
In the industrial market, a product can be designed into an application for 10 to 20 years. The engineering invested into those products can be leveraged over many years, whereas in the consumer space, the product sales rise and decline quickly, and then there are no legacy sales to pay for that research and development investment.
What do you see as being the path forward for Linear, aside from automotive and industrial?
The plan for Linear over the next five years is to stay the course. We are in all the right markets, we have a huge head start, and the amount of electronics in automobiles cannot be stopped. Everything that is mechanical in a car is at some point going to become electric because this saves weight, which will ultimately improve fuel economy. Everything in a car engine that is attached to the belts will eventually become electrical. When that happens, it will become a whole new game in terms of applications because these modifications will put more demand on the internal system in the car. The command center in a car is not going to be adequate as is, so there will be more electrical components to do all of the housekeeping.
The challenge with designing for this market is that the sale of products in the automotive industry revolves around providing a part that solves a specific problem. However, as the car becomes more electrical, this becomes a suboptimal way of addressing problems. Automobile electronics are becoming
“It is one thing to provide good products, but you have to provide good service as well. LTspice® is just one of those products that hits both those marks.”
more of a system, not just a collection of mechanical components bolted together. Also, in the past, we have dealt with the suppliers to the automotive companies, and now we are dealing more closely with the car companies themselves.
Switching gears a little bit, could you talk about the philosophy behind LTspice®—why do you give it away for free, and why do you feel it is the best simulation program with integrated-circuit emphasis (SPICE) tool on the market?
The application goes back 15 years. Over the years, the original developer of our CAD tool developed the first version of LTspice®. Obviously, it is tailored to run Linear’s products, but we also designed it to run some competitors’ products as well. As I said before, it is one thing to provide good products, but you have to provide good service as well. LTspice® is just one of those products that hits both those marks. From an ease-of-use standpoint, it is an intuitive system, and we get much praise from our customers on the overall product.
“Seven years ago, it would have seemed unwise to become so involved in the automotive and industrial markets, but these markets today have become some of the fastest-growing segments of the analog market.”
What are the differences in engaging with the automotive and industrial markets as opposed to the consumer market?
The design-in and design-out cycles are much different. In consumer markets, you can get designed in relatively quickly, so you need a treadmill of new products ready to go at any time. A few years ago, we were doing well at keeping up with the high demand of consumer products. We had great margins, and we were being designed into first-generation and second-generation devices, but it became
obvious with the third-generation products that oftentimes, cost became more important than performance. We could make products that would extend the battery life of a consumer product from 10 hours to 20 hours, but nobody was willing to pay more for that—those types of advancements were not highly valued. With the automotive market, it takes much longer to develop a product that gets designed into an automobile. We are creating designs in vehicles right now that will not show up until model-year 2018. But once the product is designed in, it’s there for a long time.
2626
Power Developer
By Sachin Gupta, Kannan Sadasivam Cypress Semiconductor
Part 2
Designing
ULTRALOW-forEmbedded Applications
POWER
Generally, a system on a chip (SoC) supports more low-power modes compared to a traditional microcontroller unit (MCU). The reason for this is that due to the high level of integration, they have more
on-chip components, and multiple power profiles are needed to support different operating needs. The number of power modes and the resources available during each mode vary from device to device. For example, in a particular low-power mode, one device may power down everything and just retain register and RAM content while another may just power-down the central processing unit (CPU) while keeping other resources up and running. Different manufacturers may name these modes differently as well. For this article, we’ll cite the example of PSoC 4 devices from Cypress Semiconductor to explore power modes in detail.
To read the previous article in this series, click on the image above.
By Sachin Gupta, Kannan Sadasivam, Cypress
Part 1
Designing
ULTRALOW-forEmbedded Applications
POWER
Considering the importance of conservation, the portability of embedded systems is a key design consideration. Portable systems are generally battery-powered with
battery life dependent upon the systems’ power consumption. These days, as part of “Go Green” initiatives, power consumption is a selection criterion even for wall-powered applications.
27
TECH COLUMN
27
By Sachin Gupta, Kannan Sadasivam Cypress Semiconductor
Part 2
Designing
ULTRALOW-forEmbedded Applications
POWER
Generally, a system on a chip (SoC) supports more low-power modes compared to a traditional microcontroller unit (MCU). The reason for this is that due to the high level of integration, they have more
on-chip components, and multiple power profiles are needed to support different operating needs. The number of power modes and the resources available during each mode vary from device to device. For example, in a particular low-power mode, one device may power down everything and just retain register and RAM content while another may just power-down the central processing unit (CPU) while keeping other resources up and running. Different manufacturers may name these modes differently as well. For this article, we’ll cite the example of PSoC 4 devices from Cypress Semiconductor to explore power modes in detail.
To read the previous article in this series, click on the image above.
By Sachin Gupta, Kannan Sadasivam, Cypress
Part 1
Designing
ULTRALOW-forEmbedded Applications
POWER
Considering the importance of conservation, the portability of embedded systems is a key design consideration. Portable systems are generally battery-powered with
battery life dependent upon the systems’ power consumption. These days, as part of “Go Green” initiatives, power consumption is a selection criterion even for wall-powered applications.
2828
Power Developer
The following power modes are supported by most other devices from other manufacturers too:
• Active Mode
• Sleep Mode
• Deep-Sleep Mode
• Hibernate Mode
• Stop Mode
Let us see generally what these power modes offer.
1. Active In this power mode, the CPU, as well as all other resources on the chip, are up and running. This mode contributes towards the major portion of a system’s overall power consumption. Various peripherals available on the chip can be powered down individually in this mode when they are not in use.
2. Sleep This is another common power mode for controllers. This power mode primarily relates to the CPU. When
the CPU goes to sleep, its clock is removed. The only contribution by the CPU towards the total power consumption is static power consumption, as there is no switching due to the clock and hence no dynamic power consumption. Other peripherals like analog-to-digital conversions (ADCs) and comparators are available in this mode.
3. Deep Sleep In this power mode, even the system clock is disabled. Therefore, none of the high-frequency resources are available in this mode. However, the current state of these resources remains intact (i.e., CPU registers, SRAM). As the high-frequency clock is disabled, power that would have been consumed due to switching is saved. Generally, deep-sleep modes have an option to keep a low-frequency clock running which can be used to drive low-frequency resources such as a timer. Also, this mode allows developers to use communication protocol blocks, such as a I2C slave, which do not need a clock to be generated by the device itself. This can be achieved given that a primary way to enter this mode is to
disable the main clock to the system. However, power is kept on for the blocks. This mode mainly contributes to power consumption through the static power consumption from all the blocks on the chip.
4. Hibernate In this mode, all clocks are switched off, including the low-speed oscillator, and power is removed from all resources on the chip with the exception of those resources used as a wake-up source when triggered by an external event. As almost everything is powered down in this mode, this mode yields the least power consumption as both static and dynamic power consumption components have been reduced.
5. Stop Stop mode, as the name suggests, powers down all peripherals and does not even retain RAM and CPU register contents. In devices like PSoC 4, only IO pins’ states are retained. Wake from this mode leads to chip restart.
When analyzing power consumption for an application, It is important to check power consumption across all power modes used.
It’s important to determine which wake-up sources are available to come out of a particular power mode. For example, to come out of sleep mode, any interrupt may do the job. In hibernate mode; it may be just the I2C address-match interrupt that can wake up the device. It is worth understanding which resources are working in each mode and which wake up resources are available. For example, it is useful to have a comparator interrupt as a wake-up source in a system to be able to use an analog input to wake the system when it exceeds a set threshold. In the case of the application shown in figure 1, a wake-up using general-purpose input-output (GPIO) interrupt would be needed or even a hard reset would do as the RTC will be running all the time anyway, and the controller does not need to retain its previous state.
“It’s important to determine which wake-up sources are available to come out of a particular power mode.”
Display activeController
sleep
Sys
tem
pow
er d
own
Sys
tem
pow
er d
own
Sys
tem
cur
rent
Con
trolle
r act
ive
10s1ms
20nA
Time
1nA
300nA
Figure 1. Current drawn in various states.
29
TECH COLUMN
29
The following power modes are supported by most other devices from other manufacturers too:
• Active Mode
• Sleep Mode
• Deep-Sleep Mode
• Hibernate Mode
• Stop Mode
Let us see generally what these power modes offer.
1. Active In this power mode, the CPU, as well as all other resources on the chip, are up and running. This mode contributes towards the major portion of a system’s overall power consumption. Various peripherals available on the chip can be powered down individually in this mode when they are not in use.
2. Sleep This is another common power mode for controllers. This power mode primarily relates to the CPU. When
the CPU goes to sleep, its clock is removed. The only contribution by the CPU towards the total power consumption is static power consumption, as there is no switching due to the clock and hence no dynamic power consumption. Other peripherals like analog-to-digital conversions (ADCs) and comparators are available in this mode.
3. Deep Sleep In this power mode, even the system clock is disabled. Therefore, none of the high-frequency resources are available in this mode. However, the current state of these resources remains intact (i.e., CPU registers, SRAM). As the high-frequency clock is disabled, power that would have been consumed due to switching is saved. Generally, deep-sleep modes have an option to keep a low-frequency clock running which can be used to drive low-frequency resources such as a timer. Also, this mode allows developers to use communication protocol blocks, such as a I2C slave, which do not need a clock to be generated by the device itself. This can be achieved given that a primary way to enter this mode is to
disable the main clock to the system. However, power is kept on for the blocks. This mode mainly contributes to power consumption through the static power consumption from all the blocks on the chip.
4. Hibernate In this mode, all clocks are switched off, including the low-speed oscillator, and power is removed from all resources on the chip with the exception of those resources used as a wake-up source when triggered by an external event. As almost everything is powered down in this mode, this mode yields the least power consumption as both static and dynamic power consumption components have been reduced.
5. Stop Stop mode, as the name suggests, powers down all peripherals and does not even retain RAM and CPU register contents. In devices like PSoC 4, only IO pins’ states are retained. Wake from this mode leads to chip restart.
When analyzing power consumption for an application, It is important to check power consumption across all power modes used.
It’s important to determine which wake-up sources are available to come out of a particular power mode. For example, to come out of sleep mode, any interrupt may do the job. In hibernate mode; it may be just the I2C address-match interrupt that can wake up the device. It is worth understanding which resources are working in each mode and which wake up resources are available. For example, it is useful to have a comparator interrupt as a wake-up source in a system to be able to use an analog input to wake the system when it exceeds a set threshold. In the case of the application shown in figure 1, a wake-up using general-purpose input-output (GPIO) interrupt would be needed or even a hard reset would do as the RTC will be running all the time anyway, and the controller does not need to retain its previous state.
“It’s important to determine which wake-up sources are available to come out of a particular power mode.”
Display activeController
sleep
Sys
tem
pow
er d
own
Sys
tem
pow
er d
own
Sys
tem
cur
rent
Con
trolle
r act
ive
10s1ms
20nA
Time
1nA
300nA
Figure 1. Current drawn in various states.
3030
Power Developer
Sachin Gupta has several years of experience in embedded solution development. He holds a bachelor’s degree in electronics and communications from Guru Gobind Singh Indraprastha University, Delhi. He can be reached at [email protected].
Kannan Sadasivam is a Staff Applications Engineer with Cypress Semiconductor Corp. He has spent a considerable amount of his past career designing and integrating satellite subsystems. He loves working on different types of analog circuits and applications. He can be reached at [email protected].
Hibernate and stop mode can bring down the power consumption to as low as 100 nA. For the real-time clock (RTC) itself, you can easily find very low power RTCs that consume in the range of 100–200 nA. Assuming that the controller is directly driving the LCD, we can consider the liquid-crystal display (LCD) off-state power consumption to be zero.
This puts our average power for the system shown in figure 1 in the range of 300 nA. If we assume that the design uses a CR2032 as the power source, then we are looking at around 225 mAh of capacity. With a 300 nA of current, this battery would support operation for approximately 70–80 years but only if the device was in power-down mode the entire time.
Every time a button is pressed, the controller wakes up. This can pump up the controller’s power consumption into the 500 uA-1mA range. Assume a 1 mA range where the controller is fetching data from the RTC and displaying it on the LCD. While the controller can
perform this activity in a very small amount of time, the display needs to be kept active for longer (10 seconds for the user to read the time). Since this is a directly driven LCD, this means the controller has to remain active for longer, thus consuming more charge. For such instances, devices like Cypress’s PSoC4 have low-power modes where the device shuts down all peripherals and only runs those blocks that are necessary to drive an LCD. In these devices, the LCD drive is run in a specific low-power mode called digital correlation mode. The result is significantly reduced current consumption.
Every key press would go through a current profile as shown in figure 1. The area under the curve is the typical power consumption during a single key press. The charge consumed can be calculated as:
Q = (1mA*1ms) + (20uA*10s)
Based on the above number, we can calculate how many key presses can be supported with a give power source.
In the next part of this series, we will examine a larger system as an example and explore how to reduce average power consumption. Some system-level techniques will also be discussed that can help to reduce average power consumption.
“A faster MCU can finish tasks quickly and spend more time in low-power modes, yielding a system that consumes less power.”
Time spent in the active mode is critical since that draws the maximum power. One option is that the MCU can stay in active mode running at a slow CPU clock with reduced active power. However, this might result in a higher, average power consumption because the time spent in the active mode depends on the clock frequency. Here, power consumption will be determined by the time taken by the MCU to process data since the MCU must stay in active mode. A faster MCU can finish tasks quickly and spend more time in low-power modes, yielding a system that consumes less power. Depending on the system requirements, it is up to the system designer to determine the most optimal configuration.
31
TECH COLUMN
31
Sachin Gupta has several years of experience in embedded solution development. He holds a bachelor’s degree in electronics and communications from Guru Gobind Singh Indraprastha University, Delhi. He can be reached at [email protected].
Kannan Sadasivam is a Staff Applications Engineer with Cypress Semiconductor Corp. He has spent a considerable amount of his past career designing and integrating satellite subsystems. He loves working on different types of analog circuits and applications. He can be reached at [email protected].
Hibernate and stop mode can bring down the power consumption to as low as 100 nA. For the real-time clock (RTC) itself, you can easily find very low power RTCs that consume in the range of 100–200 nA. Assuming that the controller is directly driving the LCD, we can consider the liquid-crystal display (LCD) off-state power consumption to be zero.
This puts our average power for the system shown in figure 1 in the range of 300 nA. If we assume that the design uses a CR2032 as the power source, then we are looking at around 225 mAh of capacity. With a 300 nA of current, this battery would support operation for approximately 70–80 years but only if the device was in power-down mode the entire time.
Every time a button is pressed, the controller wakes up. This can pump up the controller’s power consumption into the 500 uA-1mA range. Assume a 1 mA range where the controller is fetching data from the RTC and displaying it on the LCD. While the controller can
perform this activity in a very small amount of time, the display needs to be kept active for longer (10 seconds for the user to read the time). Since this is a directly driven LCD, this means the controller has to remain active for longer, thus consuming more charge. For such instances, devices like Cypress’s PSoC4 have low-power modes where the device shuts down all peripherals and only runs those blocks that are necessary to drive an LCD. In these devices, the LCD drive is run in a specific low-power mode called digital correlation mode. The result is significantly reduced current consumption.
Every key press would go through a current profile as shown in figure 1. The area under the curve is the typical power consumption during a single key press. The charge consumed can be calculated as:
Q = (1mA*1ms) + (20uA*10s)
Based on the above number, we can calculate how many key presses can be supported with a give power source.
In the next part of this series, we will examine a larger system as an example and explore how to reduce average power consumption. Some system-level techniques will also be discussed that can help to reduce average power consumption.
“A faster MCU can finish tasks quickly and spend more time in low-power modes, yielding a system that consumes less power.”
Time spent in the active mode is critical since that draws the maximum power. One option is that the MCU can stay in active mode running at a slow CPU clock with reduced active power. However, this might result in a higher, average power consumption because the time spent in the active mode depends on the clock frequency. Here, power consumption will be determined by the time taken by the MCU to process data since the MCU must stay in active mode. A faster MCU can finish tasks quickly and spend more time in low-power modes, yielding a system that consumes less power. Depending on the system requirements, it is up to the system designer to determine the most optimal configuration.
Sierra Circuits:A Complete PCB Resource
PLUS: The Ground ” Myth in PrintedCircuits
“
PCB Resin Reactor+
Ken BahlCEO of Sierra Circuits
Let There Be
How Cree reinvented the light bulb
LIGHT
David ElienVP of Marketing & Business
Development, Cree, Inc.
New LED Filament Tower
Cutting Edge Flatscreen Technologies
+
+
M o v i n g T o w a r d s
a Clean Energy
FUTURE— Hugo van Nispen, COO of DNV KEMA
MCU Wars 32-bit MCU Comparison
Cutting Edge
SPICEModeling
Freescale and TI Embedded
Modules
ARMCortex
Programming
From Concept to
Reality Wolfgang Heinz-Fischer
Head of Marketing & PR, TQ-Group
Low-Power Design Techniques
TQ-Group’s Comprehensive Design Process
+
+
PowerDeveloper
Octobe r 20 13
Designing forDurability
View more EEWeb magazines— Click Here