haymeron
Member level 3
Point-of-Use Power Converters Trade-off Simplicity,
Efficiency, and Cost
David G. Morrison, Power, Packaging & Components
Editor, Electronic Design
~~~~~~~~~~~~~~~~~~~~~~~~~~
Texas Instruments' (TI's) recent introduction of the
TPS5461x, a series of synchronous dc-dc converters with
on-chip MOSFETs, raises the level of power that can be
obtained with a single-chip point-of-use power
converter. Also referred to as the SWIFT (switcher with
integrated FET technology) family, the chips are buck
regulators that can deliver 6 A of continuous output at
low voltages. With their ability to stepdown a 3.3- or
5-V bus to values as low as 0.9 V, these chips will be
candidates to power the many DSPs, FPGAs, and ASICs
migrating to the lower supply voltages.
The SWIFT chip's low input-voltage range sets it apart
from many of the dc-dc converter chips that are
currently available. So, any comparisons with existing
dc-dc converter ICs must be made with that in mind.
Nevertheless, the chip's ease-of-use should be its
major attraction. The SWIFT converter requires just a
half a dozen passives to complete a stepdown regulator
design. That task is further simplified by a software
application that steps the reader through component
selection
Nevertheless, such simplicity does not come free.
Building a monolithic converter to deliver 6 A at an
85°C ambient with no airflow is an endeavor that
requires a few trade-offs. The value of the TI part in
a given application may depend largely on the
customer's priorities when it comes to factors such as
ease of use, board layout requirements, required
voltages and currents, desired efficiency, and--last--
but-never-least--cost. A comparison of TI's TPS5461x
with a sampling of the existing dc-dc converters
illustrates some of the trade-offs that ensue when
selecting a dc-dc converter chip for point-of-use power
applications (see the table entitled "How TI's TPS5461x
Compares With Other Point-of-Use Converters" under the
News Briefs section on the PlanetEE website at
http://www.planetee.com).
Comparing the TI converter with current chip solutions
is not exactly an apples-to-apples comparison. As noted
previously, the input voltage range on the TI part goes
lower than other chips, but does not extend as high. In
addition, differences in other electrical specs such as
transient response may be significant for a given
application. And then there are the variations in
external circuitry, which dictate different component
counts and values depending on the chip selected.
However, putting aside those distinctions for the
moment, the basic functional similarities will put TI's
SWIFT converter in competition with existing two- and
three-chip dc-dc converter solutions.
Above all else, TI's part offers ease of use. For
system designers who are less experienced in power-
supply design, the SWIFT chip's simplicity may be a
major attraction because it could reduce power-supply
development time significantly. On the other hand,
designers more seasoned in power conversion may not be
daunted by the selection of external FETs and passives
or by board layout. Order quantity is another issue
that comes into play. Design support among chip vendors
is directly proportional to the expected business. If
the expected volume is high enough, the vendors'
applications engineers will do the whole design for
you!
Layout is another important consideration. The SWIFT
reduces the design to just one chip. But that chip is a
rather larger 28-pin TSSOP with an exposed copper pad
that must be soldered to the board for best power
performance. There must be sufficient copper on the pc
board to dissipate the heat generated by the chip. The
use of external FETs may provide the designer with
added flexibility when laying out the board. That's
because the designer has more leeway in placing the
FETs away from the controller to obtain the best
thermal management. The end result may be a board with
less copper (and less cost) when external FETs are
employed. The thermal requirements may also be
mitigated by the use of external FETs because of their
higher efficiency.
The size of the TI converter is itself a consideration.
A single 28-pin TSSOP is larger than the two or three
S0-8s required by other solutions. However, that factor
must be considered in light of the external passives
that are required. It's overall board space that
counts. Keep in mind that the TI part typically
requires fewer external parts.
Efficiency is another issue to consider. The one-chip
approach offers good efficiency, but cannot match
what's available when external FETs are employed. The
numbers in the table illustrate this point in a general
way. However, be aware that these numbers reflect
datasheet values that may involve differences in
operating conditions. Moreover, efficiency is a
function of external components as well as the FETs and
controllers.
Cost is another area where multichip solutions
currently have an edge. In the relatively small, 1000-
piece quantities cited in the table, TI's part is
significantly more expensive than the two- and three-
chip solutions. But as quantities go up, and as the new
TI device ramps up production, the price gap between
these components is likely to narrow.
As with any component that breaks new ground, TI's
TPS5461x offers some exciting possibilities for new
point-of-use power designs. Designers considering this
part for their applications will have to weigh their
requirements carefully against the capabilities of the
TI device and the competition. In doing so, designers
may also want to consult with vendors about the
chipmakers' roadmap for future products. An awareness
of looming product developments can help designers
adopt power-supply circuits and components that serve
not only current product development needs, but those
of future product generations.
For more information about the TPS5461x, see "DC-DC
Converter IC Packs Two 12-A MOSFETs," Electronic Design
Magazine, October 1, 2001, page 40. Also, see TI's
website for complete data on this product family. The
direct link is **broken link removed**.
Readers may also call the
company's Literature Response Center at (800) 477-8924,
ext. 4500.
Efficiency, and Cost
David G. Morrison, Power, Packaging & Components
Editor, Electronic Design
~~~~~~~~~~~~~~~~~~~~~~~~~~
Texas Instruments' (TI's) recent introduction of the
TPS5461x, a series of synchronous dc-dc converters with
on-chip MOSFETs, raises the level of power that can be
obtained with a single-chip point-of-use power
converter. Also referred to as the SWIFT (switcher with
integrated FET technology) family, the chips are buck
regulators that can deliver 6 A of continuous output at
low voltages. With their ability to stepdown a 3.3- or
5-V bus to values as low as 0.9 V, these chips will be
candidates to power the many DSPs, FPGAs, and ASICs
migrating to the lower supply voltages.
The SWIFT chip's low input-voltage range sets it apart
from many of the dc-dc converter chips that are
currently available. So, any comparisons with existing
dc-dc converter ICs must be made with that in mind.
Nevertheless, the chip's ease-of-use should be its
major attraction. The SWIFT converter requires just a
half a dozen passives to complete a stepdown regulator
design. That task is further simplified by a software
application that steps the reader through component
selection
Nevertheless, such simplicity does not come free.
Building a monolithic converter to deliver 6 A at an
85°C ambient with no airflow is an endeavor that
requires a few trade-offs. The value of the TI part in
a given application may depend largely on the
customer's priorities when it comes to factors such as
ease of use, board layout requirements, required
voltages and currents, desired efficiency, and--last--
but-never-least--cost. A comparison of TI's TPS5461x
with a sampling of the existing dc-dc converters
illustrates some of the trade-offs that ensue when
selecting a dc-dc converter chip for point-of-use power
applications (see the table entitled "How TI's TPS5461x
Compares With Other Point-of-Use Converters" under the
News Briefs section on the PlanetEE website at
http://www.planetee.com).
Comparing the TI converter with current chip solutions
is not exactly an apples-to-apples comparison. As noted
previously, the input voltage range on the TI part goes
lower than other chips, but does not extend as high. In
addition, differences in other electrical specs such as
transient response may be significant for a given
application. And then there are the variations in
external circuitry, which dictate different component
counts and values depending on the chip selected.
However, putting aside those distinctions for the
moment, the basic functional similarities will put TI's
SWIFT converter in competition with existing two- and
three-chip dc-dc converter solutions.
Above all else, TI's part offers ease of use. For
system designers who are less experienced in power-
supply design, the SWIFT chip's simplicity may be a
major attraction because it could reduce power-supply
development time significantly. On the other hand,
designers more seasoned in power conversion may not be
daunted by the selection of external FETs and passives
or by board layout. Order quantity is another issue
that comes into play. Design support among chip vendors
is directly proportional to the expected business. If
the expected volume is high enough, the vendors'
applications engineers will do the whole design for
you!
Layout is another important consideration. The SWIFT
reduces the design to just one chip. But that chip is a
rather larger 28-pin TSSOP with an exposed copper pad
that must be soldered to the board for best power
performance. There must be sufficient copper on the pc
board to dissipate the heat generated by the chip. The
use of external FETs may provide the designer with
added flexibility when laying out the board. That's
because the designer has more leeway in placing the
FETs away from the controller to obtain the best
thermal management. The end result may be a board with
less copper (and less cost) when external FETs are
employed. The thermal requirements may also be
mitigated by the use of external FETs because of their
higher efficiency.
The size of the TI converter is itself a consideration.
A single 28-pin TSSOP is larger than the two or three
S0-8s required by other solutions. However, that factor
must be considered in light of the external passives
that are required. It's overall board space that
counts. Keep in mind that the TI part typically
requires fewer external parts.
Efficiency is another issue to consider. The one-chip
approach offers good efficiency, but cannot match
what's available when external FETs are employed. The
numbers in the table illustrate this point in a general
way. However, be aware that these numbers reflect
datasheet values that may involve differences in
operating conditions. Moreover, efficiency is a
function of external components as well as the FETs and
controllers.
Cost is another area where multichip solutions
currently have an edge. In the relatively small, 1000-
piece quantities cited in the table, TI's part is
significantly more expensive than the two- and three-
chip solutions. But as quantities go up, and as the new
TI device ramps up production, the price gap between
these components is likely to narrow.
As with any component that breaks new ground, TI's
TPS5461x offers some exciting possibilities for new
point-of-use power designs. Designers considering this
part for their applications will have to weigh their
requirements carefully against the capabilities of the
TI device and the competition. In doing so, designers
may also want to consult with vendors about the
chipmakers' roadmap for future products. An awareness
of looming product developments can help designers
adopt power-supply circuits and components that serve
not only current product development needs, but those
of future product generations.
For more information about the TPS5461x, see "DC-DC
Converter IC Packs Two 12-A MOSFETs," Electronic Design
Magazine, October 1, 2001, page 40. Also, see TI's
website for complete data on this product family. The
direct link is **broken link removed**.
Readers may also call the
company's Literature Response Center at (800) 477-8924,
ext. 4500.
