In my previous column (“Applications Drive Component
Power Designs,” Sept. 11, 2008, p. 18, ED Online
19486), I contended that the application drives the
selection of a specific power component—or it even
forces the creation of a custom component. But, it must be
admitted, there is another side of the coin.
Many price-sensitive engineers see value in changing their
architecture to accept a component that is readily available
from many manufacturers because of the benefits of low price
and product honed to near perfection by intense competition.
There are no more popular power components than those with
48-V inputs.
The telecom industry brought a lot of commodity bricks
to the 48-V realm as a voltage settled on in the early 1980s.
Other markets, such as data communications, followed that
in a migration to 48 V slowly gathering momentum in that
industry, which was booming in the late 1990s, as well as the
emerging distributed power architecture of the time. So anybody
making 48-V converters then had a sweet business.
If designers can gain cost advantages by changing whatever
input voltage is native to the market or application, they’re
likely to consider it. That’s probably what happened as the
intermediate bus architecture (IBA) emerged. Surely it was a
consideration in the selection of 48 V for the first VI chip.
48 V EVERYWHERE
Now, everything in the blade server market is 48 V. Much
medical equipment now has gone to 48 V. Even the 42-V automotive
initiative could take advantage of many 48-V power
products if that ever takes hold, and at some level it has, but it’s
not a mass-produced product yet.
What’s more, many ac front ends put out 48 V now, reinforcing
the trend to 48 V. A search on front ends shows that
it’s the popular voltage now, to bring it down to 48 V and distribute
it from there. Virtually any power designer that wants
to use a standard front end that is also at the leading edge will
select one at 48 V.
As soon as you get 48 V out of your ac front end, it doesn’t
matter what the application is. You’re in the 48-V distribution
business. So immediately, there’s a whole laundry list of
applications that can be plugged into the wall. You name it.
That could be anything, even consumer. You can put bricks
downstream, you can do intermediate bus downstream, you can
do factorized power. It’s your call.
Some ac front ends, of course, deliver high voltage, and that’s
relatively straightforward with a boost converter. But most
power architects don’t want to be distributing high voltage if
it can be avoided, because it’s dangerous, safety approvals are
difficult to get, and there are liabilities. They try to get highvoltage
ac down to some safe voltage in one package.
The highest safe voltage is 60 V, but of course the dc-dc
converter input range is typically 36 to 75—a nominal 48 V—
leaving some headroom from the absolute maximum. At this
level, safety approvals can readily be gained, so a 48-V input is a
perfect call. And it’s been done for years, so it’s a natural.
An ac front end additionally could deliver a lower voltage.
But from a system performance standpoint, busing low voltages
around usually means higher current. As current increases,
the amount of copper increases, the infrastructure of busing
the voltage around becomes larger and more costly, and I2R
losses increase.
Consequently, system efficiency is also impacted by the
interconnect. But for most systems, 48 V represents probably
the best tradeoff for bus voltages in regards to being able to bus
large currents effectively without incurring significant losses
that would occur at 12 V or lower. It’s a safe voltage, easily
accessible for serviceability.
SOME RESTRICTIONS MAY APPLY
But 48-V input converters aren’t for everybody. Some market
segments require a certain operating voltage because of the way
the industry works. As an example, railway applications specify
input voltages of 72 or 110 V. That’s the way the trains are built.
It’s the same with the military, which requires input voltages of
28, 155, or 270 V. In such market segments, a change of input
voltage would require massive and impractical architectural
change. They’re not going to do that any day soon.
Finally, the main driver for this distribution voltage, however,
is cost. And when you peel back the onion, other clear
advantages are present. There have to be. It can’t be a wash
performance-wise. There have to be efficiency, safety, and performance
benefits.
The 48-V input converter is a competitive marketplace,
so a great deal of engineering effort has gone into designing
these products in a commodity world, where manufacturers
have battled each other on cost for decades—not just years,
decades.
So the fact that there’s been so much engineering talent put
toward the development of 48-V converters means there’s a
large selection, and they’ve been cost-reduced many times over.
It’s the best of both worlds—a well-engineered product at a
low cost.
Tom Curatolo holds a BSEE from Worcester Polytechnic
Institute, Worcester, mass. look for more Power Design columns
from him in print and online.