Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: current limit on 48 VDC


I have designed 48 volt power feed circuits for Lucent's propritary
digital sets for too many years.

I've used PTCs.  They are too slow to limit startup surges and may not
reset automatically.

The latest circuits use custom electronic switches.  These custom
switches contain current limiting and thermal shut down.  They are
protected by MOVs which clamp the interface to about 200V wrt ground.

In a previous message, roger stated "using silicon on the wire will
prove that it
has its own challanges from the transient perspective".
I have not found this to be the case.

I designed the PSE unit for the Lucent group.  The reference design that
Bob presented passed 9KV (contact) and a separate unit passed 10kV (air
discharge).  I would like to take credit for a special design, but this
is simply not the case.  Actually no protection components were

So why did the units survive?
The simple answer is that we cheated.  Instead of protecting the PSE,
the units simply pass the surges on to the PD and DTE.  Both contain
large capacitors which absorb the surges while creating only a small
change in the 48 volts.  My thanks to the PD and DTE designers for
protecting the PSE.

In this pass it on technique, the MOSFET can be considered a short in
all cases.  In the power on state, the MOSFET is saturated.  In the off
state, the MOSFET is a 60 volt power zener.  For reverse polarities the
MOSFET is a diode.

This techniques eventually fails.  At the 9kV level there are sufficient
sparks to cause the micro controller to latch up.  Not sure what the
exact cause is.  It could be the current source sense resistor, or one
of the transistors which are "connected to the wire".

Latch up is usually a problem.  In the PSE case, the 5 volt supply is
limited to about 15ma.  This is sufficient to maintain latchup, but
insufficient to cause damage.  Note that if the micro had been on the
PHY side where a 15A supply is more likely, it probably would have been
fried.  Hence "connected to the wire" can be safer for IC's than on the
PHY side.

Dieter Knollman
Lucent - soon to be retired

Why pay for something you could get for free?
NetZero provides FREE Internet Access and Email