Power Inverter Dead Time and Propagation Delay Specica-
tions
The HCPL-M454 includes a specication intended to help
designers minimize “dead time” in their power inverter
designs. The new “propagation delay dierence” speci-
cation (t
PLH
- t
PHL
) is useful for determining not only how
much optocoupler switching delay is needed to prevent
“shoot-through” current, but also for determining the
best achievable wort-case dead time for a given design.
When inverter power transistors switch (Q1 and Q2 in
Figure 15), it is essential that they never conduct at the
same time. Extremely large currents will ow if there is
any overlap in their conduction during switching transi-
tions, potentially damaging the transistor and even the
surrounding circuitry. This “shoot-through” current is
eliminated by delaying the turn-on of one transistor (Q2)
long enough to ensure that the opposing transistor (Q1)
has completely turned o. This delay introduces a small
amount of “dead time” at the output of the inverter dur-
ing which both transistors are o during switching tran-
sitions. Minimizing this dead time is an important design
goal for an inverter designer.
The amount of turn-on delay needed depends on the
propagation delay characteristics of the optocoupler,
as well as the characteristics of the transistor base/gate
drive circuit. Considering only the delay characteristics
of the optocoupler (the characteristics of the base/gate
drive circuit can be analyzed in the same way), it is im-
portant to know the minimum and maximum turn-on
(t
PHL
) and turn-o (t
PLH
) propagation delay specica-
tions, preferably over the desired operating temperature
range. The importance of these specications is illustrat-
ed in Figure 16. The waveforms labeled “LED1”, “LED2”,
“OUT1”, and “OUT2” are the input and output voltages
of the optocoupler circuits driving Q1 and Q2 respec-
tively. Most inverters are designed such that the power
transistor turns on when the optocoupler LED turns on;
this ensures that both power transistors will be o in the
event of a power loss in the control circuit. Inverters can
also be designed such that the power transistor turns o
when the optocoupler LED turns on; this type of design,
however, requires additional fail-safe circuitry to turn o
the power transistor if an over-current condition is de-
tected. The timing illustrated in Figure 16 assumes that
the power transistor turns on when the optocoupler LED
turns on.
The LED signal to turn on Q2 should be delayed enough
so that an optocoupler with the very fastest turn-on
propagation delay (t
PHLmin
) will never turn on before
an optocoupler with the very slowest turn-o propaga-
tion delay (t
PLHmax
) turns o. To ensure this, the turn-on
of the optocoupler should be delayed by an amount no
less than (t
PLHmax
- t
PHLmin
), which also happens to be the
maximum data sheet value for the propagation delay dif-
ference specication, (t
PLH
- t
PHL
). The HCPL-M454 speci-
es a maximum (t
PLH
- t
PHL
) of 1.3 µs over an operating
temperature range of 0-70°C.
Although (t
PLH
- t
PHL
)
max
tells the designer how much
delay is needed to prevent shoot-through current, it is
insucient to tell the designer how much dead time a
design will have. Assuming that the optocoupler turn-
on delay is exactly equal to (t
PLH
- t
PHL
)
max
, the minimum
dead time is zero (i.e., there is zero time between the
turn-o of the very slowest optocoupler and the turn-on
of the very fastest optocoupler).
Calculating the maximum dead time is slightly more
complicated. Assuming that the LED turn-on delay is still
exactly equal to (t
PLH
- t
PHL
)
max
, it can be seen in Figure 16
that the maximum dead time is the sum of the maximum
dierence in turn-on delay plus the maximum dierence
in turn-o delay,
[(t
PLHmax
-t
PLHmin
) + (t
PHLmax
-t
PHLmin
)],
This expression can be rearranged to obtain
[(t
PLHmax
-t
PHLmin
) - (t
PHLmin
-t
PHLmax
)],
and further rearranged to obtain
[(t
PLH
-t
PHL
)
max
- (t
PLH
-t
PHL
)
min
],
which is the maximum minus the minimum data sheet
values of (t
PLH
- t
PHL
). The dierence between the maxi-
mum and minimum values depends directly on the to-
tal spread of propagation delays and sets the limit on
how good the worst-case dead time can be for a given
design. Therefore, optocouplers with tight propagation
delay specications (and not just shorter delays or lower
pulse-width distortion) can achieve short dead times in
power inverters. The HCPL-M454 species a minimum
(t
PLH
- t
PHL
) of -0.7 µs over an operating temeprature
range of 0-70°C, resulting in a maximum dead time of 2.0
µs when the LED turn-on delay is equal to (t
PLH
- t
PHL
)
max
,
or 1.3 µs.
It is important to maintain accurate LED turn-on delays
because delays shorter than (t
PLH
- t
PHL
)
max
may allow
shoot-through currents, while longer delays will increase
the worst-case dead time.
