"Ideal" Power Diode
There is no
such thing as an ideal diode of course, but the following circuits can
provide a much closer approximation to diode ideality in ways that suit
the reverse current blocking requirements of certain power circuits. These are sometimes referred to as "smart diodes".
application in mind is a battery charger. When charging from a source
that can drop below the battery voltage, there is a danger of the
battery discharging back through the charger and possibly damaging the
latter. As such a diode is usually inserted between the charger and
battery to block this reverse current flow. An ideal diode circuit may
be valuable in high power applications to reduce power and voltage
and heat dissipation demands.
Care needs to
be taken when considering its use in
non-synchronous switch mode power supplies to replace a catch diode.
The MOSFET takes a certain amount of time to turn on and off, and this
may result in destructive shoot through currents if other MOSFETs in
the circuit have turned on. In these cases it is preferable to make use
of a synchronous mode circuit in which the controller inserts delays to
avoid the situation.
controller ICs are available from some manufacturers, for example
Linear Technology and Maxim produce a small range of these that work
with or incorporate a power MOSFET, sensing the voltage across the
MOSFET and turning it off when the voltage reverses. They can be
somewhat expensive but can provide a range of additional safety
circuit only needs readily available BJTs and a handful of resistors,
as well as the power MOSFET. It requires a supply to the BJTs that will
allow the MOSFET gate-source voltage to be raised to a level to allow
full turn-on of the MOSFET. I
haven't yet seen this particular circuit in the literature but it may
well exist. There are several forms of this circuit:
this one places the MOSFET on the negative terminal of the battery.
Other similar circuits are described below.
circuit can be adapted easily enough to other topologies. To switch the
positive supply line the circuit can either use a P-type MOSFET and PNP
BJTs, or use the same circuit as above in conjuction with an auxiliary
low power supply, such as a charge pump, to allow the BJTs to pull the
MOSFET gate above its turn-on voltage.
How it Works
The MOSFET Q1
is N-type in this circuit and operates in reverse mode, that is, it
conducts when the drain-source voltage is negative. This voltage is
intended to be less than the forward voltage of the body diode so the
gate has control over its operation. If the current through the MOSFET
is too high, the drain-source voltage may rise to the point where the
body diode turns on, which defeats the main purpose of the circuit.
Therefore a MOSFET with a suitably low on-resistance needs to be chosen
to match the expected current load. The Schottky diode protects the
MOSFET from large body diode currents during turn-on transients.
The circuit takes advantage of the strong dependence of BJT collector current on the base-emitter voltage. Q3 clearly controls the gate voltage while the purpose of Q2 is to cancel the base-emitter voltage of Q3 and so measure the voltage difference across the MOSFET. If the voltage across the MOSFET is positive, the emitter of Q2 rises above ground, causing Q2 to turn off and Q3 to turn on. This pulls down the collector of Q3 to turn off the MOSFET. Note that the base voltage of Q3 can rise quite high, pushing it into saturation. However this doesn't appear to affect the circuit operation greatly.
voltage across the MOSFET goes negative, the emitter of Q2 is pulled
below ground and Q3 starts to turn off. The MOSFET gate voltage rises
until the MOSFET starts to turn on. The gate voltage settles at a point
which is just above the turn-on voltage. This will increase with drain
current to keep the voltage drop across the MOSFET at a level that
activates the BJTs.
The circuit doesn't appear to be particularly sensitive (within limits) to the choice of Q2, even if a different but similar transistor type is used. High gain BJTs will result in a smaller voltage across the MOSFET in the on-state (and hence lower power loss). As such it was found that Sziklai and Darlington pairs could be made to work, although the higher base-emitter voltage of the latter caused the circuit behaviour to be degraded. To reduce the adverse effects of turn-on transients, high speed switching BJTs should be used (such as the 2N2222 used here).
It is possible to improve the on-state drain-source voltage of the MOSFET by amplifying the gate signal with an operational amplifier between Q3 and the MOSFET gate. A high speed operational amplifier with low output impedance needs to be chosen, and this introduces difficulties with instability particulary at higher gains.
circuit offering the same functionality as the above is given in
reference 1. This circuit uses a PNP-NPN pair of BJTs in a
configuration whose operation is somewhat less transparent than the one
described above. The circuit below, based on this configuration, was
tested with SPICE simulation. The connection between the BJTs is such
that the PNP's base-emitter voltage cancels that of the NPN. When the
MOSFET drain-source voltage drops below zero the PNP turns on and
forces the NPN to turn off since its base-emitter voltage is tied to
that of the PNP's collector-emitter voltage. This turns off the MOSFET.
circuit amplifies the voltage across the MOSFET in a cascade pair of
BJTs, thus giving higher current gain than the circuit described above.
Nevertheless the SPICE simulations show that the
behaviour of the circuit is almost identical to that of the circuit we
presented earlier, with a very similar on-state voltage across the
MOSFET despite the high gain. In addition a significant amount of
occurred when the MOSFET turned on. This may indicate that this circuit
is less stable than the earlier one. A
possible cause of this oscillation may be due to the high gain, which
did occur in our circuit when additional gain was added (although the
Sziklai and Darlington
configurations did not show oscillations). A 100pF capacitor placed
across the base resistor Rp seemed to resolve the problem, at least as
far as simulation showed.
This is essentially the same circuit form as that used for type 1 but by making use of p-type MOSFET and BJTs it can be placed in a different part of a power circuit. The circuit used for Spice simulations is as follows:
The IRF5305 is a p-type MOSFET
with an on-resistance of about 20 mΩ. The circuit works in a
similar way to the first one described above. It measures the voltage
across the MOSFET and amplifies it to produce a gate-source voltage
that is pulled down to ground to turn on the MOSFET. This type of
circuit is useful at the output of a battery charger to prevent the
battery discharging back through the power circuits.
The following SPICE simulation was setup to study the behaviour of the type 1 circuit using a pulsed voltage source:
The IRF1404 is a low voltage, logic-level, low on-resistance MOSFET capable of supporting over 160A of drain current. A time-dependent voltage source (sinusoidal or pulse) is used to push the drain voltage of the MOSFET below ground. The 2N2222 transistors are high speed NPN switching BJTs. Vp2 and Vp3 are zero voltage sources used to measure currents. The resistance values shown are appropriate for the MOSFET to ensure that its gate charges and discharges quickly.
plot shows the voltages at the drain, gate and BJT bases for a pulsed
voltage source. The first transient occurs when circuit starts to turn
on the MOSFET. During the turn-on phase the Schottky diode conducts for
brief period until the MOSFET has turned on, thus limiting the
transient peak voltage. The second transient at the end of the pulse is
much shorter, but results in a heavy drain current flow for a short
shown in the second plot below. The impact of this current spike can be
reduced by using faster BJTs.
Finally the drain current flow with a sinusoidal voltage source shows that the transitions behave reasonably well:
static measurements were made to check the efficiency of the circuits.
A type 1 circuit was built in which the MOSFET used was an
IRL1104 and the BJTs were 2SC1815, neither of which are particularly
high performers. The following plot
shows the variation of the on-state current and
drain-source voltage when the MOSFET is on. The voltage drop is
for this low current, so a higher gain BJT or Sziklai configuration
should be used, along with a MOSFET having a lower RDS.
A second circuit was tested using 2N3904 BJTs and an IRF1404 MOSFET, and provided with a "battery" voltage of 12.8V and a source voltage of -11.7V. The results showed a much lower drain-source voltage in the on-state. The following table gives some of the measurements, although they tended to vary strongly as the circuit began to warm.
Measurements made with a Schottky diode across the MOSFET gave consistently higher VDS.
A type 2 circuit was built with a SQD45P03 P-MOSFET having a RDS of about 10mΩ, and two 2N3906 PNP BJTs. R1 and R2 were 15K and RB was 130K. Measurements were made at a range of source voltages and loads, giving the following curves:
The VDS of the MOSFET was
less than 150mV even at 12A current.