A Non-synchronous Buck dc-dc Converter, high side P-type MOSFET
This is one of the simpler circuits available for a moderate current dc-dc buck converter. The use of a p-type MOSFET allows for very simple gate drive circuitry. An MC34063 controller chip provides the basic control and drives a p-type enhancement mode MOSFET as a switch.
The circuit design is described for a 12V to 6V dc-dc converter. The output current will be effectively constant at 1A. The output peak-peak ripple voltage is not critical but will be chosen to be 1% of the output voltage, that is 0.06V. The application is intended to power a 6V light from a 12V source.
The MC34063 is set to switch at approximately 60kHz by using a 220pF timing capacitor. This is close to the 100kHz limit of the device and allows relatively small reactive components to be used. In this design a current limiting resistor is not included. A suitable value can be selected using the design equations in the MC34063 data sheet for the desired limiting current.
The MC34063 compares the output (using a very simple voltage mode feedback in this design) with a 1.25V reference. To achieve an output of 6V it must be divided down by 4.8 as shown to a 1.25V level and passed to the COMP input.
The output of the MC34063 consists of an emitter follower driver feeding an output BJT that is intended to be used as the switch in very simple circuits. This can be made into a Darlington pair by connecting the two collectors (pins 8 and 1) together as shown. This combination does not saturate and therefore can be made to be quite fast.
The output is pulled high via a 1K resistor and switches the MOSFET gate through a totem pole driver. This driver circuit amplifies the drive current to charge the MOSFET gate rapidly. The current is limited by the 1K base resistor and the large signal beta of the BJTs in the driver. The totem pole configuration is very suitable for this application as the BJTs do not saturate and are protected from breakdown. A limitation is that it can drive the output only to within about 0.3V of the power rails, but this is not a difficulty for this circuit.
The IRF5305 p-type MOSFET is inexpensive and readily available. It has a typical on-resistance of 60 mΩ and a peak current of 31A. The on-resistance is much lower than is needed for this application but the device will be useful in other contexts. The gate charge needed to switch on the MOSFET to a gate-source voltage of 12V is about 45nC. This could be reduced to 30nC if a gate-source switching voltage of 8V were used. The gate charge tends to have an inverse relationship to the on-resistance and a different MOSFET choice may give a lower gate charge at the expense of on-resistance.
The 1N5819 is a 1W, 1A maximum average current Schottky diode which drops about 0.4V at 1A forward current. For the design the average current limit of the diode will restrict the current output to 2A.
C4 and C3 are low ESR capacitors. The 16V rated 47μF capacitor has a nominal ESR of 0.3Ω C1 and C2 are decoupling capacitors for the controller and driver separately.
The inductor used for the circuit is a toroidal inductor measuring about 200μH and scrounged from an old computer monitor. No further detail is available at this stage concerning the core saturation characteristics or the ESR.
The measured efficiency of the circuit was 88%, giving 6.08V at 1.06A into a 5.6Ω load, and drawing 0.63A from an 11.5V bench supply.
Note: The totem-pole driver as shown above doesn't completely discharge the MOSFET gate capacitance and results in a somewhat slow turn-off time. This can be an issue at the higher switching frequencies. The effect can be ameliorated by providing power to the totem-pole driver and pullup resistor that is a few volts above the circuit input voltage. Typically this will need a charge pump boost circuit. With this the upper BJT will remain on as the MOSFET gate charge reduces. One might argue however that if a charge pump voltage boost is used, the circuit may as well use better performing N-MOSFETs.
An SMPS is normally specified in terms of the output ripple voltage in addition to the output voltage and current. For this exercise the ripple voltage is computed rather than specified. The choice of the output capacitor is arbitrary, although a low ESR type was obtained. The design here covers the selection of the basic converter components. The various equations can be found in any of the vast number of publications available on the Internet and will not be derived here (but beware that some of them are wrong).
Let VI be the input voltage (12V), VO be the output voltage (6V), VF be the diode forward voltage (nominally 0.4V at 1A) and VS be the switch on-state voltage drop (0.06V at 1A). The duty cycle is:
D = (VO + VF ) / (VI - VS + VF ) = 0.516
Choice of inductor affects the system response to load changes, the ripple current and the boundary below which the circuit operates in discontinuous mode. High ripple current affects the output voltage ripple via the capacitor ESR. Below the discontinuous mode boundary the circuit is much harder to control. Given a choice of 60kHz operating frequency and a ripple voltage of 0.06V, the selected 47μF capacitor limits the ripple current to 0.2A. The ripple current is
IR = (VI - VO - VS) DT/L
where T is the switching period (17 microseconds) and L is the inductance. This gives a required inductance of 256μH. The continuous-discontinuous mode boundary occurs when the current at the start and end of the switching period is zero, and occurs at an average current which is half of the peak-peak ripple current (0.1A).
As an aside, the output capacitor will ideally filter the ripple current to give an average ripple voltage VR of
VR = IR(D2 + (1-D)2)T/8C
which gives a value of 7mV for the 47μF capacitor. This is negligible compared to the 60mV developed across the capacitor ESR, and is a typical situation unless particularly low ESR capacitors are selected. It has been pointed out also that ripple current has a lower limit set by the feedback comparator switching imprecision, which is a few millivolts.
The losses in the circuit come from the following sources:
The quickest way to obtain estimates of the transient losses is by means of a SPICE simulation. This will also allow us to observe the dynamics of the MOSFET turn-on phase. The following circuit was used:
Models for the active devices were obtained from various Internet sites. The output stage of the MC34063 is modelled with a pair of 2N2222 BJTs. Voltage sources V2 and V3 are set to 0V and are inserted to allow current to be plotted. The output shows 6.08V with a peak-peak ripple of 0.06V or 1%. The current in V2 (lower red trace), the gate voltage (blue) and the MOSFET drain voltage (orange) are shown below.
The power loss in charging and discharging the MOSFET gate is given by the area under the product of the gate current and the gate to source voltage, and indicates a loss of about 18mW or 0.3% of the output power. The commonly used equation for the power required to turn on the MOSFET:
PG = QG VG FS
where QG is the gate charge, VG is the gate voltage and FS is the switching frequency, gives a value of 25mW for this device.
Finally the controller itself contributes a quiescent loss of 48mW plus 72mW in the output transistors (which pull 12mA for the on period of the cycle), which is 2% total of the output power. The total comes to an efficiency of 95%. This doesn't include the losses in the inductor. The highest contributor is the diode voltage drop.
The circuit below indicates the compactness that can be achieved with prototyping board. It is used to power a 6V fluorescent light from a 12V supply.
The following publications provide excellent resources for understanding and designing SMPS power supplies, and are obtainable on the Internet.
First created 5 August 2010
Last Modified 31 January 2014
© Ken Sarkies 2010