 |
 |
|
 Remember that the gate of a MOSFET is just a capacitor. The mfgs. data sheet usually only gives Ciss, which is the gate capacity at Vgate = 0. But in truth, the gate capacity rises as Vgate increases. During turn-On, the capacitance rises to 4 x Ciss through Vthresh as the FET switches from Off to On. Vthresh is the threshold voltage of the MOSFET. At this point, one must continue to supply current to the gate, though Vgate rises only slowly here. Once past this point, increasing Vgate lowers the ON resistance of the FET. A good rule of thumb is that the average gate capacitance is 2 x Ciss. Thus a gate drive circuit must, in the end, supply a net charge Qg = 2CissVgate. Remember that FETs are fast enough that there is a tradeoff between losing power by failing to turn the FET fully on, and losing energy per cycle by overcharging the gate each Ts. Vgatemax is 15 volts for most FETs, but I have found ~10 Volts more efficient when drive power is included in the equation! Since power is energy x frequency, and Cgate is charged to Vgate each cycle, drive power is Pdrive = Ugate x fs Ugate = (1/2)Cgate x Vgate**2 Pdrive = Ciss x Vgate**2 x fs A more useful form for gate drive design is Pdrive = Ugate x fs Pdrive = 1/2 x Qgate x Vgate x fs. Note that Qgate is the total charge (in coulombs) impressed on the Gate to turn the FET ON. Thus, on an oscilloscope, the area under the drive current waveform is just Qgate!   What if I chose the wrong topology?
Remember that the superiority of the boostbuck topologies is a function of the arrangement of the power components, not of the components themselves! So, if your design doesn't meet spec, there is an easy way out! A simple board layout change will rearrange you topology to become a Cuk Converter. Just reroute the power traces to form the "Optimum Topology Converter" instead of what you're using now. A buck with input filter and isolation transformer can usually be converted into an isolated Cuk Converter in one pass!  These old standbys from the crystal radio days still feature the lowest forward drops available, at 0.3 Volts. This is versus 0.5 for Schottkys, and 0.7 for Si. This makes them ideal for diode OR situations! Though they are slow, the L2C2 time constant of a converter will hold up the output! I was working on the Quantum the other day, and needed a metric hex wrench, which I didn't have. It was just one hex head bolt, so I tried wrapping a couple of layers of tin foil over the end of the 1/4" key. Worked like a charm! Background | Tips | Engineering vs. Science | | Isolation of the Cuk Converter | Historical Perspective | | Return Home | Safety First | Lesson in Ethics | The Consequences of Bad Design | Real World Applications | Deals | e-mail me! | |
||
 |
 |