How MOSFET leaks

[ayeaye]

Helpful would be the proven circuit to switch mosfet at nanoseconds time (below 100ns) 

Ok, simulated with 100 ohms circuit resistance, the pulse voltage source was

dc 0 pulse 0 5 0 5n 5n 10n 700n

Which means 5V positive pulses, with the initial delay 0, rise time 5ns, fall time 5ns, and length 10ns. Which means that the total length of the pulse was 20ns. The cycle length was 700ns. The transient analysis was done with

tran 1ns 700ns

Which means transient analysis during 700ns with 1ns steps. The current in the circuit in the ngspice diagram below, is clockwise. But calculate yourself please.

For these who have not used ngspice before, helpful may also be the thread that i created about ngspice here http://overunity.com/16064/ngspice-and-geda-circuit-simulation/#.VhFThW943iY .

PS Why the capacitance Cgd decreases when the mosfet opens, i think i didn't describe it well before. The n channel mosfet consists of a slightly p semiconductor, but at the source and gate electrodes, there are n areas. This n area conducts, and thus it is like an extension to the capacitance's plate. When the mosfet opens and a current starts to go through it, more electrons come and these n areas widen, and thus the gate side plate of the capacitance decreases. This is also what should cause the Miller plateau. It's just as i see it by now.

[ayeaye]

I think i didn't say it quite right again. It looks like when the mosfet opens and current starts to go through it, then the n areas decrease. The decreasing of the n areas also causes the decrease of the input capacitances, and the Miller plateau. The n areas conduct, remember. And the p area is only slightly p, so it doesn't conduct. Why, because the electrons that come to the p area and make it conductive, come from the n areas. At least this is how i understand it by now.

I had to explain it, because i think one should know what happens there. This is not anything high level. It's just mechanical. Like understanding how a coffee machine works. It's only necessary to know what exactly happens.

[forest]

I don't understand it all ;-) but I feel that keeping mosfet right around the Miller plateau would allo very fast opening/closing. Am I right ? I know there are scientific papers about circuits switching ordinary mosfets (with trr time in the 2us range) in 5ns
We should concentrate on this, at least i would if I had a team working on solutions :-)

[ayeaye]

I feel that keeping mosfet right around the Miller plateau would allo very fast opening/closing.

Yes i guess. And should decrease the input energy too, as all the charge in Cgs gets wasted when the input goes to zero, and it will be completely discharged to ground. And should also decrease the leaking to the circuit after the pulse, because Cgd will also not be fully discharged.

If you don't understand the leaking, then think about it like this. When the (n-) mosfet is closed, then all the circuit is really connected from the ground to the drain. When we charge Cgd, then the current goes from the ground to drain. And when we discharge Cgd, then the current goes from the drain to ground. This evidently is all the leaking to the circuit there.

[ayeaye]

My hand drawing was a kind of bad, so i added below a drawing made with gschem, which explains mosfet leaking. I hope that all understand it now.

On that drawing, Cgd is not a capacitor, but the (n) mosfet's gate to drain capacitance. Mosfet is not drawn, because on both diagrams above and below, it is closed, that is, it is the same as not connected. The gate to drain capacitance is in both cases connected to the ground, so the leaking to the circuit does not depend on it anyhow.

The arrow shows on both diagrams the direction of the negative current, that is current which is considered to be a movement of negative particles.

Hope it was useful for someone for some purpose.