How MOSFET leaks

[ayeaye]

Below is the circuit symbol of a mosfet (an n-channel mosfet). The two input capacitances of the mosfet can be seen on that symbol, as circled there. The gate-source capacitance is greater of these.

Mosfet can leak at two times. When switching on, then the input capacitances are charged, when the mosfet still has a high resistance. This period is very short though, as mosfet switches on fast, and its resistance when it is switched on, is extremely small, so there is no considerable leakage then.

The second time is when the mosfet is switching off. Then its input capacitances discharge, but this doesn't happen instantly, so some current goes to the circuit. For the circuit the input capacitances are connected in series. The input capacitance of a mosfet is quite great, may be some 500pF.

I wrote it for not to avoid mosfets to be used in the overunity experiments. Their switching is very fast, and they are the most widely used transistors. When the output power gain of the circuit is more than the energy necessary for charging the mosfet's input capacitance once in a switching cycle, then i think that it's quite certain that there is an overunity. The same as bipolar transistors, they leak, but this leakage has to be considered in the circuits and calculations.

Saying that, it looks like, when looking at the simulation diagrams, that mosfet also has some 42 k ohms resistance, between the gate and the drain. So maybe that should be also considered. And it also looks like that it doesn't leak when switched off, at all. Or maybe it only leaks because of that resistance.

I hope that what i wrote here would be useful for someone for some purpose. Sorry if not.

[ayeaye]

Ok, i now simulated it with ngspice (gEDA), to find out, the first figure below is the simulated test circuit. The third figure below shows the currents in the simulated circuit, blue is the current in the circuit, and red is the gate circuit. When talking about currents here hereafter, it is about positive currents, which is considered to be conventional. The direction of the currents on that diagram is what is considered positive, for the gate current, the current towards the gate, and for the current in the circuit, the current which is considered positive for the mosfet, that is current towards its source. The 600 units on that diagram is 600 uA. The second figure below is the equivalent circuit of the mosfet.

Look at the currents diagram. The current direction in the circuit is always opposite to the change of the gate current. And the shapes of the signal are rounded and delayed. This can only be inductance, nothing else. And look at the mosfet's equivalent circuit, there is an inductance Ls before the source.

So it now seems to be clear how the leaking happens. What is important for leaking is only the gate-source capacitance Cgs, and the inductance before the source Ls. Notice that in the circuit, Cgs is not connected directly to the ground, but there is Ls between it and the ground. This causes the leakage exactly as it happens, and i will explain how.

In the test circuit, when Cgs charges, there is an increase of current towards the ground. This increase of current current induces current in Ls in the opposite direction, away from the ground, and this current goes to the circuit. That current on the currents diagram has exactly that direction.

When Cgs discharges, there is an increase of current away from the ground. This increase of current current induces current in Ls in the opposite direction, towards the ground. That current on the currents diagram has exactly that direction.

So by that, mosfet leaks because of Cgs and Ls, and not considerably in any other way. The capacitance Cgs can be calculated from the data in the mosfet's datasheet. The inductance of Ls can be calculated based on the test results. It also should appear in the mosfet's spice model. And based on Cgs and Ls it can be calculated how much the mosfet leaks in any particular circuit.

So TinselKoala, you were wrong wrong wrong. Mosfet doesn't leak because its internal capacitances, but because of its internal capacitances *and* its internal inductance. The mosfet's leakage is mostly said to be because of its internal capacitances, and this causes a huge confusion and makes completely impossible to understand what happens in some circuits. Because leakage due to capacitance and leakage due to capacitance and inductance are completely different, they behave in completely different ways. And the direction of some currents is the direct opposite.

Hope that this could help anyone. Sorry if not.

[TinselKoala]

No, I am not "wrong wrong wrong".  Your simulations actually show that I am "right right right".

But whatever. You can continue to fool yourself all you like, I'm tired of playing this silly game.

[ayeaye]

I was not serious but, i think anyone who has to do with electronics knows, that circuits that contain only capacitors, and circuits that contain both capacitors and coils, behave very differently.

[TinselKoala]

Here's your proof. I've taken an unused IRF624A mosfet (grabbed at random from my New Parts Stock Tin) and connected the FG's Red output lead to the Gate of the mosfet, and connected the CH1 Scope Probe to the Source of the mosfet. Nothing is connected to the Drain and there are no other components used. The FG's Black output lead and the Scope Probe Ground leads are clipped together. The FG is set to produce a positive (DC) pulse train of about 2V peak. CH1 is the signal at the Source pin of the  mosfet and CH2 is the signal coming from the FG, directly connected to the scope via a BNC patch cable (no probe). Note that the CH1 signal is _AC_ and is _in phase_ with the raw FG output. This indicates that the mosfet is acting like a capacitor, passing the signal from the FG on the Gate, through to the Source.

Certainly, the inductances of the mosfet and its connecting leads must be taken into consideration in any _true_ simulation of the mosfet's behaviour, but the leakage I have always been talking about is the Cgs leakage from Gate to Source due to the capacitances of the mosfet.

So, what.