MOSFET SWITCHING

This might provide a clue...

https://www.homemade-circuits.com/sg3525-pure-sinewave-inverter-circuit/

Dear Guru Solar Eagle,

Thank you very much for forwarding the link. Unfortunately, the problem ( not turning OFF after the gate drive is removed and turning ON before the gate pulse is issued) is not addressed anywhere in Mr. Swagatham's post.

Kind regards,

Job Thykkoottathil

I am not a ET, but I know that these switches don't react instantaneously, there is always a delay...the timing can of course be adjusted to suit the application, what problem, if any, is this delay causing?

https://talema.com/gate-driver-introduction/

https://www.ti.com/lit/ml/slua618a/slua618a.pdf?ts=1650055662647&ref_url=https%253A%252F%252Fwww.google.com%252F

Dear Guru,

I realize that a MOSFET with a large turn OFF time may cause the trasistor keep conducting fully or partially for a small period on turn OFF. But why the transistor start to conduct partially even before a gate siginal is applied?

Kind regards;

What is your oscilloscope? Are the two channels being swept simultaneously, or alternately, like in the old days of tube oscopes. I agree the transistor should not be turning on before the gate signal drives it to. So I would question the scope image.

I have two scopes. One is a Teknonics TDS220, which is an old one and the other one is a Daniu ADS 1013 D. I am not sure whether the chanels are being swept simultaniously. I believe that thay are not. I am using chanel Ch1 as trigger chanel. Both scopes give the same result.

Regards;

Well the first thing I would check is if there is an electric signal getting to the gate ahead of time, if so then trace it back to it's origin...

Thank you Guru Rexter;

I will apply a small resistive load across the secondaries and see what happens.

Kind regards;

I tried putting a small load of 1K (15 mA) across both secondary windings but did not make any difference.

Kind regards

try putting a small test resistor(1 ohm?) in the Source leads (referenced to ground) to act as a shunt to measure current. I suspect you will not see the current rise until the FET is turned on.

Hi Guru,

Do you mean inserting a small value resistor between source and ground?

Kind regards;

yes

GA

In order to prove this, one can disconnect the transformer windings temporarily and substitute resistors instead, thereby removing the possibility of an <...inductive kick...>. If the phenomenon goes away, then "Robert is your mother's brother".

Not sure but, I think that the slow turn off is because of the capacitive nature of the gate.

The apparent advanced turn on is not because Q1 is on: think about what is happening to the Q2 half of the primary. Whilst Q1 is off: D1 is effectively open circuit so that side of the winding will simply follow the other half of the primary.

Does that make sense?

The MOSFET operate properly. The problem is with your interpretation of the waveform. When the first MOSFET turns off (at the gate ~0V), the drain voltage stays high due to the inductance of the transformer. You would see the drain current go to zero if you measured it. The transformer core energy is discharged in the free wheeling diode of the other transistor. The drain voltage would eventually go down to zero if the dead time was about a millisecond and the current would decay to zero. But it is not the case here. The current is eventually diverted to the other transistor when it turns on which brings the drain voltage of the first MOSFET to zero. Everything is working as expected. I am even surprised by how clean the waveforms are. Nice work minimizing the parasitic components.

I agree with your assessment. Looking at the drain voltage does not translate to to current. The relationship with G2 could be mapped and the actual drain current measured. Without a load on the inductive components there will be a strong inductive coupling between the two windings. When Q2 is turned off the inductive kick will free wheel through the parasitic body diode and induce a negative potential on the other coil.

With just a 2 channel scope you could watch what happens with G2 and D1, and then add an output load so that you influence the L/R time constant.