1EDS5663H

[Ahmed]

I am using 4 GaN switches "IGT60R070D1 " in full bridge configuration in a PSFB DC DC converter. I am using the gate driver "1EDS5663H". The configuration is shown below.

The gate source voltage is also shown below

When taking a closer look on the gate to source voltage I notice this part here

So I wanted to ask about the reason for that.

Also I am designing this converter for ZVS (soft switching), and I noticed that if the deadtime is long the FET sometimes truns on unexpectedly in the spice simulation. I wanted to ask if the gate driver IC is capable in reality to hold the voltage at zero. Is there a possibility to use a bipolar supply to have a negative voltage all over the turn-off period?

[Daisy_T]

Hi Ahmed,

The 1EDS5663H  is an isolated driver IC,  this means the input and output are isolated.  In your circuit, the gound is the same for both input and output. Please modify the ciruit.

And we also recommend you to use the below application note as a reference during design:

https://www.infineon.com/dgdl/Infineon-ApplicationNote_EvaluationBoard_EVAL_1EDF_G1_HB_GAN-Applicati...

Best Regards,

Daisy

[Ahmed]

Hi Daisy,

Thank you for your reply. If you mean by the same ground like the one at the transformer primary, that was just to fix convergence problems.

I removed it and I still see the same behavior like shown below

And is there a solution to my second question?

Kind regards

Ahmed

[Xiangrui]

Hi Ahmed,

I have some advice for your circuit. 

Firstly, you can try to add Rg for your drive circuit. It helps to suppress spikes of Ugs.

Secondly, your Gout waveform is not correct. I don't think GaN can be turned off completely with this Gour. It might be the reason for unexpected turn on.

Finally, Can you share more details for your simulation. For example the PWM duty, frequency and failed example when dead time too long?

BR,

Xiangrui

[Ahmed]

Hi Xiangrui,

Thanks for your reply. Rg is in the circuit. The RC network recommended is used.

The switching frequency is 100 kHz. duty cycle is 50%.  This is the Vgs waveform.

My circuit uses soft switching and increasing the deadtime helps me achieve ZVS. But because the negative turn-off is only available for a short time. Increasing the deadtime leads to unexpected turn-ons. So I wanted to ask if there is a possibility to use a bipolar supply.

[Xiangrui]

Hi Ahmed, 

"IGT60R070D1 " is actually a normal OFF switch, and this driver can also provide negative voltage as a bipolar driver. After that Vgs is switched back to zero to ensure identical conditions for the next switch "on" event and to minimize losses from diode operation.

However, as you said, the time with negative voltage should be longer than dead time to switch off GaN, because the Vth is only about 1V.

So could I know how long is your dead time?

BR,

Xiangrui

[Ahmed]

Hi Xiangrui,

Thanks for zour reply. I am useing an adaptive delay. So, the deadtime ranges from 70 ns to 250  ns.

I am also still curious about this part here

Kind regards

Ahmed

[Xiangrui]

Hi Ahmed,

you can add a RC circuit in TENG Pin to extend the interval with negative voltage by turn off. In my simulation it's enough to use R=50k and C=10.8pF for 500ns dead time.

I didn't see this ripple which is shown in your pic in my simulation. But I notice that after changing your circuit it looks better than before. Could you share your current circuit? And does it happen each time when turn on or just with first pulse?

BR,

Xiangrui

[Ahmed]

Hi Xiangrui

Thank you very much for your help. I added a 50K resistor and that fixed my problem with the deadtime.

But I still have the other problem mentioned earlier. These are the waveforms for the voltage and the current

Kind regards

Ahmed

[Xiangrui · Accepted solution]

Hi Ahmed,

Refer to following picture in application note. I think it's normal to have this waveform with last and dead time.

BR,

Xiangrui

[Ahmed]

Thanks, Xiangrui!