Hello, I am trying to make a simple inverter pcb but my output is a square wave instead of the expected spwm. I suspect that some capacitance or inductance is built up due to the pcb layers that alters the gate signal. My traces going to the mosfet gates are quite wide (1.5mm) and their length is approximately 25mm.

Picture 1 is the output of an earlier version of the board. This is close(r) to the expected result.


Picture 2 is my newest version of the board.


Note that I went from a 2 layer pcb to a 4 layer pcb and the traces from the microcontroller to the mosfets are in the inner layers. Also, the earlier board had thinner traces to the mosfets' gates.

Any advice is appreciated. Thanks.

 

Schematic?

 

Schematic?

Sorry about that. The full schematic is pretty messy (I am still bad at them) so here is the part you are probably interested in. Both inputs are from a 5V microcontroller. There is also another copy of it for the reverse polarity. N and L are the outputs.

 

If nothing is overheating, I suspect that the gain set in the control circuit is wrong, or that the transformer ratio is wrong.
The entire schematic would be needed, not just the power stage.

 

The entire schematic would be needed, not just the power stage.

Here you go.

 

What PWM frequency are you using?
A 4mA output From a microcontroller is going to take 16us to charge the 66nC gate charge of a IRLZ44.

 

Wider traces and increased length can introduce more capacitance and inductance. It can distort high-frequency signals.

 

What PWM frequency are you using?

I am using (a custom version of) this code. Connecting the microcontroller (not connected to the rest of the board) to the oscilloscope and manually counting the rising edges of the spwm signal I get a period of 100μs so 10kHz.

A 4mA output

How did you get to that value? I was under the impression that the ATmega328P can do up to 40mA per IO pin (under abs. max. ratings on the datasheet).

 

Wider traces and increased length can introduce more capacitance and inductance. It can distort high-frequency signals.

I can't really change the length but I can make them significantly thinner. Will moving them to an outside layer help?

 

I am using (a custom version of) this code. Connecting the microcontroller (not connected to the rest of the board) to the oscilloscope and manually counting the rising edges of the spwm signal I get a period of 100μs so 10kHz.


How did you get to that value? I was under the impression that the ATmega328P can do up to 40mA per IO pin (under abs. max. ratings on the datasheet).

4mA is a typical value for microcontrollers, even at 40mA it is still far too weedy for directly driving a MOSFET gate. They need at least half an amp to get switching times in the tens of nanoseconds.

 

4mA is a typical value for microcontrollers, even at 40mA it is still far too weedy for directly driving a MOSFET gate. They need at least half an amp to get switching times in the tens of nanoseconds.

Interesting, so I will need a mosfet driver?

Also, how does that explain the different oscilloscope reading?

 

Interesting, so I will need a mosfet driver?

you certainly will. No question about it.

Also, how does that explain the different oscilloscope reading?

You could be suffering from poor rise and fall times.

 

you certainly will. No question about it.

Ok, thank you very much for the help