Worried about short-circuiting H-Bridge using PWM

[HBRIDGE]

Basically, I'm trying to implement a simple single-phase inverter. I have a sine wave generator, a triangle wave generator and I have created PWM signals out of them using some op-amps and comparators (I haven't yet done a schematic, just a mix of different schematics I've seen online that I've put together to work for me)

I've created the PWM signals, or so I think (tri-level PWM) and before I go connecting this to the MOSFET's I just want to be sure that it's OK, because I have a feeling I'm going to end up short-circuiting the supplies. So if you consider the four H-BRIDGE switches like this:



Whilst S1 is having PWM input to its gate, S4 is ON, then whilst S3 has PWM input to its gate, S2 is ON. The lower switches (S2 and S4) are controlled by a square wave signal, so it's either ON or OFF. Only the upper switches have PWM.

Here is the waveform for the upper left-hand side switch and lower right-hand side switch, CH1 (yellow) is upper and CH2 (blue) is lower, there are sections where there will be short-circuiting (roughly 100us), is this acceptable or have I got this completely wrong? I'm thinking that when S2 is OFF, S1 should also be completely turned off, why would I be getting a PWM signal whilst it's turned off? See below



Here is a zoomed-out version:



You can see that when S1 is turned on and has PWM, S2 is completely off, that's fine! But when S2 is ON, S1 is off, but it starts having PWM in the middle for a bit.

Here is my trilevel PWM, the reference sine wave superimposed on it.





Any advice appreciated, I'm hoping someone can tell me why, I know it's hard without a schematic, but I haven't drawn one out yet.

 

[kappa_am]

I am not sure I have understood your meaning. please add your pictures. You cannot keep below switch ON or OFF, and switch upper switches.

 

[HBRIDGE]

Hey, here's a schematic that I followed, this is not mine, mine is very similar except for some differences which I don't think will make too much of a difference. I have annotated as necessary, it's not very complicated:

1) A bubba oscillator make a reference sine
2) A triangle wave generator
3) Creating a modified triangle square wave for Trilevel PWM
4) Comparing sine with modified triangle square
5) Comparing square with DC offset voltage

So there are 2 PWM outputs and 2 square signal outputs, the people who made this schematic have appeared to make it work, so I'm not sure why you're saying we can't use a square wave signal

Any advice

 

[BradtheRad]

There is the issue whether you plan to put an inductor in the output, to smooth the SPWM.

Do you plan to make the output see high impedance between pulses of current? This is likely to cause high voltage spikes from the inductor. You will need a robust snubber/smoothing network.

This simulation shows the switching method which seems to give the least problem in smoothing the SPWM. Notice one end of the output is grounded, while the other end is alternately connected to ground and supply+.



The op amps create the concept of an H-bridge, to simplify the layout. Their inputs receive similar waveforms as you have in your schematic. It is important to pay attention to the range of input. If you apply a positive and negative waveform, then your op amp must have a bipolar supply.

Notice the op amp output alternates between two states: (a) sourcing and (b) sinking current. The output does not go to high impedance.

My switching action may be difficult to implement. If you use mosfets, then the body diode may be sufficient to do the job. Or, consider adding external diodes.

 

[chuckey]

In your wave forms the PWM is not running through out the square wave period, i.e. you will be producing bits of a sinewave during each AC half cycle.
Frank