MOSFET H-bridge switching spikes

[earckens]

Downspikes of about -2V appear during MOSFET switch time on the rectifier output, no matter what value C2 (up 1.000uF) or R4 I tried. Is there any way to get rid of these spikes on the rectifier output?

 

[dknguyen]

Do these spikes appear on the rectifier input?

You might need schottky diodes in antiparallel with each MOSFET...though I didn't think they would be needed since your H-bridge load is not inductive. 2V is also a pretty small fixed number but it seems too large to be the voltage drop across the rectifier.

 

[earckens]

Schottky or high speed diode anti-parallel to each MOSFET.

Drain to source, antiparallel: no change; the load is resistive.

 

[dknguyen]

Yeah, that didn't seem like it would be the cause. What's the voltage drop across any one of the diodes in the rectifier?

What is this circuit anyways? Is it just an intermediary test circuit? Because the rectifier on the output of the H-bridge defeats the purpose of the H-bridge.

 

[earckens]

Forward voltage drop on rectifier diodes: 0.7V.
The reason for the rectifier is to serve as input to an optocoupler. The H-bridge serves to send a square wave ac signal to sensor probes. When open (no liquid), the full voltage is forwarded to the rectifier, subsequently to the optocoupler. When closed (ie sensor probes shorted, liquid present) no voltage is present and optocoupler not activated.
When the optocoupler is not activated it serves to send signal to controller to activate whatever. There is no issue with the spikes when the optocoupler is not activated, so not to bad in that respect, but I would like the signal to be clean too when the optocoupler is activated, if only not to perturb the anxillary components.

 

[dknguyen]

Forward voltage drop on rectifier diodes: 0.7V.
The reason for the rectifier is to serve as input to an optocoupler. The H-bridge serves to send a square wave ac signal to sensor probes. When open (no liquid), the full voltage is forwarded to the rectifier, subsequently to the optocoupler. When closed (ie sensor probes shorted, liquid present) no voltage is present and optocoupler not activated.
When the optocoupler is not activated it serves to send signal to controller to activate whatever. There is no issue with the spikes when the optocoupler is not activated, so not to bad in that respect, but I would like the signal to be clean too when the optocoupler is activated, if only not to perturb the anxillary components.

What are you feeding the opto-output too? It might be easiest just to debounce the signal with hardware or code rather than dealing with the spike because I'm not sure where that's coming from, but you had something similar with 2V in an earlier incarnation.

 

[earckens]

What are you feeding the opto-output too? It might be easiest just to debounce the signal with hardware or code rather than dealing with the spike because I'm not sure where that's coming from, but you had something similar with 2V in an earlier incarnation.

Here is the preliminary schematic. Part of interest is upper right.

 

[MikeMl]

Have you checked the shoot-thru current on the h-bridge. That is a very poor H-bridge circuit because with this very simplistic (naive) method of driving the gates there is a huge potential for having the upper and lower fet on each side of the bridge on at the same time.

 

[earckens]

Yes, has been checked: it actually is more of a problem of microsecond delay between shutting down one side and starting up the other side.

 

[dknguyen]

Yes, has been checked: it actually is more of a problem of microsecond delay between shutting down one side and starting up the other side.

Maybe just throw a clamp diode across the output of the rectifier just to make sure the spike isn't damaging and then debounce the thing to deal with the blip in the signal. A schottky should clamp the undershoot to -0.3 to -0.5V.

 

[MikeMl]

Yes, has been checked: it actually is more of a problem of microsecond delay between shutting down one side and starting up the other side.

First, convince me that this is even an H-bridge. After the first edge from the 555, it never changes state again. It looks more like a latch to me.

 

[dknguyen]

First, convince me that this is even an H-bridge. After the first edge from the 555, it never changes state again. It looks more like a latch to me.
View attachment 111649

It is a latch, but this was pointed out previously (in another thread) and he's added pull-up resistors to the drains of the two lower MOSFETs which solved the issue.

 

[MikeMl]

It is a latch, but this was pointed out previously (in another thread) and he's added pull-up resistors to the drains of the two lower MOSFETs which solved the issue.

Then why the hell didn't he just add to the existing thread instead of starting a new one?

So post the "real" schematic!

 

[dknguyen]

The why the hell didn't he just add to the existing thread instead of starting a new one?

Same circuit but the thread was started for a specific issue in that circuit and once that was solved he didn't want to muddy things up by mixing everything into the same thread.

 

[MikeMl]

Same circuit but the thread was started for a specific issue in that circuit and once that was solved he didn't want to muddy things up by mixing everything into the same thread.

How could I be more confused?

I'll bet that if he posts the real h-bridge schematic, and we simulate it, it will have shoot-through up the whazoo...

 

[earckens]

First, convince me that this is even an H-bridge. After the first edge from the 555, it never changes state again. It looks more like a latch to me.
View attachment 111649

Measured across the load.

 

[earckens]

Maybe just throw a clamp diode across the output of the rectifier just to make sure the spike isn't damaging and then debounce the thing to deal with the blip in the signal. A schottky should clamp the undershoot to -0.3 to -0.5V.

Clamp diode: when no voltage is present, no spikes. When voltage is present, the spikes are to ground.

In my opinion the spikes are due to MOSFET switching delays, in effect creating very short moments of no conduction across the load. What puzzles me is that a cap across the rectifier output cannot smoothen these downward spikes.

 

[MikeMl]

...What puzzles me is that a cap across the rectifier output cannot smoothen these downward spikes.

Where is the scope probe ground clip? You have a floating load. If the scope chassis is grounded to the AC line, then the only way you could read the voltage across the filter capacitor is with a true-differential measurement.

 

[earckens]

What is strange is that when the rectifier output is high, hence the optocoupler is conducting and pin 5 is to ground, that spikes are measured at pin 5 of the optocoupler.

 

[earckens]

Where is the scope probe ground clip? You have a floating load. If the scope chassis is grounded to the AC line, then the only way you could read the voltage across the filter capacitor is with a true-differential measurement.

Interesting.

However, when the scope ground clip is connected to the circuit ground then there should be no issue: the power supply is not connected to ground, hence floating, and will follow the scope ground. No?