[SOLVED] Drain-Source Diode And RC Snuber Filter In Paralleled MOSFETs/IGBTs

[mcmsat13]

I have two identical diagrams here with the placement of the drain/source diode and RC snuber filter, the only difference. I like presenting my question with a diagram because it makes the question a visual subject. These diagrams are for reference purposes only.

When I using a switch module, MOSFET/IGBT, because they are a kind of single device, you place one single diode across the drain and the source likewise a single RC Snuber circuit.

Now, when using multiple descrete devices in parallel, must the Drain/Source diode and the RC Snuber soldered across each individual device? Some people have asked me this question and I have no better answer for them because I lack full explanation to this theory.

Now my question is:


Is the circuit "A" the correct one or the circuit "B"?

Please I need full/fair explanation here as answering me "A" or "B" alone will not make much difference in my learning.

 

[FlapJack]

My personal opinion is, it depends. If you have only two devices to parallel and they are close to each other you could probably put one RC snubber between them.

As you add more and more devices, your snubber becomes less effective as the distance from it increases. The inductance of the PCB trace starts to add up and you will see higher voltage peaks as you get farther away from your RC snubber.

The frequency and switching speed will also make a big difference on how far you can be away from the RC.

In the end you would probably have to make a prototype and look at each FET with a good scope for each different case.

But if you had multiple RC’s you could probably make each one smaller and still have good snubbing and maybe get by with just a SMD resistor.

 

[BradtheRad]

Remember that it is common for mosfets to contain a body diode.

For this configuration, I believe you'll benefit more if you install snubbing networks across each of your primary windings.

The snubbers will not necessarily be identical to those in your schematic.

 

[mcmsat13]

Remember that it is common for mosfets to contain a body diode.


For this configuration, I believe you'll benefit more if you install snubbing networks across each of your primary windings.

The snubbers will not necessarily be identical to those in your schematic.

It is common for mosfets to contain a body diode.

Do you mean that the external diode is not needed?

PRIMARY WINDINGS, do you mean as in circuit B? In B, it is connected to the left and right drains and the ground. This is how I understand your expression. Am I right?

 

[BradtheRad]

PRIMARY WINDINGS, do you mean as in circuit B? In B, it is connected to the left and right drains and the ground. This is how I understand your expression. Am I right?

The center-tapped primary creates spikes, which go backwards through the power supply.

This is my simulation demonstrating how snubbing networks would operate, placed across the windings.

Without the snubbing networks, there would be more severe spikes going backwards through the power supply.

I'm not saying this is the optimum snubber for all inverter topologies.

Do you mean that the external diode is not needed?

Mosfets have an unpredictability about them. The internal body diode has its own fwd V. If it is greater than your external diode's fwd V, then current spikes will go through your external diode instead of the mosfet. This would spare the mosfet undue stress (I think).

 

[FvM]

Do you mean that the external diode is not needed?

Yes, very clearly. In addition, you can ask if the diode is needed at all. But that's a theoretical question because it's present in any MOSFET.

The dominant problem of the transformer push-pull converter is overvoltage after switch-off, caused by the transformer leakage inductance. You discussed a RC snubber, Brad suggested a RCD snubber. In any case the leakage inductance energy is burnt in the snubber resistor, and often additional losses are caused by the snubber. I think, the first check would be if the converter can be operated without any snubber. Then the leakage inductance energy will possibly force the switch transistors into avalanche operation, but recent MOSFETs have considerable avalanche capability and can bear this operation.

 

[mcmsat13]

Thanks Bradtherad, I am putting the pieces together now. Will the values be the same for 24V? Please what program did you use for this simulation?

Thanks

- - - Updated - - -

Thanks Bradtherad, I am putting the pieces together now. Will the values be the same for 24V? Please what program did you use for this simulation?

Thanks

 

[BradtheRad]

This is Falstad's animated interactive simulator. Free to download and use at:

www.falstad.com/circuit

It can export a link, containing an entire schematic.

Click the link below and it will:
* Open the falstad websie
* Load my schematic into the simulator
* Run it on your computer

https://tinyurl.com/me98lzu

You can see current bundles traveling in wires.

You can change values. Right-click on a component and select Edit.

Will the values be the same for 24V?

It depends on how much energy is in a spike coming from the transformer. The load is a major factor. If you change the supply voltage, but you draw the same amount of watts from the secondary, there is a chance you will not need to change snubber values at the primary.

For the capacitors you can emphasize their Farad value, or their volt rating.
They must absorb a spike (charge), then they discharge through the resistor.

The resistors need to be rated several watts. That is, if you are drawing substantial power from the secondary, say, 100 W.

 

[FvM]

Does the Falstad transformer model implement leakage inductance? Otherwise there's no meaningful simulation of snubber necessity and snubber action.

 

[BradtheRad]

Does the Falstad transformer model implement leakage inductance? Otherwise there's no meaningful simulation of snubber necessity and snubber action.

There is not a specific parameter for leakage inductance listed in the edit window for transformers (plain or tapped).

Nor do I see leakage inductance referred to in the Java source code.
This comment may be relevant:

// mutual inductance between two halves of the second winding
// is equal to self-inductance of either half (slightly less
// because the coupling is not perfect)

I confess I don't know how leakage inductance will manifest itself.

I simplified the schematic, in hopes to observe stark theoretical action.



Both windings conduct reverse current at turn-off. But only one winding generates a high V spike. (In theory.)

The mosfets conduct current upward at turn-off. This suggests the body diode is included in the mosfet model.

You may be correct in saying the simulation is not meaningful.

However I remember getting similar spikes from a transformer, when I tried to make a similar push-pull inverter years ago. I burned up a few mosfets. I had too many gaps in my knowledge, as to what kind of snubbing components to try.

 

[FvM]

I see that the Falstad simulator offers a coupling parameter for transformers, that's another way to specify leakage inductance, So the answer is, yes it does model leakage.