Discrete MOSFET vs MOSFET Module (Need Help)

[Glenn0010]

Hi all so I need some expert advice!

I have made 2 different BLDC Motor Controllers they have the exact same functionality and code except for one difference.  In one I am using discrete MOSFETs and in the other I am using a MOSFET module the MTI85W100GC.

I have tested them under the same conditions. 50V 20A phase current. I have probed over the high side mosfet using a ground spring to measure my switching transients both under the exact same conditions in the exact same way.

But there is a massive difference in the results

Here is the first one which is 'correct'



An here are the results every thing is within reason





Here is the second version and I am probing across the two points circled in red.



And here is the nasty result



The transient on turn off which is typically the worst case, is better than that of the first version.

However, As can be seen there is a HUGE negative transient on turn on which makes 0 sense to me. I do not know why this is. Could it be that since I am probing over the  GND Trace of the module it is emitting some noise into my measurement or could it be something else. I think if that negative transient was really present the magic smoke would have come out

I also probed from ground to phase (i.e. lowside FET) therefore over no traces, and got a clean waveform with no negative transient so that is also pointing to the fact that since I am probing over the ground trace it may be emitting some noise

Hope someone can help me on this because I am stumped!!!

Cheers

[T3sl4co1l]

Probe from GND to GND and show the waveform.  Yes, in the same location.

Tim

[Marco]

You could try putting a ferrite on the probe cable close to the probe. The floating DUT's ground plane is moving 50V relative to earth during switching because of the way the probe is connected, its parasitic capacitance to earth is being charged/discharged through the ground lead.

[Glenn0010]

Probe from GND to GND and show the waveform.  Yes, in the same location.

Tim

Do you mean from phase to gnd (that would be lowside)? I have done that but unfortunately the scope did not seem to save my data. What I had ws a pretty large overshoot of around 88V and an undershoot of -2V but nothing anywhere near the -40V

Post the part number of your original MOSFET.

The part number of the discrete MOSFET is IPB039N10N3G. They are very similar to that of the Module

You could try putting a ferrite on the probe cable close to the probe. The floating DUT's ground plane is moving 50V relative to earth during switching because of the way the probe is connected, its parasitic capacitance to earth is being charged/discharged through the ground lead.

I am guessing you mean a ferirtee bead ground the ground spring? If I am honest I have never worked with them and do not really know their effects. Also I do not have any available. If I use one could it deistort my signal in a way that I would not get the true reading during MOSFET turn off for example?

[Marco]

No, I mean taking a clip on ferrite and winding the probe cable through it one or two times, putting it close to the probe.

[Glenn0010]

I'll  see if I can find one and give it a shot

[Glenn0010]

I have a dead time setting of around 333ns.
And rise and fall time of around 250ns.

So basically  the difference in the body diode could be an issue. But wouldn't the -40V transient kill the fets?

[T3sl4co1l]

1. What's your dead time setting? Does it match Td/Trf of your power devices with a healthy margin?
2. With your MOSFET module, it has 1300nC Qrr at 50V, 80A, while the OptiMOS3 has 1/10 of that Qrr at 50V, 100A. That should make a huge difference especially if your dead time is too long.

More like 2-3 times?  Where did you see 1300?

Tim

[T3sl4co1l]

I have a dead time setting of around 333ns.
And rise and fall time of around 250ns.

So basically  the difference in the body diode could be an issue. But wouldn't the -40V transient kill the fets?

No, 50V supply + 40V spike = 90V, rating is 100V, probable actual breakdown 110-120V.

The 40V spike does not actually appear across a MOSFET diode, it does however appear across the ample stray inductances in the circuit, which you are probing across.

Tim

[T3sl4co1l]

More like 2-3 times?  Where did you see 1300?

Datasheet. Qrm=1.3uC.

Oh yeah the diode...

Let's see...... the IPB's data is unreasonable, as it's measured at 100 A/us.  Peak recovery current is not given.  The MTI's is at 1500 A/us, so the charge is much larger.  t_rr is similar, though we should expect the MTI performs more poorly at the same conditions.

Tim

[Glenn0010]

So what would be the best course of action to correct such an issue? would you try to reduce the dead time or add a better body diode in parallel with each fet?

The previous values of 240ns rise and fall times were for the discrete fets

For this Module the rise time is aroun 140ns and fall time is around 170ns

I have some magnified view for you guys with the time values maybe it is helpfull:

Turn Off




Turn on:



Some nasty ringing in there