Half-Bridge Low-Side Gate Oscillation When High-Side MOSFET Switched On

Hi all,

After many h-bridge revisions (4 in total), I have almost got it working. The one remaining problem that I can't seem to solve is that the low-side MOSFET experiences oscillations when the high-side MOSFET is swtched on. I think this is causing shoot-through and causing the power rail voltage to sag momentarily. The bulk capacitors on the power rail where heating up, supporting this theory. The MOSFET's also get un-naturally warm once you start drawing more than an amp through them (by my calculations, they should still be dead cold at this point). These oscillations get worse as the current increases, so the shoot-through is more severe at larger currents.

I am driving the brushed 36V, 800W at the outputs.

This oscilloscope image shows it better than I can explain it.



The blue trace is the low-side MOSFET gate voltage, shown turning off from 12V to 0V. The yellow trace is the high-side MOSFET switching on momentarily later, going from 0V up to Vdd +12V (around 55V in this case). Notice the low-side oscillates when the high-side is turned on.

This is the schematic of the half-bridge



The driver IC is the LM5104M (sorry for it not being on the schematic).

Series resistors on the gate of the MOSFET's were present (10R), but they made the problem worse (they lengthened the period of the oscillations), so I removed them.

Also, increasing the dead-time or the boot-strap capacitance (I doubled it from 420nF to 840nF) did not have any effect.

The circuit is made on strip-board, and I have tried to make the gate wires as short as possible. The image below shows the board in-case its relevant. The SMD IC in the middle is the driver IC, the two TO-220 packages bolted to the heatsink are the MOSFET's (the other two are schottky diodes), and that plug-in module is a 35-75V to 12V buck converter for powering the IC from the 60V rail voltage.

The specific shape of the waveforms looks pretty strange. The lower FET gate voltage is actually pushed negative, just as the upper gate voltage is peaking... Also you can see on the upper gate voltage that despite having a fast rise time, it then droops down quickly and slowly recovers to its steady state. I think this implies that the DC bus voltage is drooping briefly.

I don't see any bypass capacitors in your picture. It's critical that you have good high frequency bypassing (large film or ceramic capacitors) right on the FETs. I think you made a reasonable decision by removing the 10ohm gate resistors, but that will also increase the rise times on the gate, which can cause harm. Slowing down the rise time of the upper gate voltage with an RD network may help, while still providing a low impedance during turn off. Also I would recommend looking at the half bridge output for excessive ringing. If there is, you'll want to add a RC snubber (adding the high frequency bypass caps will help also).

The bus voltage does drop, I was trying to work out what was cause and effect. If the low-side FET gate voltage rose and then turned on, shoot-through would occur which would drop the bus voltage. However a drop in bus voltage due to other reasons (such as if the battery was not powerful enough to supply enough current to the motor), could cause the driver IC to stop working momentarily, causing the oscillations on the gate. To rule out the latter, I powered the driver IC of a separate 12V battery rather than the 60V-12V buck converter. However this didn't fix it. But yes, the bus voltage does drop momentarily at that point, sometimes getting as low as 10V!.

By having bypass on the FET's, are you referring to connecting them across drain-source? If that's what you mean, then no, I don't have any caps there. The only bypass caps I have are on the bus rail, and they are not close to the FET's at all.

gbmhunter said:

The bus voltage does drop, I was trying to work out what was cause and effect. If the low-side FET gate voltage rose and then turned on, shoot-through would occur which would drop the bus voltage. However a drop in bus voltage due to other reasons (such as if the battery was not powerful enough to supply enough current to the motor), could cause the driver IC to stop working momentarily, causing the oscillations on the gate. To rule out the latter, I powered the driver IC of a separate 12V battery rather than the 60V-12V buck converter. However this didn't fix it. But yes, the bus voltage does drop momentarily at that point, sometimes getting as low as 10V!.

Click to expand...

Shoot through probably won't hurt the gate driver IC (unless the FET fails first due to overcurrent).

By having bypass on the FET's, are you referring to connecting them across drain-source? If that's what you mean, then no, I don't have any caps there. The only bypass caps I have are on the bus rail, and they are not close to the FET's at all.

Click to expand...

Not a capacitor for each FET, but one on the DC bus, right where it connects to the FETs. This is likely part of your problem.

Shoot through probably won't hurt the gate driver IC (unless the FET fails first due to overcurrent).

Click to expand...

It could if it causes the bus voltage to drop below the minimum operating voltage of the IC (around 10V in my case).

Not a capacitor for each FET, but one on the DC bus, right where it connects to the FETs. This is likely part of your problem.

Click to expand...

O.K., I'll give it a go. Also recommended from some friends is adding snubbers across the drain-source of each FET. They think a likely cause in the fast switch-on of the high-side MOSFET causing oscillations to ripple through and appear on the gate of the low-side FET.

gbmhunter said:

O.K., I'll give it a go. Also recommended from some friends is adding snubbers across the drain-source of each FET. They think a likely cause in the fast switch-on of the high-side MOSFET causing oscillations to ripple through and appear on the gate of the low-side FET.

Click to expand...

Yes, that's one way of putting it. Hence my recommendation to use an RD gate network to slow the turn on of the FETs while keeping the turn off strong and fast. And having good bypassing on the half bridge is effectively a type of snubber network, since it effectively decreases the inductance in your DC bus which causes ringing and droop. Additional RC snubbers on the H bridge output may still be necessary though (but usually isn't if your layout is good).