3

My high and low side gate measurements to GND show a tiny little shoot through spike.

Adding dead time, Zener bypasses over R(g) and from G-S, and TVS diodes does not do anything.

The load is 100K and it's running off 12V supply. Any ideas as to how to get rid of the spike? I'm assuming it's shoot through, but even giving it way more dead time simply just spikes it anyway. It's like it never fully discharges the gate.

enter image description here

enter image description here

MicroservicesOnDDD
  • 2,557
  • 1
  • 13
  • 48
ARMATAV
  • 603
  • 7
  • 19
  • How should anyone have any idea without schematics and bom? – PlasmaHH Apr 19 '15 at 21:42
  • 2
    Reduce load resistor to 100 ohms or less. You're getting capacitive coupling of your gate drive into your load. It's intrinsic to MOSFETs. With a much lower load, you won't see it. – WhatRoughBeast Apr 19 '15 at 22:56
  • Why would you drive a 100K load with an H bridge?? That's a bit extreme for such a tiny load. If you want bi-directional control over a 100K load, use an op-amp with rail-to-rail output and dual +-12V supplies.. – KyranF Apr 19 '15 at 23:21
  • Which trace is red, and which trace is yellow? Also, when taking a picture of your oscilloscope, please place the camera *in front* of the oscilloscope to minimize skew. This angle makes me feel nauseous. – Zulu Apr 19 '15 at 23:42

3 Answers3

2

Looks like dV/dt. When for example the lower FET is being held off, but the upper FET is driven on, current will be driven from drain of lower FET through Cgd into Gate. If the gate circuit impedance is high enough, lower FET will be turned on by induced voltage across gate circuit impedance. The impedance will include Rg (like R6, R8, R9, R10) and driver resistance (for IR21xx parts is often ~ 10Ohms) and the rest of gate routing including back to the source. In your circuit it looks like there is a round about path from gate to source on the bottom FETs. Also, looking at the rise time of the gate voltage, there may be too much resistance. Are you sure Rg is only 4.7Ohms? If so, maybe that's too much.

gsills
  • 7,133
  • 16
  • 22
  • Actually. Removed the resistance, removed that loop. Nothing changes whatsoever. Changing the load resistor to 100ohms or even nothing at all doesn't seem to change that spike. – ARMATAV Apr 20 '15 at 00:50
2

The bloody C8 and C12 were in completely switched sides of the board, and tied, therefore, to the wrong driver, meaning that whenever a driver's low side fired, the charge left on the bypass capacitor was built up on the bootstrap capacitor, and as soon as the 'switch' came, it would discharge the bootstrap momentarily for the wrong cap, meaning you get a spike at the beginning of the wave. At least I think.

The reason the yellow wave is slightly higher DC bias is because the bypass being on the wrong side, and thus, I think, undergoing a bit of hystereses due to the really long trace.

Unfortunately you guys weren't given the actual PCB layout by me, I thought I had it down, but silly mistakes like this will cost you.

ARMATAV
  • 603
  • 7
  • 19
0

Mostly the problem seems to be in IC2. Ground connection is not clean, so just try to remove IC2 and connect solid thick connection between source of bottom MOSFET and gnd. You can use two-three connecting wires for same and will see the difference significantly.

All the best.

Electroholic
  • 729
  • 9
  • 34