DC Motor Controller Problems

[Rapidrory]

Hi guys, I've had some strange issues with my latest motor controller project.

The controller has a single H bridge with 3 IRFS7530 MOSFETs per leg. The issue is that when the controller is running in a particular direction above about 2 amps, the unused low side FETs for the other direction start to heat up significantly. This was obviously a shoot-through issue, but the cause of the shoot-through seems to be caused by the capacitive coupling between drain and gate within the FETs themselves. Basically, there is a square wave being applied to the drain of the low side FETs by the high side FETs, and this is being coupled through to the low side gates. This causes the low side FETs to partially switch on in a resistive state, causing them to rapidly heat up. The picture below helps illustrate the issue:

and here is the signal measured at the gate the low side FETs

The spikes are about 3.5v at peak.

Weirdly however, even if the gates are connected to 0v with a 15ohm resistor, the voltage spikes are still there, and even if the FET gates are shorted directly to ground the FETs still heat up. This isn't just a case of having damaged FETs; I have made 3 different FET boards with IRFS7530s, IRFS7534s, and IRF1405s, and the issue still occurs across all 3.

Nothing I've tried has reduced the heating to an acceptable level. I know it's not a fundamental issue with the concept as I've seen 2 different commercial ESCs with this FET layout with similar FETs. Just wondering if anyone had come across something similar and if so any advice on fixing it?

Below are some details of the ESC in question:

FET Schematics

FET Driver Schematics

Any help would be greatly appreciated!

 

[TerryBarnaby]

I suspect you are seeing issues due to circuit inductances and the very fast switch times of the MOSFET's in some way. I note your MOSFET gate drive chip is quite a distance from the MOSFET's themselves on a separate board and possibly rely on a shared ground pin across the boards. This can cause problems. I have seen voltage transients of 30 Volts across 20mm of 10mm wide PCB track taking the current to a MOSFET source pin when switching 40 Amps!
This can inject spikes into the SOURCE GATE circuits. Here is a rough example where R1 is the motor (probably inductive as well) and L1 is inductance of a power track to a MOSFET to (hopefully!) show the issue:

See the bit on transients in: http://www.greenpower.beamweb.co.uk/groups/electronics/BuckBoostController/index.html

Now if something like this is causing the issue, how this relates to your circuit would take some working out. Some random points/ideas:
1. Where you earth your scope probe makes a huge difference. To see the voltage on a MOSFET's gate wrt to its source you have to connect directly to those points and often with a twisted pair of wires. Your scope could be seeing a spike across some inductance in the circuit or magnetic pick-up on the scopes ground lead. Ground is not Ground

2. Your gate drive IC may not be able to switch hard and fast enough to drive the high capacitances of the MOSFET's and keep the gates under control. The input capacitance of those MOSFET's are about 12nF each (36nF in total) and as you say feedback from the drain and source pins to the gate will occur through this capacitance. Your gate drive IC has to charge/discharge and hold the voltage on the gates through the 15 Ohm resistors with spikes occurring through these capacitances.

3. Your gate drive IC should drive between the source and gate pins closely. This means connecting the drive chip closely to those pins both the gate and ground (or S for high side). On the low side especially, the ground from the drive IC has to connect close to the MOSFET chip and not somewhere such that it passes through tracks shared with the high drive currents otherwise large spikes will be injected in the GATE SOURCE circuit (normally causing fire emitting transistor syndrome )

4. Your main power storage capacitors are electrolytic and thus will have highish impedances to the high frequencies present in switching. Some smaller 100nF film or ceramic capacitors placed across the power lines closer to the MOSFET's would help.

5. The motor load is probably inductive. This will generate voltage spikes on the switching that can be injected. A snubber circuit (series resistor/cap) across the motor terminals on the board could help.

6. Actively driving the low side MOSFET in anti-phase, with finely tuned timing to stop feed through, could help (Will help dissipate motor spikes).

However, I'm not sure why you see the heat up when the GATE is shorted to the SOURCE though. Was this connection close to the MOSFET ?
Anyway some ideas.

 

[Njay]

First thing is Terry's point one.

Then I would remove FETs leaving only 1 per leg and re-test (would be easier if you had 1 pull-down resistor per FET).

 

[TerryBarnaby]

Duh, it just clicked why your low side MOSFET's are heating up, you are not currently driving them, so this is normal and to be expected.

I presume you are driving a motor with that 20 Amps. A motor is highly inductive. When the high side MOSFET's switch off the inductance causes the motor current to continue to flow. The potential on the motor pin will fly down negative but will be arrested by the low side MOSFET's internal source/drain body diode (thankfully or the huge voltage generated would blow the MOSFET's). This continues to pass the 20 Amp current across its 0.7 to 1V drop diode. This current will drop of with time, but, depending on the inductance and your switching frequency this current might not drop off by much during the half cycle. So you would have about 20 Amps flowing through the three low side MOSFET's for half the cycle. If these diodes drop 0.8 Volts, with this level of current, then this produces 8 Watts of heat.

To reduce the heat generated you will have to switch on the low side MOSFET's during the high sides off period, with an appropriate dead band to stop shoot through. The voltage drop across the low side MOSFET's will now be low during the high sides off period and the power dissipated by them much lower. I think your spikes were probably due to scope probe ground positioning. Mind you all of the above still applies especially 3. If the gate to source voltage exceeds the MOSFET's limits (about 20 Volts) for more than a few micro seconds it will blow holes in the MOS gate. This leads to the "random" fire emitting transistor failure.

 

[Njay]

Jeeez, I think you're right Terry, he seems to be doing asynchronous rectification, which definitely has the other FET heat up as hell.