0
\$\begingroup\$

I am developing the BLDC motor controller board and I have problems with low-side gate driving. I am using MC33035 BLDC motor controller. PWM for speed control is fed into low-side MOSFETs. High-side MOSFETs are used for commutation. I have one ground-referenced shunt for current protection. I found out that low-side MOSFETs have oscillations on turn on and I can't get rid of them. (High-side MOSFETs don't have such problems)

30Ohm gate resistor

I tried to increase the gate resistor (30Ohm, 56Ohm, 100Ohm, 200Ohm)- did not help. I tried to put ferrite in series with the gate resistor it helped a little, but not enough. I tried to put a capacitor between the drain and the gate. I tried 470,900,2200pf capacitors, but this did not help. I tried to put the capacitor between gate and source, but this did not help.

And I found out that oscillations increase as PWM is fed to low-side MOSFET. First PWM edge does not have gate oscillations.

Maybe my driver is a bad choice? Because all low-side gate PWM currents have to go through a single shunt, although I have a solid ground plane, maybe parasitic elements are just too big and cause oscillations.

But maybe you guys know what can I try to get rid of this oscillations.

Schematic of one leg:

enter image description here

Top side PCB: enter image description here

UPDATE 1: I tried to decrease the gate resistor value, but oscillations become worse.

100Ohm + Ferrite enter image description here

10Ohm + NO ferrite enter image description here

\$\endgroup\$

2 Answers 2

2
\$\begingroup\$

It's likely caused by drain-gate capacitance and you won't see it on your upper N channel MOSFET because its drain is connected hard and fast to your DC bus voltage. The lower side N channel MOSFETs have their drains at the switching node and, under these circumstances, when they start to switch on, the drain voltage rapidly starts to fall and that action, in turn, hits the gate (via the inter-capacitance of the device, and causes the gate voltage to reduce (and sometimes oscillate) or wobble a bit.

If you reduced R17 down from 56 ohm, you might see an improvement provided that the MOSFET driver chip (MCP1402) can handle the peak current into the gate changing that rapidly at the switching point.

Basically it's miller capacitance and negative feedback.

Here's a picture of MOSFET turn on transition with insufficient gate drive: -

enter image description here

Improved response with bigger drive capability: -

enter image description here

Pictures from here.

It's also not helped much by having the shunt resistor in the source circuit - as soon as drain/source current flows, the shunt voltage rapidly rises and slightly switches off the lower N channel MOSFET. It's probably a bit of both things in reality. Both are negative feedback actions.

\$\endgroup\$
3
  • \$\begingroup\$ Isn't it called Miller capacitance? \$\endgroup\$
    – fifi_22
    Nov 30, 2020 at 15:09
  • \$\begingroup\$ If the shunt resistance is for overcurrent protection, the source issue (last para) issue may be helped by a parallel C across the shunt : C such that RC is long compared with the oscillation period and short compared with a PWM cycle. \$\endgroup\$
    – user16324
    Nov 30, 2020 at 15:15
  • \$\begingroup\$ @fifi_22 that's what I wrote - a couple of lines above the embedded pictures. \$\endgroup\$
    – Andy aka
    Nov 30, 2020 at 15:17
0
\$\begingroup\$

Additional observations to compliment AndyAKA's

Your low-side has a ferrite on the gate while the top FET does not. While this is a "recommended" practice to mitigate EMI concerns, the choice of ferrite needs to be carefully considered since it will act as an open-circuit to HF.

As a result if there are any HF perturbations around the gate of the device, say due to the miller capacitance during switching events, the ferrite will appear as an open-circuit and thus impede the ability of the gate-drive to hold the gate voltage at a given potential

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.