0
\$\begingroup\$

I'm debugging an electronic speed controller I designed that uses a 3-phase gate driver, DRV8300DRGER, to control BLDC motors. The circuit is shown here (the 2 other phase aren't shown) with the following component values:

  • DT resistor -> 100k (results in ~500ns delay, I'm not complementing the high-side and low-side MOSFETs anyway so there's no worry of shoot-through)
  • Gate resistors: 10Ohm
  • LS and HS snubbers: R = 10Ohm, C = 10nF (motor operation sounds cleaner with these populated)
  • C_bst = 330nF
  • C_GVDD = 30uF

Gate driver circuit:enter image description here

My motor is spinning well using open-loop control (I'm speeding up the frequency of the stator gate driver inputs using a MCU, and hoping the rotor can keep up with the stator energizings) but some waveforms do not look as expected. I'm sensing the back EMF, which I'll feed to a comparator to get zero-crossings (for control purposes), but my back EMF waveform doesn't look ideal (I'm expecting it to look trapezoidal). Names of signals are circled in the image. enter image description here

When I increase the time-scale, I see that the max values of the back EMF (and the other signals) appear sinusoidal. enter image description here

A zoomed in image of this scope capture is shown below where you can see the motor goes in-and-out of the "trapezoidal behaviour" as shown in my first scope capture. enter image description here

Some information that could be useful:

  • I'm using a PSU that's current-limited to ~5V and 4A. My PSU voltage setpoint is set to 12V. At max speed, the PSU voltage rises to ~6V (due to increased back EMF) but it's still current-limiting.
  • The gate driver supply undervoltage lockout (GVDDUV) is listed at ~4.6V and boot strap undervoltage lockout (VBSTx - VSHx) is listed at ~4.2V. The deglitch times to get out of both conditions is listed as ~10us.

Responses to comments/answers

  • Back EMF shape with the motor powered by a drill. The drill I used couldn't spin very fast. enter image description here
  • Below I have scope captures of another BLDC motor supplied at 2 different current limits. The 2A current limit case does not exhibit the sinusoidal behaviour, but the 6A current limit case does. Again, when I observe the sinusoidal behaviour, there's instances where the back EMF looks slightly trapezoidal with the spikes previously mentioned, and instances where the back EMF + gate driving signals look very messy.

2A case: enter image description here

6A case: enter image description here

Questions I have

  1. What causes the voltage spikes on the back EMF waveform? Is it because the freewheeling diodes of the MOSFETs are conducting during turn on/off?
  2. What could be a reason the signals in the zoomed-out waveform look sinusoidal?
\$\endgroup\$
2
  • \$\begingroup\$ If you disconnect your motor from your inverter, hook up your oscilloscope to measure the drop between two phases, and spin the motor using a cordless drill, what is the backemf shape? \$\endgroup\$
    – Ocanath
    Mar 8 at 0:29
  • \$\begingroup\$ Updated my post with the back EMF waveform when a BLDC motor is powered from a drill and probing 2/3 phases. \$\endgroup\$
    – ken
    Mar 8 at 6:47

1 Answer 1

0
\$\begingroup\$

By definition , if you are running the motor open-loop, you are driving at a speed lower than the optimum motor speed. Another way of looking at it is that you are switching phases too late. When you switch, the motor generates torque by attracting the magnets on the rotor to the energized pole, but at the end of one of your steps, the motor becomes a generator and your driver becomes a brake. Current will then be flowing from the motor phases back through the driver. When perfectly timed, the open phase exhibits the back-emf signal that crosses zero halfway through the step, but the back emf signal comes from the relationship of the magnets and the windings, so they signal will not be symmetrical since the timing is off. I'm guessing now, but I think your power supply current limit may be causing the "sinusoidal" oscillations you are seeing.

\$\endgroup\$
2
  • \$\begingroup\$ I've updated my post with scope captures of cases with/without the sinusoidal behaviour when my current limit was set to 6A/2A respectively. Just to clarify, was your hunch that the "relatively low" current limit was causing the sinusoids and that increasing it would make the sinusoids disappear? \$\endgroup\$
    – ken
    Mar 8 at 6:51
  • \$\begingroup\$ As I said, it was a guess, but power supplies can react quickly to increases in load by increasing their duty cycle to boost the output voltage and current. When the output voltage or current is too high, they can reduce the duty cycle but generally must depend on the load to pull down the charge on their outputs. For this reason, it is usually not a good idea for a system with a rapidly varying current to run from a current-limited supply, and some variation in voltage should be expected even when not current limited. \$\endgroup\$ Mar 11 at 14:23

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.