I am designing a 3-phase inverter board for BLDC Motor control. The H-bridges i use are IFX007 from Infineon. The mentioned IC has the possibility to adjust the slew rate.

From what i have read, i've understood that higher slew rates result in a more "perfect square wave" shape which in return gives rise to more high frequency components (Fourier Analysis).

Those extra high frequency components lead to EMI ( Electromagnetic Interference).

I plan to control the H-bridges with 20 KHz PWM (suitable range for Field Oriented Control). The problem is, if i limit the slew rate and make it as small as possible, the valid duty cycle with 20kHz drops to between 20 % and 80 % due to the rise time. (the rise time consumes nearly % 30 of a whole duty cycle).

On the other hand, i can just increase the slew rate and make the rise time way faster, having a larger range of duty cycle(which i would love to have) as a result but higher EMI.

My question/dilemma is: Would increasing the slew rate with the flowing specifications be ok:

  1. 14 V Power supply feeding the motor
  2. The motor wont draw more than 5 amps in any case
  3. the H-bridges switching at 20kHz
  4. no wireless component or delicate analog circuitry that would be 'victimized' by the EMI. (only the MCU will be stacked on the far side of my pcb.)
  • 1
    \$\begingroup\$ There is no one-size-fits-all margin between good and bad EMC compatibility. Compliance usually rests on very subtle changes that are generally unforeseen during the design phase. \$\endgroup\$ – Andy aka Nov 27 '18 at 14:34
  • \$\begingroup\$ EMI can interact on a PCB in very unusual ways. As Andy notes, subtle changes can do all manner of interesting things, as can not using the appropriate power distribution. There is far more to EMI control than simple part placement. \$\endgroup\$ – Peter Smith Nov 27 '18 at 15:09
  • \$\begingroup\$ Thanks for the input! The answer i was looking is more like "never do that! thats a cardinal sin!" or "that should be aight with your given specs" of a kind. This is one the situations that people just come to a compromise with their gut feeling that comes from experience (which i lack) i suppose. \$\endgroup\$ – Firat.Berk.Cakar Nov 27 '18 at 15:41
  • \$\begingroup\$ Use twisted-pair wiring between the controller and the motor, even if 3 or 4 or more wires are twisted together. This minimizes the "loop area" which minimizes the inductance and minimizes the radiation. (the loop area is the radiator) \$\endgroup\$ – analogsystemsrf Nov 27 '18 at 18:22

You certainly don't want a slew rate so slow that it'll have a significant impact on available duty cycle at 20kHz, there is a switching loss in the output FETs every time they are not either in full conduction or fully off, so minimizing the switching time minimizes the power loss and heating. On similar controllers, I've found a 0.5 - 1us rise/fall time typically is an acceptable compromise between switching losses and EMI generation, but it depends greatly on the load, component and trace placement, and the interaction of the various other components, including the recovery time of the freewheeling diodes - the body diodes that the FETs offer are usually pretty slow and can cause issues.

Distance between the EMI sources and any sensitive components won't guarantee immunity either, conducted noise can easily affect anything connected to common lines, and placement of decoupling caps or filters may be needed, but I doubt that there'd be many issues with a controller of the specs you quote. Just be liberal with decoupling caps, they're a cheap insurance.

These modules look to be aimed towards fairly low frequency operation, the on/off delay times are going to limit the range of duty cycles anyway - FETs intrinsically have a delay of just tens of nanoseconds, and with a microcontroller driving them you can play with the dead time that is present between the gate driving signals to fine tune the timing, but on these modules that all appears to be built in and fixed, and the delays that the control circuits introduce are dependent on the selected slew rate.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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