0
\$\begingroup\$

I am operating some mosfets in the 50-500khz range, and wish to improve performance, reduce losses and very important reduce emi. I am using a resistor on the gate on the drive with a 12v zener and 10k resistor to ground/0v to try and adsorb transient voltages at the gate.

My question is, if i wish to permit 50-500khz operation / switching, but reduce emi and parasitic oscillations from the device, while maintaining fast switching, would i benefit from ferrites and would you be able to suggest some suitable options/ where to place (is it between gate drive resistor and gate?)

I Occasionally detect parasitic of multiple mhz that i wish to suppress.posting an exact schematic is not possible at this time but looking for advice anyway.

Thanks in advance

Edit: I'm really sorry for not posting a schematic when asking the question originally, main reason is that i don't have permission for what im working on at the moment but i work with a number of switching circuits in the khz region and was postulating about how one might stop ringing at the gate, while not slowing switching, after choosing fast components and suitable gate drive etc. The question was general in that i understand a resistor on the gate of a mosfet can help provide some level of dampening, and routinely use them, but wished to know if a ferrite placed in the gate drive would be a good or bad idea in general to clean operation up more.

However to try and present something of a similar situation ive had a go this morning and dusted off an old mazilli zvs fly back driver from my tesla coil as i can switch it through a suitably sized inductor to achieve the same frequencies my circuit switches at and although the parasitics are slightly higher in frequency and amplitude in my test, they seem to occur in a similar fashion in that circuit too. I am scoping the gate and the source seperately and can see them occuring on switching. Although its a different circuit, my question, badly worded, was general as to whether a ferrite, placed on / inline to the gate of the fet would reduce ringing and emi? As a practical curiosity where i don't actually have any suitably sized ferrites in my parts boxes to test practically at the current point in time.

Link to a mazilli zvs (different situation but similar and somewhat exaggerated problem):

enter image description here

\$\endgroup\$
  • 3
    \$\begingroup\$ Can you post a simplified schematic? How much current is being switched? Have you taken other EMI-mitigating steps (improve layout to reduce loop area)? Are you seeing resonance at specific frequencies? Do you have a real problem or are you just trying to avoid problems in a design you haven't built yet? \$\endgroup\$ – The Photon Jul 5 '17 at 22:50
  • \$\begingroup\$ The first thing I'd like to say, is be careful with the gate resistor, because it forms a RC filter with gate-source capacitance (probably a few nanofarads), and if this filter chops off the sharp edges of gate drive signal then the MOSFETs will not be driven properly thus the power loss on them will be high. Another thing to consider about EMI while driving high current loads with MOSFET is ringing caused by parasitic elements. And they can be suppressed with flyback diodes and snubbers. \$\endgroup\$ – Rohat Kılıç Jul 6 '17 at 4:45
  • \$\begingroup\$ Posting an exact schematic is something THAT YOU MUST DO. \$\endgroup\$ – Andy aka Jul 6 '17 at 8:44
  • \$\begingroup\$ Sorry Andy, have edited and come up with another schematic / far simpler circuit that has same issue. \$\endgroup\$ – Rendeverance Jul 7 '17 at 14:27
  • \$\begingroup\$ Hi Photon, switching around 30A, trying to avoid problems in future designs to be fair, it was a general question i have seen from what i was currently working on, AND could replicate with the simplest circuit i know that switches large currents in khz range - the mazilli, as linked. In what i was working on i see ringing at 2 and 4 mhz, possibly higher too but 2 and 4 dominate \$\endgroup\$ – Rendeverance Jul 7 '17 at 14:31
1
\$\begingroup\$

I suggest you use ONLY PLANES for your high-current high-speed-switched-current loops. And have no breaks in the planes where outgoing and return currents need to co-exist to null out the magnetic fields (which by the way reduces the stored energy and thus lets the circuit switch faster).

By the way, you need not allocate entire planes to the high-speed high-current loops. Do plan, with sketches, your high-speed components placement; select components with pinouts and backside solder-regions etc to minimize component heights above the planes (or regions of planes).

Regarding MOSFET oscillation, a "gate stopper" resistor is used to dampen ringing. Assume 10nH to 100nH (1/2" to 4" of wire, 10,000pF Cgate, and transconductance of 1 amp/0.1 volts. Lotta potential to oscillate, in Colpitts config.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Thank you, on the circuit im working on i am designing with planes to try and suppress emi. The question was general because i often see ringing of potential at the gate in high current switching and wondered if i could do anything more to eliminate that. I don't know if it contributes greatly to emi design issues but it can't help, and the more ringing i see at the gate, generally the worse the circuit seems to perform. I use a stopper resistor already, but obviously that is dc resistance, i wondered if an inductor online with the gate could reduce ringing? \$\endgroup\$ – Rendeverance Jul 7 '17 at 14:45
1
\$\begingroup\$

What is your EMI problem, specifically? Is it conducted, or radiated? Is it magnetic or electric? Is this a buck converter, or boost converter, or H-Bridge? I'm going to assume a buck converter here.

You will have conducted noise on the input and output of the supply. This noise should be around the fundamental switching frequency of your converter (so in that kHz range), and will always be present -- you can only minimize it, but never eliminate it.

You could have radiated noise as well. The classic 100-200MHz "hump" of radiated noise is a result of your switch node, and is unavoidable parasitic RLC tank that is excited at each hard switching transition of the MOSFETs. At these frequencies, now power supply leads and other wiring can act as effective antennas for the noise to radiate away.

Magnetic interference is easy to fall victim too if using an unshielded inductor, and at high currents. If your current loop areas are large, and differ greatly in size, the changing flux will also result in problems: enter image description here

There are ways to combat / suppress these noise sources. For switch node ringing, a RC snubber is the most common way, trading efficiency / power dissipation for a reduction in EMI. RL snubbers can also be used. MOSFET gate resistors are used to slow-down the rise time of the FET, broadening the spectral content generated by a fast edge. (You will often see a diode in parallel, with cathode facing the driver to allow the fall-time to be less affected).

Common-mode chokes are common input filter devices, and (if this is a converter) the LC filter on the output is in fact a giant filter. Adding further ferrite beads or LCs will again trade efficiency / heat generation for further attenuating noise.

There are a lot of good app notes on this subject to help you out, and some walk you through the debug process as well.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Hi Krunal, thank you for your answer, im scoping the gate directly and seeing 2, 4 and possibly 8mhz ringing at the gate. This is not radiated far and i don't seem to have issues with anything but my operating frequency being radiated magnetically, except close to the fet itself. I use chokes on the positive rail supplying my circuit and in combination with some diodes this does a good job of suppressing back emf. But not sure about ringing of potential at the gate itself, which can transfer to the load if its severe. I replicated the issue with a simple zvs circuit as in my edit above. \$\endgroup\$ – Rendeverance Jul 7 '17 at 14:41
0
\$\begingroup\$

You should contact an applications engineer at fair-rite.com to be certain, but I believe that their 75 material is close to what you are looking for. See http://www.fair-rite.com/75-material-data-sheet/

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ I don't think I've seen any ferrite beads using this material but maybe you can illuminate? \$\endgroup\$ – Andy aka Jul 6 '17 at 8:22

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.