I'm currently redesigning the inverter for driving my induction heater.

While I've sorted out most problems I encountered I'm still struggling with the actual inverter.

Picture 1 shows the waveform problems I encountered.

Yellow is the actual inverter output.

Purple is the resonator voltage.

Cyan is the driving signal delivered to the driver.

Everything works fine until I go up in power levels.

Oscillations only occur at peaks of mains voltage, so it has to be some sort of current problem.

Scoping the output of the gate of the IGBT modules reveals that the signal supplied to them represents the output fairly accurately (also ringing.)

The IGBTs used are skm100gb128d which have an integrated gate resistor of 5 ohms.

The driver consists of optocouplers, voltage regulators and gate drive ICs.

  • What could cause this ringing?

  • Why is it visible in the cyan signal which is optically isolated?

  • How to prevent this?

  • Are 5 ohms of gate resistance enough if the signal at the IGBT gate looks alright (no overshoot?)

The second picture shows the setup of the IGBTs and the drivers associated with it.

UPDATE: The ringing was caused by the oscilloscope ground loop being located too close to the actual inverter. The voltage induced in the oscilloscope probe then acted back on the signal generation which was part of the reason the distorted signal was visible even before the Optoisolator.



  • \$\begingroup\$ How are we supposed to know what any of this means without a schematic diagram? \$\endgroup\$
    – Dave Tweed
    Feb 24, 2019 at 18:57
  • 1
    \$\begingroup\$ Could you take your scope display image and use numbered circles to highlight specific questions you want to draw attention towards, along with a numbered list in your text about your thoughts and questions for each? It sounds to me as though you have several questions but at the end you just ask about the cyan signal. It would help me and it might help others if you could be a little more concrete about how many questions and the details for each. A schematic would also be good, as Dave mentioned. \$\endgroup\$
    – jonk
    Feb 24, 2019 at 19:02
  • \$\begingroup\$ You should highlight the specific section of the waveform you are referring to. There’s a lot of ringing in multiple places in it and the answers will reflect different people’s assumptions. \$\endgroup\$ Feb 24, 2019 at 19:47

1 Answer 1


Random Ringing is due to "some" unknown external disturbance like a magnetic field from an old Weller soldering iron switching On/Off with a relay switch AC load current. Lack of high impedance shielding on sense lines can contribute to this or poor layout. ( Notorious cause of glitches in the 70's/80's)
enter image description here

p.s. The well-damped bottom pulse after the disturbance suggests to me that your mains current sense ( make one) should see a spike here due to cross-conduction during the disturbance)

What is the rep. rate of this random disturbance? 1ppm ? only when low load?

Steady Ringing is an inherent response due to mismatched load to source impedances in a 2nd order or higher system response to a step input. We can define this as Phase margin, Damping factor or Q and all are related to the equivalent circuit of the switched resistance before and during deadtime, the choke L, the wire inductance 1uH/m and the FET Miller Capacitance Cm * gm.

For simple series RLC low pas filter I use \$Q = \frac{1}{Rs} \sqrt{\frac{L}{C}}\$ and critical damping using this formula has Q=0.707 and has ~25% overshoot when Q=1

This is fundamental, but details are in the actual circuit model.

Wire inductance , crosstalk and stray positive feedback can amplify this Q factor as well by reducing the "phase margin".

Possible solutions

  • Tight layout with a ground plane.
  • Lead-lag filter on Vfb.
  • Snubber on FETs for impulse during deadtime commutation.
  • Complex realization of circuit impedances over every component, including all Caps that need low ESR L/R=T1, C*ESR=T2 for each component with conjugate matching., But too low ESR can reduce BW of feedback and reduce phase margin.


I compute that your ringing is at 2.5MHz so if it is a x1 probe , it could be due to probe ground resonance but if 10:1 probe, that resonance is > 20MHz from long probe ground leads.

Speaking of disturbances

( We just had a 1-minute power failure in Richmond Hill near TO probably due to melting ice on underground HVAC cables and PD (partial Discharge arc) activity, which seems to happen when the snow melts annually in our area. )

  • 1
    \$\begingroup\$ Thanks for the very qualified awnser, the ringing only appears over a certain current. I'm running it at rectified and poorly smoothed mains so those disturbances only appear at mains voltage peaks. I'm using a x10 Probe. I will try order proper pcbs for the driver circuitry with a ground plane and update the post with a schematic of the current driver circuitry. \$\endgroup\$ Feb 25, 2019 at 8:41
  • \$\begingroup\$ There will be many other possible reasons if more design details are given. \$\endgroup\$ Feb 25, 2019 at 11:37
  • \$\begingroup\$ I probed a bit more around and the ringing is already visible in the amplifiers for the Optocouplers, it is then carried on up to the Gate drive ics. I will try some shielding and a new Layout which might be causing problems. \$\endgroup\$ Feb 25, 2019 at 14:33
  • 1
    \$\begingroup\$ Small update, Shielded wire and increasing the distance of the control circuitry to the resonator and inverter has increased the stability by about 10x.... I will shield it further and will order proper pcbs with a ground plane. \$\endgroup\$ Feb 25, 2019 at 16:03

Your Answer

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

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