I am designing a 1kHz inverter in a push-pull configuration with KA3525 and IRF540N MOSFETs.

  • Input - 12V
  • Output - 110Vac - square wave filtered to sine wave
  • Power - 15W at 1000Hz

enter image description here

When I fire it up on load I get some noise on the gate pulses.

enter image description here

Also note that the gate pulse is only about 8V peak instead of 12V.

I have a feeling that the gate noise is causing the MOSFET to switch on and off very fast and cause circulating currents and heat up the MOSFETs.

enter image description here

I built this on a perf board to test before going for a proper PCB, but I don't think that should cause any problems at such a low frequency.

I have tried changing the IC but that did not change anything.

Gate resistance is 100 Ohm and gate to source pulldown resistance is 1k.

Any advice is appreciated.

I have tried to keep the layout as clean as possible.

enter image description here


Power loss:

  • Ipeak = 1.25A
  • Duty = 50%
  • I_rms = 1.25/sqrt(2) = 0.88
  • R_ds = 52mOhm
  • P_loss = 0.04W
  • Rthj_A = 62 Deg/W
  • T_rise = 2.5 Deg C (The MOSFET is overkill for this application)
  • Deadtime is about 40us. These waveforms are at max dutycycle.

I tried using shorter ground paths to eliminate external noise - no change.

enter image description here

No change - (blue is the new setting and yellow is the older setup.)

enter image description here

The negative pulse before was most likely due to improper setting of the base line.

The gate signal starts off clean but soon gets noisy. Check time stamps on the video.

enter image description here

enter image description here

enter image description here

enter image description here

  • \$\begingroup\$ Your ground noise is exceptionally large going below 0V. Make sure your probe gnd lead is short or use spring coaxial probe tip with no gnd leads. Consider a cap on R2 and watch out for positive feedback as the reduction rate of pulses after each clock edge suggests this or phase compensation needed with some hysteresis somewhere. What does I^2RdsOn *Rjc compute to? How much deadtime is there? \$\endgroup\$ Apr 15, 2021 at 17:58
  • \$\begingroup\$ @TonyStewartEE75. Thanks for your comments. I tried a few things out (added as edit to the question) but did not help. I will add a capacitor on R2 and try to play with the gate pulldown resistor and snubber circuit and update you on it. \$\endgroup\$ Apr 16, 2021 at 8:43
  • \$\begingroup\$ " watch out for positive feedback as the reduction rate of pulses after each clock edge suggests this or phase compensation needed with some hysteresis somewhere" - I did not quite understand this. KA3525 or SG3525 does not change pulsewidth while the pulse is on as far as I know. \$\endgroup\$ Apr 16, 2021 at 8:45

2 Answers 2


These are comments, really, but I need to show you some pictures so I have to put them here as an answer.

Your schematic diagram doesn't seem to be correct. There are several points that don't make sense:

  1. The MOSFETs don't appear to be connected to anything.

enter image description here

The drain goes to the transformer, but the source just snakes around and connects to the drain through a capacitor.

  1. The transformer connections are nonsense.

enter image description here

  • Pin 1 is marked "NC_1" - NC means "not connected," but you have it wired to something.
  • The PRI pins would normally be used in pairs - PRI+1 and PRI-1 in use together or PRI+2 and PRI-2 in use together.
  • There's no match for PRI-3.
  1. You have DC going into the transformer (or maybe coming out of it.)

enter image description here

Pin 1 of the transformer is connected to Vcc of the KA3525. Pin 1 is "NC" so I have no idea what that's supposed to do.

I can't tell you what's wrong with your circuit, but I can tell you that you can't use that diagram to troubleshoot it.

If your transistors are getting hot, then you must have them connected in some way that isn't shown in the diagram.

Correct your schematic, and then we can see what's going on.

Simply check everything in your circuit as it is built against the circuit diagram and correct the diagram to match the real circuit.

In doing so, you might find the actual cause of the problems, in which case you could correct both the diagram and the real circuit.

  • \$\begingroup\$ Thanks a lot for pointing out the silly mistakes. The schematic is updated to reflect the missing ground wire. \$\endgroup\$ Apr 16, 2021 at 10:29
  • \$\begingroup\$ What about the "NC" pin on the transformer? \$\endgroup\$
    – JRE
    Apr 16, 2021 at 10:36

Stupidest problem ever. The input voltage was dipping below 9V on load and the IC was going into shutdown or crazy mode.

Sorry for wating everyone's time.


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.