# Off-line, half bridge. Weird primary voltage

I'm working with this circuit:

I only show the power section, which is not working well. It is a SMPS working in open loop, so that I can manually adjust the PWM signal. The capacitor C10 is not mounted on the board.

The parameters of the circuit are:

• VDD = 15 V
• VD = 311 V
• Vo = 24 V @ 2,5 A (approx 50 W output power)
• Switching frequency = 80 kHz

For the test, I connected a load of 25 ohms and gradually increased the PWM level.

The following figure shows the voltage on the transformer primary on the TP6, taking the TP7 as a reference point. I think the signal has an acceptable shape. Corresponds to an output of approximately 3 V

I set the control signal so that the output voltage (Vo) is 7 V, and I get the following signal as voltage in the primary:

In addition to the "hump" that appears in the zone of zero voltage, the asymmetry in the amplitude of the pulses is notorious.

Taking the output voltage to 17 V, an oscillation is observed:

And this is the waveform for 24 V at the output:

My question is: why this behavior? What would be a possible solution?

At first I thought that it could be a magnetizing current (transformer) too high, but increasing the output current level (by varying the connected load) does not change the effect.

I apologize for the quality of the images, I hope they can serve as reference for the description of the problem.

EDIT: Here is the input rectifier block with reference to point PRI_MID

• Obviously, there is an asymmetry in either the drive or the load. That low-frequency ringing effect needs to be investigated. My first question is, what is PRI_MID connected to? If it isn't solidly bypassed to the primary ground, odd things could certainly happen. Try looking at all four terminals of the transformer (single-ended, not differential). Sep 21, 2017 at 15:20
• @DaveTweed PRI_MID is connected to the mid point of filter capacitors in the input rectifier. Sep 21, 2017 at 15:23
• What kind of capacitors are they? What ESR/ESL do they have at 80 kHz and above? Sep 21, 2017 at 15:26
• @DaveTweed , I dont know exactly. I'll look for the manufacturer's data. What should I pay special attention to? Sep 21, 2017 at 15:34
• The point is, if you're already blocking DC with C8, why do you need a DC midpoint connection? Usually, it's either one or the other. You really need to study where your high-frequency loop currents are flowing. Sep 21, 2017 at 17:06