Malfunction of a rectifier using a transformer

[Francisco]

Hello,

I am currently working on a three-phase AC/DC rectifier simulation in PLECS and I am observing a behavior that I do not fully understand when introducing a transformer between the AC source and the diode bridge rectifier.

I would appreciate your help in understanding whether this behavior is physically correct, caused by transformer parameterization, or due to an incorrect modeling approach in PLECS.

System Description

I built a standard three-phase uncontrolled diode rectifier feeding an RL DC load.

The system consists of:

• Three-phase AC voltage source

• Three-phase transformer

• Six-pulse diode bridge rectifier

• DC-side inductive load (R-L)

The DC current is nearly continuous because of the output inductance.

Observed Behavior

WITHOUT the transformer

When the rectifier is connected directly to the ideal three-phase source:

• The AC input voltages are sinusoidal

• The AC input currents have the expected 120° conduction shape

• The DC output voltage has the expected 6-pulse ripple

• The DC current is smooth and continuous

This behavior appears correct and expected.

---

WITH the transformer inserted

After adding the transformer between the source and the rectifier, I observe the following:

1. Secondary line-to-line voltages become highly distorted

I am measuring:

• Vab

• Vbc

• Vca

on the transformer secondary side.

Instead of remaining approximately sinusoidal with only commutation notches, the waveforms become almost “three-level” or “six-step” shaped.

The line voltages appear strongly clipped and flattened.

---

2. Input currents become almost sinusoidal

At the same time, the transformer secondary/input currents become surprisingly smooth and nearly sinusoidal.

This is unexpected to me because for a standard 6-pulse diode rectifier I would normally expect:

• nearly sinusoidal voltages

• strongly non-sinusoidal line currents with 120° conduction intervals

However, in my simulation I obtain almost the opposite behavior.

My Questions

I would like to know:

1. Is this behavior physically correct for a transformer-fed 6-pulse diode rectifier?

2. Could this be caused by:

   • excessive leakage inductance,

   • incorrect magnetizing inductance,

   • transformer parameterization,

   • or an incorrect measurement location?

3. In PLECS, are the transformer terminal voltages measured:

   • before leakage impedance,

   • or after leakage impedance?

4. Could the transformer leakage inductance alone realistically cause:

   • nearly sinusoidal AC currents,

   • and highly distorted line-to-line secondary voltages?

5. What would be considered reasonable values for:

   • magnetizing inductance,

   • leakage inductance,

   • coupling coefficient,

   • or short-circuit impedance
     for a realistic transformer model in this type of simulation?

6. Is there a recommended way in PLECS to model a “nearly ideal” transformer for validating rectifier behavior?

Additional Notes

• The transformer secondary voltages being measured are definitely line-to-line voltages, not phase-to-neutral voltages.

• The DC-side inductance is relatively large, producing continuous conduction mode.

• No active switching devices are used — only a diode bridge.

• The issue only appears after inserting the transformer.

• The rectifier behavior without the transformer looks normal.

What I Expected

I expected:

• line-to-line secondary voltages to remain mostly sinusoidal,
  with only small commutation notches,

and:

• line currents to retain the classical 120° conduction shape.

Instead, I obtained:

• highly distorted secondary voltages,

• and almost sinusoidal line currents.

Goal

I am trying to determine whether:

• the simulation is actually physically correct,

• or if my transformer model is introducing excessive impedance or incorrect dynamics.

Any guidance regarding:

• transformer modeling best practices,

• parameter selection,

• or measurement interpretation in PLECS

would be extremely appreciated.

Thank you very much for your time and support.

Without Tranformer :

With transformer

below is the simulated file (I used the blockset version with matlab)

projet1.slx (93.4 KB)

[Bryan_Lieblick]

Hi Francisco, Thanks for posting your questions to the PLECS forum. Many of your questions are open-ended - what is physically meaningful and how components are parameterized depends on the application, power rating, etc. Much of this should be discussed within your academic community, although others on the forum are welcome to provide their feedback and insights.

I expect that much of the unexpected behavior you’re observing is due to how you’ve parameterized your transformer. There is a large voltage drop across the transformer resistances a the load current you have specified. If you set the resistances to 0 and disconnect the neutral on the primary winding transformer (or insert a resistance in the neutral path) the waveforms will be more akin to your expectations. Did you also mean to remove the DC load inductance in your second model?

The reason for disconnecting the neutral is that Va+Vb+Vc must exactly sum to zero at the source neutral and transformer neutral nodes, but due to floating point round-off that isn’t the case and there will be a small error.

You can always build your own transformer model out of primitive components (ideal transformer, inductances, resistances) to better understand the device operation.

[Francisco]

Hi Bryan, thank you so much for your help/answer.

I reduced the values of resistance and inductance and disconnected the primary neutral point, and now the waveforms are like I was expecting.