Why might an inductor current waveform overshoot/oscillate?

[state_of_flux]

I am simulating a power supply I have designed in LTSpice, which is a buck converter feeding a half bridge LCC (not LLC) converter. In the buck output inductor, I get a waveform that overshoots (almost oscillates slightly?) then settles at steady state with required output current and ripple I designed for, but I am confused as to why the oscillation occurs. My guesses are that it is resonating with something, maybe the switch capacitance. As this is a current-fed topology I do not have the output capacitor and instead just have the bridge switches following the inductor.

Any suggestions on why this is occuring and how it can be remedied?

[soldar]

What's the load?

With a resistive load and without an output capacitor I cannot see how a buck converter could not show ripple.

A Bode plot analysis would be useful. A damping factor of less than 1 would lead to ringing/oscillation.

http://support.motioneng.com/Utilities/bode/bode_16.html

[T3sl4co1l]

Check control compensation.

It... is current mode controlled, right?

If you're just simulating a fixed duty cycle, that's not very meaningful in terms of low frequency dynamics.  Just replace the buck stage with a Thevenin equivalent source.

Tim

[state_of_flux]

Check control compensation.

It... is current mode controlled, right?

If you're just simulating a fixed duty cycle, that's not very meaningful in terms of low frequency dynamics.  Just replace the buck stage with a Thevenin equivalent source.

Tim

The idea is that the I will use buck inductor current and output voltage to control the system, so two error amp stages. At the moment I do not have a control system and I am just trying to simulate the main converter electronics, and design a filter before I look at the compensation.
I am (at the moment) simulating with a fixed duty cycle as you describe because I have yet to design the feedback.

[state_of_flux]

What's the load?

With a resistive load and without an output capacitor I cannot see how a buck converter could not show ripple.

A Bode plot analysis would be useful. A damping factor of less than 1 would lead to ringing/oscillation.

http://support.motioneng.com/Utilities/bode/bode_16.html

At the moment I have not designed the feedback for the system as I am still having a few issues with the actual converter circuit. The ripple on the inductor is fine, but for some reason the actual waveform is fluctuating before reaching a steady-state. Is it possible to do a bode plot without having first designed the feedback?

[soldar]

At the moment I have not designed the feedback for the system as I am still having a few issues with the actual converter circuit. The ripple on the inductor is fine, but for some reason the actual waveform is fluctuating before reaching a steady-state. Is it possible to do a bode plot without having first designed the feedback?

Yes, a Bode plot just shows the transfer (amplitude and phase) of anything.

It would be useful if you would post the schematic you are working on.

[MagicSmoker]

The idea is that the I will use buck inductor current and output voltage to control the system, so two error amp stages. At the moment I do not have a control system and I am just trying to simulate the main converter electronics, and design a filter before I look at the compensation.
I am (at the moment) simulating with a fixed duty cycle as you describe because I have yet to design the feedback.

You can't expect normal behavior if you just operate a switchmode converter (much less a cascade of two or more) at a fixed duty cycle (and load).

That said, when you see what looks ringing the first thing to do is find out what could be forming a resonant tank by measuring the period/frequency and then looking for the unknown C (or L) that would resonate at that frequency with the known L (or C).

Note that when you cascade converters you can avoid a lot of misery by operating one stage at a fixed duty cycle (that is, with no control loop; pulse-by-pulse current limit on that stage is ok, though). Especially when one of those stages is a multiresonant converter like an LCC or LLC.

[state_of_flux]

You can't expect normal behavior if you just operate a switchmode converter (much less a cascade of two or more) at a fixed duty cycle (and load).

That said, when you see what looks ringing the first thing to do is find out what could be forming a resonant tank by measuring the period/frequency and then looking for the unknown C (or L) that would resonate at that frequency with the known L (or C).

Note that when you cascade converters you can avoid a lot of misery by operating one stage at a fixed duty cycle (that is, with no control loop; pulse-by-pulse current limit on that stage is ok, though). Especially when one of those stages is a multiresonant converter like an LCC or LLC.
 

Yeah, the plan is to operate the LCC at fixed duty cycle and have some protection features but not controlled. The buck converter is controlled only.
It was my logic that I shouldn't start the control loop until I get the response I desire (Vout, efficiency etc) from the converter. That's where I am now, but I am now stumped how to design the control.

[state_of_flux]

Hi all,

I have attached the LTSpice schematic alongside the .asy files and .lib files that I use for simulating. As mentioned I have not yet designed the control system. I have done a bode plot on the output of the filter to try and identify a starting point but I am struggling to do so as I have only just started learning about the compensation/control part of power electronics.

Any advice on where I am going wrong/where I can improve/steps to take to get to a good supply would be greatly appreciated, even if that means completely ripping into my design and finding millions of faults. 

[soldar]

Designing SMPS is very complicated and way beyond my level of expertise. I think it is not something for amateur tinkerers.

Having said this, it seems to me if you have ringing and you want to suppress it the simplest way is to increase damping by increasing resistance.  Again, a Bode plot will show the system's response to different frequencies and you want to avoid a peak at the resonant frequency.

This is even before you get into feedback control loop.

Once you design a control loop you will want to see the response of the system to a step in the load, i.e. switching the load on/off.

Maybe others more knowledgeable can give better advice.