Current Mode Control of Boost Converter using SIMPLIS Simulator

Hi everyone,

I've been trying to implement a compensated closed-loop current mode(CM) controlled boost converter using the SIMPLIS engine. However, I am having convergence issues with the simulator.

Here are the converter specs I followed: Vi = 12V, Vo=18V, Io=1A, fs=100khz.

To have a solid understanding of CM, I used primitive components available in SIMPLIS such as discrete comparators, controlled sources, an S-R latch and a parameterised op amp. I started off my attempt to close the loop by generating a control-to-output bode plot using SIMPLIS's AC analysis. From these plots, I designed a type II compensator.

I wanted a crossover frequency of fc=fs/10=10khz and a phase margin of 45 degrees. From the bode plots, the converter had a gain of -10dB and a phase of about -120 degrees at f=fc. This meant the compensator needed to give a gain of 10dB and phase boost of 45-(-120+90)=75 degrees. Using the k-factor method of determining the R's and C's of the compensator, I came up with the standard op-amp based implementation. I confirmed my design by generating the transfer function of my compensator in SIMetrix software. The op amp used in this design was the standard LM358.

Now what's left to do was to close the loop - connect the compensator to the converter. However, SIMPLIS POP analysis is unable to converge. I tried to set initial conditions on the inductor and capacitors, but the analysis still won't converge. I have been stuck with this problem for the past week :-( still no progress.

Could anyone point mo to the right direction? What am I doing wrong? Is my understanding of CM, based on the circuits, flawed? Are there other factors I failed to account for? Any help would be appreciated :grin:

Here are the simulations files for reference:
**broken link removed**
**broken link removed**
**broken link removed**

Questions:

* What is S1? It is in a position where there normally is a transistor or mosfet.

* Have you ensured that oscillations start up and continue, and do not go into a 100% duty cycle?

* Have you created a simple boost converter (without the feedback section)? Has that design worked in the simulator?

I presume you are using Simplis .POP analysis?

As far as I see, you have the AC source shorting the feedback part in all simulation circuits, which can't work.

P.S. Forgot my previous comment. I didn't understand the working of the SIMETRIX bode probe.

S1 acts as the switching FET. It's basically a voltage controlled ideal switch. It has parameters such as off resistance, saturation resistance, and a turn-on voltage threshold that you can easily change (see picture below). In the circuit and image attachment above named "boost_currentmode_openloop.sxsch", the turn-on threshold is set to 2V, while the S-R latch U2 outputs a low voltage of 0V, and a high voltage of 5V. E1 is a voltage controlled voltage source that acts as a voltage buffer between the latch and ideal switch, as it has gain set to 1 as seen in the picture. This ensures that S1 switches properly.

And yes I have tested a simple open loop boost converter and it works just fine. You can see below some of the steady state waveforms generated using the circuit in "boost_currentmode_openloop.sxsch". Switch current in blue, gate drive voltage in green, output voltage in red, and inductor current in yellow.

I ensured that the open loop current mode converter works properly. And I also ensured that the compensator gave the correct magnitude and phase response. When I combined the two, SIMPLIS complained that the periodic operating point (POP) analysis could not converge :sad:

I'm not regularly using SIMETRIX, but the closed loop circuit analysis converges in my test. The waveform shows a strange excursion near the end, it disappears with decreased feedback gain. Thus I assume, the POP analysis requirements are somehow violated. When I look at the inverting comparator input signal, there's much 100 kHz ripple, more than I would expect for a practical CM switcher.

SIMPLIS is an interesting tools but obviously needs some initial training.