Hi, i am trying to design a maximum power point controller for a bank of photovoltaic panels. The panels have have about 10kW of power, with voltage up to 900V and 12A. As the load there is a bunch of IR heating elements which pose a high load for the panels when they´re cold.

Now the owner wants me to design what is basically a PWM controller that should control an IGBT module powering the elements in such way that the voltage coming from the panel is kept relatively constant. The pwm will be controlled by some simple micro at ~15kHz.

The problem I need help with is designing the input and output filter in such way that it minimizes ripple current on the input line, that I guess should give me best efficiency right? The output doesn´t need to be too clean beacause the elements don´t really care, but the input I think should be so that it stays close to the set voltage.

 

Seems like you could use a buck boost calculator like this one to set the desired Inductor ripple.

http://www.daycounter.com/Calculators/Switching-Converter-Calculator.phtml

 

Ok, the input filter works like a treat, but I get absurd results with the switch.
My period is set to 66us, V3 has rise and fall times of 2us which aims to mimic the switching times of the actual igbt that will be used later. The voltage source X1 is putting out around 800V. L2 and RL is my guess at the load, but the inductance doesn´t change anything.

I started to try different pulse widths, and I can´t understand why for anything lower than about 20us the peak Vce voltage gets extremely high, when for higher pulse widths it stays around 150Vp. Is this some resonance in the output filter or something or is it just simulation bollocks? I feel the panel will have to be separated into lower voltage sections anyway, but still I want to get this right.