Pulse skipping buck converter

Horrible efficency 😁 cuz you eliminated the synchron emulation with diode! Which will cause (with diode) again overshoot by ringing if the filter, but it does increase the efficency drastical!!! See with diode and compare. http://everycircuit.com/circuit/5939036245196800

Holy sheet i made this so fast i forgot the diode! Now it works much better

But this is a good thing, cuz you might see the difference now. Ringing / overshoot / filter / synchron / not synchron / efficency. All this is related and need highest attention for a good buck converter. The digital part is just boring to make. The analog part is the importante one.

The digital part boring?? I love making the digital part! making digital controls and semiconductor bridges is my favourite part

This is good and ok, and I hope this will also be the case if have done 100 more similar complex digital circuits 😁 but I hope you agree or you see and understand why I am pointing you on the analog part. You maybe remember our first discussion about this buck converter concepts, a few month ago? You were only Interested in the digital part and you avoid having contact to the analog things.

Yes i remember that the analog part is really a delicate thing and you have to design it based on the load it is going to supply power to because it can easily go in resonance

Resonance, which is of its dampening here called ringing, which cause here at startup an overshoot. Yes?

I don't thing resonance is the cause of that overshoot

You dont think?

Ok, to understand, this might help. Make a so called "small signal model" of your analog part. Power supply, mosfet, filter coil, filter cap, shunt resistor, load resistor (attention load might be more then just a resistor!). Only the linear behaving properties are needed in such a model. The power supply is now an ac source to sweep for you over the visible frequency range and plots the resulting proportional amplitude (and phase). You can see, its a low pass filter, no surprise, right? Such a low pass does resonate at a fixed frequency given by L and C. Unloaded, your filter resonate/ring amplitude is 9.3dB @5kHz, dB is just a factor referenced to the input one and says 2.91 more then input! So have voltage gain! Loaded, its much less, about 2dB. In this setup you have a potential to overshoot to about 29V unloaded! Play with the settings, but importante are the load unload scenarios. http://everycircuit.com/circuit/4578414228144128

In reality i'm making an lmz12008 based buck converter. It has a 2.2uH internal inductor and goes at fixed 350khz. With 660uF (3x 220uF) capacitor this should never go in resonance with or without a load!

L and C always resonate if R is low enough. You have choosen here Rds_on quite high with 110mOhm. Tune it to 20mOhm and use 2.2uH with 660uF it does nicely resonate/ring at 4.18kHz. Check that with the "small signal model" I prepared above for you.

TI spec does not says anything about resistance, but some reverse calculation over power dissipation and output power I think R is about 50mOhm This will cause 1.34dB ringing at 3kHz and cause 17% overshoot. Anyway, the regulator will take care with softstart and for PID regulators the D part will slow down if voltage rises or falls to quickly and also avoids over- undershoots.

Give it a try. I still think about an even cheap and more elegant hack. But I havn't tested it so far. Let me see if I find time, cuz I need it anyway for a little other application then PS. I need a solution for a balancer. When is your lmz12008 finished?

That buck converter is a side project i'm making while i make the main project. I'll order all the parts this days. I've already designed a pcb for that converter and i'm going to order it with all the other pcbs i'm making for other projects