Why frequency responce is needed for DC-DC converters?

Thread Starter

onlyvinod56

Joined Oct 14, 2008
363
Hello,

Why frequency response (for ex: Bode Plot) will be taken for stability analysis for buck converter? We don't use AC in buck converter. But why it needs to be studied for frequencies zero to infinity? One of my friend tested a buck converter by giving a small AC voltage at one of its ports. I didn't get the logic there?
 

Papabravo

Joined Feb 24, 2006
12,405
Because a DC-DC conversion is basically an AC process with the ability to change either the switching frequency and/or the duty cycle. A figure of merit is how fast the conversion process responds to load transients.
 

ci139

Joined Jul 11, 2016
716
"stability analysis" -- is the keyword
what is happening is the varying "nature" of the load and the nature of this "varying"

for example if you test simple linear voltage references - then in.. (my test set's) ..common they have an apx. stability island at 1.5 to 15mA.. (or more tight at 7 to 10 mA) .."operating current" --e.g.-- the operating range that does not need a special compensation is very narrow ← applies to very simple designs consisting of "few" parts only

so - you need to compensate your specific topology +also accounting it's operating "environment" --or-- select a more suitable one for

// as with voltage ref.-s -- the SMPS has best regulation --and/or-- stability somewhere in the "middle" of it's op.-ng range ← this and the behaviour at the edges of frequency spectrum is somewhat illustrated by Bode Plot -- while a more specific response characteristics are gained by frequency×load(×power.src)
 

crutschow

Joined Mar 14, 2008
23,349
The circuit that accurately maintains the output voltage of the DC-DC converter for changes in input voltage and output load is a negative feedback DC loop that also has an AC response, so it can be unstable if not properly compensated, primarily for the inductor and capacitor in the output which form a resonant tank.
A frequency response versus phase-shift (Bode plot) test is thus usually done to verify that the loop design has the desired gain/phase stability margin.
 
As crutschow and DarthVolta said, it's all about making sure the regulator, whether it's a switcher or linear, is stable with respect to load and line changes. A 3.3V regulator for digital stuff isn't much use if, for instance, it's going to ring up to 5V if there's a step change in it's output load.

Our power group spends a large amount of their time making sure their switching power supplies are unconditionally stable over load and input voltage changes.
 
Top