Post by William WestfieldPost by William WestfieldIs there a relatively simple equation for estimating the maximum
possible performance of a (buck) switching regulator?
Post by William WestfieldI mean, if I have Vin, Vout, an inductor with an internal resistance of
R ohms, and a switching element with a drop of Vf volts, I’m going to have
an efficiency that is limited by those factors, regardless of how perfect
the rest of the circuit is, right?
Lets see if we can put an upper bound on the efficiency.
I was 'back of brain' considering how many factors you'd need to add
before getting into the noise.
I useful point is that in an ideal system Vin/Vout = Tout/Tin for one cycle.
or Tout = Tin x Vin/Vout
ie for a given charge time the discharge time DECREASES as Output voltage
INCREASES.
Tin = inductor charge time.
Tout = discharge time.
Then, once you add inefficiencies you need a longer Tin to supply the same
Vout.
ie effective or apparent Vout is increased by inefficiencies.
So for efficiency Z, 0 < Z < 1 you see
Tout = Tin x Vin/Vout x Z
This is useful for determining how long inductor current flows, and how
long the switch and 'rectifier' elements conduct in a cycle in a given mode.
To the basic ohmic losses consider adding various junction capacitance
charges and discharges and switching losses due to eg partial conduction
during switching and turnoff delays due to charge seepiyt times and ....
agh! :-). (And more).
Somewhere you get down into the noise.
98% is awesome no matter what.
95% is extremely good and often only for a selected part of the operating
range.
90% is often achieved for many systems using best practices.
80%+ can be hoped for from any mass market kluge.
Lower has been seen :-).
Very high Vin:Vout ratios hurts efficiency.
Very high frequencies reduce inductor sizes but demand superior switch
performance.
Very low frequencies increase core sizes and losses and are usually avoided
unless special reasons dictate.
...
E&OE.
Corrections welcomed.
Russell
--
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/