I was trying PWM by giving a sine wave and triangle wave in a comparator. Could anybody explain what should be the frequency of triangle and sine wave? and Why?
In some applications it may be desirable to have a higher PWM carrier frequency even though the resolution may be
limited. The higher carrier frequency can allow using smaller inductors in switching circuits and is also easier to filter
from the output. The limited resolution can also be improved by “dithering” the PWM output. This method effectively
creates intermediate duty cycle values by alternating between two duty cycle values
One point is the transients in PWM as well as response time Tr for the product goes very high with higher carrier freq.
However , this is compensated by higher noise levels , RFI/EMI and spike in the output .
It further improves the efficiency/performance and lowers down on ferrite heat losses upto 30KHz.
You set your frequency based on what you need your loop
bandwidth to be (load step response) and what size of
ferrite you can stand (physical size). Tens of kHz is so
'80s. Maybe somebody was using a comparator with a
sporty 1uS prop delay or something.
Sine wave is going to get real jittery when you approach
the crest or valley. I can't think of a reason to want a
sine wave reference oscillator other than one you get
for free, which 15kHz would not be. It's bone simple to
make a triangle wave (or switched-RC) in CMOS.
PWM spells pulse width modulation. Without having a particular modulating signal (e.g. a sine), why using a PWM? The question refers to your PWM application. What is it?
A popular sine PWM application is UPS or motor inverter. In the former case, it's e.g. 50 Hz. Some criteria for PWM frequency have been mentioned: core size, "noise", core losses. Not yet mentioned but even more important are switching losses.