# Confusion over PWM.

#### cjdelphi

Joined Mar 26, 2009
272
I just need help understanding it (no project, no nothing in mind to do just been nagging away at the back of my mind)

Is this understanding correct? a PSU from 110vac 50hz, the PWM side could simply compare the voltage and if it's between 0 - 15v the PWM allows the electric/current to pass through to the caps and then onto the regulators and so on eventually feeding the components.

If it is, how exactly would PWM work if you're using say 8 - 12vdc but you want to feed something that connects to 3volts.... since there's no sinewave it's going to be a constant 8-12vds...

So how does it supply the 3.2volts (could be any voltage but argument sake 3.2v) from 8v or higher?... ?? this is driving me insane i find lots and lots of information about Switching Power Supplies/PWM using 110/240acv but nothing about 8-12v DC stepping down voltage?.... (only thing i can think of is it it converts it to AC and goes through a transformer)....

Anyone want to put my understanding right so i can stop thinking about it lol it's driving me insane.

#### ChrisHelvey

Joined May 22, 2004
45
I am not sure I understand the question exactly, but if you take 12V DC to start, and then make a 25% duty cycle PWM output from it, and then smooth it with a proper RC circuit, you'll get 3V.

Is that what you are asking?

#### Wendy

Joined Mar 24, 2008
21,840
I am not sure I understand the question exactly, but if you take 12V DC to start, and then make a 25% duty cycle PWM output from it, and then smooth it with a proper RC circuit, you'll get 3V.

Is that what you are asking?
Actually, that depends somewhat on the load. If the load has any peak detection features your in trouble, but then, that's what filters are for. The RC network you mention is now part of the load, and can adversely affect your regulation. I used to have the same thought, and had to work it though.

A LC network tends to absorb less power overall, and it is possible to uses a flywheel diode to make the conversion very efficient, as the coil dumps it's charge into the load when the series pass switch (ie, transistor) is off. But a feedback mechanism is still required for it to be effective, as is the case with linear power supplies.

#### cjdelphi

Joined Mar 26, 2009
272
25% duty cycle of 12volts dc... i still don't see how that can translate into anything but 12vdc if i flash a 6v bulb on and off 1 time a second it gets the full 6volts not 2volts...

how's the voltage being knocked down.....

#### rjenkins

Joined Nov 6, 2005
1,015
The PWM carrier frequency (eg. switching speed) has to be higher than the reaction time of the load.

eg. with your bulb, if you switch it thousands of times per second, the filament never has time to heat up or cool down excessively, it works *as if* it's on the average voltage.

With inductive loads, you need 'flywheel diodes' so the load current can still circulate between power pulses.

As Bill says above, if you put a flywheel diode and inductor straight on your PWM switch and add a smoothing capacitor after the inductor, you will then get DC out at a voltage proportional to the duty cycle - as long as the switching frequency is high enough. This is how one type of switched-mode power supply works.

Have a look at the link below. This is for a very simple switch mode voltage regulator which uses PWM.
There is a an example circuit on the first page. The internal power switching transistor is between V+IN & OUTPUT.
The diode allows current to circulate in the inductor when the transistor switch is off and the capacitor smooths the output & acts as a reservoir so load changes don't affect the voltage faster than the chip can respond.

The voltage sensed at the Feedback pin is used to vary the PWM duty cycle to keep the output at 5V for any load from nothing up to one amp.

The info on the first page says the switching frequency is 52KHz; this is the PWM carrier frequency.

http://www.national.com/ds/LM/LM2575.pdf

Last edited: