multiphase switching advantages

Mar 26, 2008 7 Replies

Hi,



Is there any advantage to doing multiphase switching (ie. a



3 phase stepdown buck) with synchronous switching, or is it only advantageous to use interleaved switching?

I know the input and output ripple will be reduced only with interleaved switching, but I am not sure if the inductor requirements will be lessened with synchronous switching or not. I was thinking that since an inductors size is proportional to inductance*current^2, that a 3 phase synchronous or interleaved switcher would have 1/3rd the overall inductance capacity requirement of a single phase switcher.



Is this correct? I know if you parallel inductors the inductance goes down but I am not sure if that is important here.



cheers, Jamie


Hi Jamie,

It sounds like you mean "Synchronous Rectification" - using a FET rather than a freewheeling diode.

the 2 issues are entirely separate. SRFETs typically have a lot less conduction loss than diodes. And for moderate currents and voltages you can get stupidly low Rdson, so if your budget allows you can buy "negative watts". Its surprising how big a rectifier you can heatsink to a 4" square PCB.

staggering the phases of paralleled converters is also advantageous. One way in which it is useful is to give significant reductions in output capacitor ripple current - this can be flipped around, and used to allow very low inductances (with large ripple) which are physically very small, without taking a hit on output ripple voltage. this doesnt work so well when the inductor current goes discontinuous.

SRFETs ensure your inductor current is always continuous, so is ideal for use in polyphase converters, as it ensures the output ripple current always cancels.

and as an added advantage, the ripple frequency is

Fripple = Fswitching*Nphases

bucks always have discontinuous input current. the net result of N phases is the input capacitor ripple current goes up as 1 x sqrt(n) rather than 1 x n.

The input ripple frequency is also Nphases higher

The two are basically the same thing. Interleaved switching is what multiphase converters do and single phase ones don't.

There are multiple advantages:

Both the input and output ripple are shifted up in frequency by N.

The amplitude of both ripples are reduced by N or more.

The ripple in the ground plane is reduced by a lot more than N and this can help a lot if there are analog circuits sharing this plane.

The Nyquist of the control loop typically is increased by N.

The EMI radiation may be reduced by as much as N.

With the sharing of part of the working inductor, a slightly smaller total amount of core may be needed. The shared part needn't be as good at the high harmonics as the seperate parts, but this is too little of an effect to matter.

Some other advantages not mentioned -

- power capability and consumption are directly related to nphase giving a possible difference in efficiency over the load range.

- differences in inventory, parts count, reliability, redundancy, thermal distribution, mechanical profile, multiple-source availability, assembly method.

All either pro or con, depending on your specific application and your point of view.

RL

What about a multiphase converter that does synchronized switching? Would that be the same performace as a single phase converter? (ie. no reduction in the input/output voltage ripple)

I am just curious if there is any benefit at all for a multiphase converter that has synchronous switching (not synchronous rectification) over a single phase converter.

cheers, Jamie

Please explain exactly what you mean by synchronous switching.

If you mean all the devices switch at the same time, the converter is not multiphase.

If you mean synchronous with some external pulse then multiphase can have a large advantage there. It is a fairly common trick to time the DC-DC converters switching at the point where some other part of the system doesn't mind the noise so much or to make the DC-DC converter's noise alias to a frequency you don't care about in an ADC. In these cases, a multiphase converter can make things a lot easier by allowing the DC-DC converter to run at a lower frequency.

Yes that is what I meant, all devices switching at the same time. I just was wondering if the overall inductor core volume would be lessened in that case rather than a single switch, since the separate inductors for 2 phase for example would each carry half the current and inductor volume increases with current^2.

cheers, Jamie

That isn't multiphase. It is just paralleling smaller converters. The core requirements etc are not any easier to meet except you could perhaps a practical consideration. If you can find the smaller cores that exactly meet the requirements but at three times the size you can't and have to go up to a bigger model then there is an advantage.

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required