What does Interleaved mean when talking about SMPS.

[si2030]

Hi there,

This thread is focused on the above concept however i have three questions - the first in two parts. I have heard the term "interleaved" used a fair bit around the internet in regard to the area of SMPS's. What I am unsure about it just exactly what it means...



1. My basic understanding at this point is that it is used for basically two purposes:

A: To reduce the ripple output by (from what I understand) creating an intermediate peak between the two peaks made up of a rectified sinwave. So you achieve (say with a 90 degrees lag from the second transformer) four peaks "interleaved over each other... In a high frequency stepdown SMPS this means that the filter capacitor has another two peaks to fill it or a peak every 90 degrees of the sine wave signal - four peaks rather than two in a single Hz cycle....

B: the second reason is to increase the power of the system as a whole... so given a maximum core size on hand you can, instead, have two or even three of these smaller transformers and you roughly triple the power output... e.g. three smaller transformers rather than making one bigger transformer... by interleaving them (in this scenario applying it to the simple topology of high frequency, off line and voltage transformation from 240V to say 30V) while reducing the ripple (120 degree lag for each transformer). Am I on the right track...



2. Further I have heard of interleaving the secondaries on a single transformer... Why do this and would this not affect affect (complexity) whats happening in the coil (If they do this).



3. Finally, What is Discontinuous Mode compared to continuous mode and why??
( this might have something to do with Buck or Boost converters rather than a line frequency power transformer with which I intend to work with)



Would be really good to clear up my confusion on this... Will appreciate any input here..

Kind Regards

Simon

 

[colin55]

Are you talking about interleaving the laminations?

 

[Diver300]

Well this thread has managed to find three uses for the word "interleaving" in just two posts!

1) As I understand it, the buck regulators that supply the processors on PCs are run like a multiphase supply, so that they all produce current at different times. That will reduce the output ripple, the size of capacitors needed, and the size of the input current peaks.

2) With the very small number of turns on the transformer of an SMPS, there can be a real problem with stray inductance. That is the part of the inductance of one winding that is not linked to the other windings. The stray inductance can be reduced by winding half the primary, then the secondary, then the rest of the primary. It reduces the distance of the furthest part of the primary from the secondary.

3) Switch-mode supplies use ferrite cores and so don't have laminations, generally. The same applies to toroidal transformers. The rectangular transformers where the windings are wound onto a bobbin, and then the iron core is assembled usually have laminated cores. The cores have to be split so that they can be assembled around the bobbin. A common arrangement is EI where the lamination form a capital E and a capital I. If all the E laminations are stacked together, and so are all the I laminations, there is the possibility or a gap going the whole way through the transformer. Instead, the laminations are interleaved, and assembled EI on odd layers, then I3 on the even layers. **broken link removed**

 

[si2030]

Diver300,

So as I understand from (1) in your reply the system does indeed contain multiple SMPS transformers in a multi phase format. Would I correct in inferring that this multi transformer, multi phase setup will also work with the topology "high frequency, off line and voltage transformation from 240V to say 30V" as well as the described Buck topology?

Seems, with regard to number (2) that this is a good idea for winding any SMPS transformer using a strip (length of core window) and interleaving the multiple windings of the primary/secondary to minimise stray inductance.

 

[Ubergeek63]

Diver...

you get 50% on #3. laminations are used whenever the core is slightly conductive.

cores are split for ease of assembly, bobbins are just another ease of manufacture thing.

there are laminated toroids... they are wound from metal straps and epoxy coated.

Dan

 

[si2030]

Diver300,

So as I understand from (1) in your reply the system does indeed contain multiple SMPS transformers in a multi phase format. Would I correct in inferring that this multi transformer, multi phase setup will also work with the topology "high frequency, off line and voltage transformation from 240V to say 30V" as well as the described Buck topology?

Seems, with regard to number (2) that this is a good idea for winding any SMPS transformer using a strip (length of core window) and interleaving the multiple windings of the primary/secondary to minimise stray inductance.

Wondering if someone out there might confirm this or correct it if possible...

 

[ronsimpson]

There are systems where multiple SMPS operate out of phase. There are controller IC that drive 3 sets of (MOSFETs+inductors) and have inside 3 PWMs that run 1/3 out of phase. There are many benefits.

In side of a transformer; Most often the primary is wound first, then insulation, and then the secondaries. Interleaving involves winding the primary in sections along with the secondary.

Example: 1/2 Primary, Secondary, 1/2 Primary
PPPPPPPPPPP
SSSSSSSSSS
PPPPPPPPPPP

Example:1/3P,1/2S,1/3P,1/2S,1/3P
PPPPPPPPP
SSSSSSSS
PPPPPPPPP
SSSSSSSS
PPPPPPPPP

 

[si2030]

So the upshot of this is say:

6 Mosfet drivers, three separate high frequency transformers being run 1/3 out of phase with interleaved half primary, Secondary, Half primary and three bridge rectifiers feeding into a single filter capacitor.

Means lower filter capacitor for the combined output, lower current peaks and smaller transformers - a third the size of the a single unit so more manageable. Also reduced ripple as you have 6 ripple peaks for the same Hertz cycle instead of two.



Feeling much better about the whole area... Very many thanks for this Ron



I did have another different question I asked above... will re-ask it to complete the reduction in confusion

What is Discontinuous Mode compared to continuous mode when considering interleaved (multiple transformers) and why use it (or not)??
( this might relate to Buck or Boost converters rather than a line frequency power transformer with which I intend to work with)

 

[MrAl]

Hi,


It's true that the laminations (if using them) can be interleaved, EI and then IE, but also the interleaving can vary as to the number of laminations in each group...
Laminate interleaved "1x1" means use one set of EI and then the next set IE, and then repeat.
Laminate interleaved "3x3" means use three sets of EI and then three sets of IE on top of that, then repeat.

To keep the leakage inductance down the best bet is to use bifilar windings. That's where the primary and secondary are wound at the same time. You get a 1:1 ratio that way, but if you wind trifilar you can use (say) two windings for secondary and one for primary creating a 1:2 transformer.
Bifilar winding puts the secondary right next to the primary on every layer:
SPSPSPSP layer 1
SPSPSPSP layer 2
etc.

Alternating primary and secondary windings would give better coupling too but also provide for better secondary to primary electrical isolation:
PPPPPPPP
IIIIIIIIIII
SSSSSSS
IIIIIIIIIII
PPPPPPPP
IIIIIIIIIII
etc.
where IIIIIII is a layer of insulation like tape.

 

[si2030]

thanks MrAl, Great answer that complements Ron's answer..

Also added a comment on your reputation if that's OK.

Simon

 

[MrAl]

Oh ok you're welcome, and thanks for the nice comment.