# Flyback converter - winding order

• posted

When I did that sort of thing I always preferred split bobbins, so that there was definite insulation between primary and secondary... but I was designing for a commercial application that had to pass VDE.

...Jim Thompson

```--
|  James E.Thompson, P.E.                           |    mens     |
• posted

It can matter tremendously, due to proximity effect. For instance, in theory a 2-layer winding can have 6 times!!! the I^2R losses of a single layer due to proximity effect. In practice this will be less because of round conductors and insulation thicknesses (the initial theory considers rectangular foil conductors infinitesimally close together.)

But the point is that it can matter a great deal.

There seem to be several approaches:

1. Ignore the proximity effect, design based on only considerations of AC resistance of wire not in proximity to other conductors. Then you can test it, and it will likely be less efficient and get hotter than you predicted, but if it's good enough move on. At least you'll know why it's the way it is.

1. Consider the proximity effect qualitatively, but don't treat it in mathematical detail. Thus you will use rules of thumb to minimize it. For instance:

a) use a long bobbin instead of a deep one, so that the number of layers stacked is minimized. b) interleave the windings to as much a degree as possible.

1. Fully treat the proximity effect. (Ouch!) I have been reading how to do this in the following text, which covers the subject in grueling detail:

Robert W. Erickson and Dragen Maksinovic, "Fundamentals of Power Electronics, 2nd ed."

It's quite overwhelming. I have gotten about a 33% grasp of it qualitatively after several skims through the sections about proximity effects.

I am still mired in my basic EM text trying to fully grasp the mechanics of the skin effect for isolated conductors.

Note that the winding approach that may minimize proximity effect might not be the best one for winding capacitance, though it appears well interleaved windings are advantageous for both leakage inductance and proximity effect minimization. For a flyback with the step-up ratio and power level you are implementing, winding capacitance isn't a big deal.

Good luck!

```--
_______________________________________________________________________
Christopher R. Carlen```
• posted

I've designed a 2-watt flyback converter, 12V to 200V, by using a round RM5 ferrite core with air gap. I wonder if it is necessary to first wind the primary and then the secondary on top, or does it make a difference if I do it vice versa, with the primary on top ? For example could the efficiency be higher in one way ?

Any idea ? Thank you. Jacques

• posted

IME, flyback converters that use the feedback winding to regulate just won't work acceptably without the windings on top of each other- preferably with one winding sandwiched between two halves of the other or intermixed. ISTR a degradation from 5% regulation to 30 or 40%.

Best regards, Spehro Pefhany

```--
"it's the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com```
• posted

I do not think it matters - often one interleaves the windings for better coupling thus reducing leakage inductance/losses. In your case, it probably does not matter - you can try it, though.

• posted

In the two-section configuration, there is a difference in winding length that has to be taken into account when determining copper loss. Some topologies have current stress that is more severe on one winding than another - flyback isn't one of these. Although HV windings usually have a poorer copper fill factor, LV windings in smaller parts can be just as bad, due to conductor dia/winding breadth ratios.

Hazardous windings are often more simply isolated when the bobbin or ground insulation serves a dual purpose for the first section - reducing functional safety requirements for the outer wrap on non-hazardous layers. With a 200V secondary, this might be a consideration.

Noisy layers in any section are also more easily screened if they are buried - the core and quieter windings can be employed as screening elements. When buried, their capacitance to other layers and the core are agravated, however.

RL

• posted

For single output designs, I usually wind the primary first. The Start (Nearest the core) layer is the switch end. The second (third) layer is the input voltage. This gives the lowest primary capacitance, since the former is fairly thick.

If you can, it is usually best to mix the windings together to reduce leakage inductance.

```--
--
kensmith@rahul.net   forging knowledge```
• posted

In article , Jim Thompson wrote: [...]

If you have to pass VDE, you have to live with the leakage inductance. If not, Putting the windings as close together as other considerations allow is the way to go on a flybacker.

Any energy that goes into the leakage inductance must be dealt with. With a slightly tricky snubber, you can put the energy back onto the input rail. You still loss energy in the switch sloshing the power around.

```--
--
kensmith@rahul.net   forging knowledge```
• posted

I know of a flyback converter that used a small 4 fet bridge input, full square converter to take the current from a diode snubber and put that energy into the secondary side, with the output voltage controlling the DC bus the snubber diodes dumped into, by the turns ratio on that converter. Very low loss for a big flyback unit.

```--
John Popelish```
• posted

Hi Ken, Would you enlighten us on that tricky snubber? Sounds very interesting. Resistors don't slosh well. Regards, Harry

• posted

[....]

Basically this is it:

Vin ---------+------------ ! ! ! ) --- ) ^ D1 ) L(working) ! ) ! ) ! C1 ! +----!!------+ ! ! --- ! ^ D2 ! ! ! ) ! ) Switch ) ) L(Snub) ) ! GND

When the switch turns on, L(snub) and C1 go sprong but not much energy is transfered anywhere. The end of C1 near the diodes ends up with a small positive voltage on it.

When the switch turns off, the rise is caught at 2*Vin minus a bit and L(Working) charges C1 from there.

D1, and D2 have to be fairly fast diodes a simple 1N400X won't do here.

```--
--
kensmith@rahul.net   forging knowledge```
• posted

Please be reminded that the OP is discussing a 2W flyback converter with a 12V primary.

RL

• posted

Sure - Still might be worth the bother if the OP is running off batteries or something.

• posted

....at 12V, 2W... in % loss, yes, in w's: nada.

Initially, I still would not care unless running off battery or in a tiny box with no dissiation capability.

I think that you can wind and test a great deal of winding configurations before you get a decent mathematical model in place - that is my experience, anyway ;-)

This is 3-D electromagnetic stuff needing either:

A hefty understanding of Maxwell, Arcane Fortran Libraries and Numeric computing (and no Life),

A hefty purse for licensing a commercial software package containing the above.

Induced currents from changing magnetic field surrounding conductor adds to the current in the inductor - changing said field (and current distribution in inductor), What's so hard about that ;-))

• posted

In article , legg wrote: [....]

Are you trying to imply that overkill is a bad thing?

```--
--
kensmith@rahul.net   forging knowledge```
• posted

When we vary from the specific, the likelihood of wasting time is increased. His inquiry was pretty specific.

Designing for ruggedness can make sense. Overkill is just overkill.

RL

• posted

And although its only a 2W smps, if he dumps all the leakage into a SOT23 "zener" it could still get pretty hot (it doesnt take many watts at 300K/W), so reducing leakage by interleaving is probably warranted.

Unless I have a requirement for poor coupling I always interleave windings. Once you learn how to do an MMF plot the interleaving becomes self-evident. Winding asymmetries are easy to compensate for too.

Cheers Terry

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.