Flyback winding questions

[abstruse1]

Greetings.  I'm using FBs to power CW multipliers and want to make a more rugged FB.

As to winding a flyback primary:

1.  How important is it that the primary is wound tightly around the ferrite?
2.  How important is it that the primary turns are right up against each other?
3.  Has anyone noticed a difference between using solid magnet wire v. stranded wire?  Any benefit to using Litz?

As to the secondary:

4.  Is much lost by using a less than perfect wire laydown (giving lower fill factor)?
5.  Do you lose much with a segmented (many grooves) bobbin?
6.  Any guidelines as to thickness of bobbin core (the part that goes around the ferite)?  Or the gaps between winding groups in case of a segmented bobbin?

Overall:

7.  There are many different kinds of ferrite.  How important is getting the best one?
8.  As to the gap in the ferrite, how wide should it be?  Does it matter if it's under the primary or secondary?

BTW, I'm in the habit of putting coils under oil in a vacuum chamber overnight, at ~28" Hg.

Thank you.  I genuflect to your knowledge.

[AstRii]

Those are all interesting questions, I will try to answer those for which I know the answer.

"3.  Has anyone noticed a difference between using solid magnet wire v. stranded wire?  Any benefit to using Litz?"
This depends on the frequency of the AC supply. Due to skin effect, litz wire will have lower resistance as the frequency goes up. Usually either the frequency is low or the flowing current is not that big to notice much difference between magnet wire and litz, but when I designed my SSTC II which was oscillating at 300kHz+ with 20A of switching current, the solid wire was getting very hot even though it was massive (5.5 millimeter diameter). With litz wire the losses would be much better.

7.  There are many different kinds of ferrite.  How important is getting the best one?
Some ferrites are not made to work with high frequencies, there are even audio ferrite cores which have high losses at higher frequencies. Though vast majority of ferrite cores will handle high frequencies just fine. Some materials will give better performance than others. You can always check the datasheet and make the calculations to see for yourself when will the core saturate and how far can you push it.

8.  As to the gap in the ferrite, how wide should it be?  Does it matter if it's under the primary or secondary?
This depends on the coupling you want. If you need coupling close to 1 you should not use any gap at all. For lower coupling factors you make the gap wider. Keep in mind that even a 0.1mm gap makes a big difference.
And no it doesn't matter where is the gap but usually you split the gap equally between primary and secondary.

Now the questions I'm not sure about:

1.  How important is it that the primary is wound tightly around the ferrite?
In a solenoid all the magnetic flux should be inside the coil so I assume it doesn't matter that much if there is a gap between the core and the windings, as the magnetic field lines should close around the core.
And if there is any leakage I think it won't be really noticeable.

2.  How important is it that the primary turns are right up against each other?
Not important, but with gaps between primary turns you're loosing precious space for more turns or thicker wire.

[abstruse1]

Marek, thank you for your careful answers.

I'm starting to wind a FB secondary tomorrow.  It will be on a divided bobbin with about 6 grooves and I'm using 24AWG wire since it's easier to work with than smaller diameters.  Is it true that there's no reason to use smaller rather than larger wire as long as you can get enough turns with a larger wire?

[davekni]

Is it true that there's no reason to use smaller rather than larger wire as long as you can get enough turns with a larger wire?

I think generally true, although there can be a couple small disadvantages:
1) Windings get farther from the core, so leakage inductance goes up.  Some leakage inductance can be useful for ZVS driving, so not necessarily bad.
2) At some frequencies, larger wire diameter can actually have higher resistance when wound into a multi-layer coil, since proximity effect increases.  Don't know if you would be in this realm, but it's not likely to be a huge effect.  Could have lower resistance instead depending on frequency and geometry.  Would need to run a FEMM simulation of your specific case to know.
Nothing else is coming to mind.  Hopefully someone else will reply if I'm missing anything obvious.

[abstruse1]

ASTRii, you wrote:

8.  As to the gap in the ferrite, how wide should it be?  Does it matter if it's under the primary or secondary?
This depends on the coupling you want. If you need coupling close to 1 you should not use any gap at all. For lower coupling factors you make the gap wider. Keep in mind that even a 0.1mm gap makes a big difference.
And no it doesn't matter where is the gap but usually you split the gap equally between primary and secondary.[/color]

Could you (or anyone) explain the "...coupling close to 1..." part?  I've read a bit about transformer coupling, and it sounds straightforward in theory, but I have no idea how it affects the output of my flyback.  Thanx.

[johnf]

Coupling close to 1
or K as it is often referred to. The perfect transformer has a coupling ratio of 1 ie all the input energy is translated to the output. In practice we can get very close ie coupling ratio of 0.96 is as good as I have measured mine own designs. To measure this one measures the inductance of the primary with the secondary open circuit and again with the secondary short circuit
So coupling factor k = 1-(shorted inductance divided by the open circuit inductance)
Lets say the primary measures 10 micro-Henrys open circuit and 100 nano-Henrys short circuit  = 1-(0.1/10) = .99
The measured value of the shorted inductance is called the leakage inductance
As coupling factor goes down the leakage inductance goes up which with flybacks manifests itself as high reflectance from the secondary to the primary and this energy gets absorbed by the snubbers on the switching device with out the snubbers your switching device will see much higher voltages than your DC bus

[klugesmith]

The coupling question came up in context of air-gapping the core.

If you are making a flyback converter (the SMPS topology, not the physical configuration)
then ideal coupling factor K = 1, and there is no benefit from reducing it on purpose.   (And short airgaps don't greatly reduce the coupling.)

Typically, the core needs an air gap in order to store enough energy in primary inductance.
The gap reduces inductance, for a given primary turns count.
But the gap greatly increases the volt-seconds and ampere-turns that can be applied without saturating the core.   
If you consider the energy (0.5 * L * I^2) to be stored in magnetic field (0.5 * B * H), it's mostly proportional to airgap volume and Bmax^2.   Only a small fraction of the energy is in magnetized ferrite.

[davekni]

So coupling factor k = 1-(shorted inductance divided by the open circuit inductance)
Lets say the primary measures 10 micro-Henrys open circuit and 100 nano-Henrys short circuit  = 1-(0.1/10) = .99

I think you are missing a square root.  k = sqrt(1 - shorted_inductance/normal_inductance), so about 0.995 for your example.

If you are making a flyback converter (the SMPS topology, not the physical configuration)
then ideal coupling factor K = 1, and there is no benefit from reducing it on purpose.   (And short airgaps don't greatly reduce the coupling.)

Agree.  Especially for conventional SMPS flyback topology, leakage inductance wants to be minimal.  Any energy stored in leakage inductance is dumped into snubber R/C/D network each chop cycle at switch turn-off.  K not changing significantly depends on how it is viewed.  Leakage inductance doesn't change significantly.  If a gap reduces normal inductance by 10x, K gets 10x farther from 1, say from 0.999 to 0.99.  Small absolute change.

For ZVS driven transformers, K is better at 0.85 or less.  Avoids ZVS oscillation from dropping out at mid-loading due to low Q.