Flyback transformers, small, high power, where?

[Joerg]

Folks,

Got to do a flyback design, 120VAC to 24VDC or so, at about 15-20W. Will
run in full PFC fashion with the ouput voltage sloshing around, meaning
peaks more like 20W but average power under 15W. 100kHz switcher
frequency, because of a dearth of available PFC-flyback chips at higher
and because EMC above 150kHz would be tough here. Most likely we'll use
the LT3798.

Here is the pickle: The transformer cannot be more than about 250mils
high and a little over 500mils wide. Gapped ferrite gets us to less than
5W. Soooo ... got to look at newer stuff like Kool Mu, Ni-Fe High-Flux,
Sendust. Probably will come with some core loss penalty but the main
issues is that I can't find small E-cores there.

Does anyone have an idea? Does anyone know a switcher transformer
manufacturer who'd likely be willing to tackle this transformer design?

 

[Tim Williams]

Have your board guy lay out a planar transformer. Disclaimer: you didn't
specify board area per se, only height. Though you'll still have a hard
time with overall dimensions coming in under a parallelepiped of the
specified dimensions...

Don't think I've heard of Kool-Mu in planar E's.... would be cool though...

Tim

 

[Joerg]

Kool-mu toroid? Kool-mu seems to be about as good as permalloy powder,
at a fraction of the price.

Believe it or not, this time cost is not the main concern. Size is
though. If the core loss amounts to a watt or so that'll be ok.

A full PFC supply at 15 watts seems like a lot of overhead per watt to
me.

We have to do it because PF rules start to apply when you offer
installations where there's dozens of these in one location. Because of
the small space we must PFC via flyback first to get to a non-isolated
LV section as fast as possible. That LV are would contain post
converters and won't need so much in creepage distances. It'll be a
small and most of all very skinny product.

I'm buying a 65 watt MeanWell PFC supply for $16 in small quantities!

Wurth did some nice custom transformers for us.

http://www.we-online.com/web/en/pas...ics/power_transformers/Planar_Transformer.php

Good idea, although AFAIK Wuerth only does ferrites.

 

[Joerg]

Have your board guy lay out a planar transformer. Disclaimer: you
didn't specify board area per se, only height. Though you'll still have
a hard time with overall dimensions coming in under a parallelepiped of
the specified dimensions...

Don't think I've heard of Kool-Mu in planar E's.... would be cool though...

Can't do that because the other side is already full of parts and must
be low voltage. It wouldn't help anyhow because there's hardly any head
room. Plus we must remain around 100kHz.

 

[Nico Coesel]

Believe it or not, this time cost is not the main concern. Size is
though. If the core loss amounts to a watt or so that'll be ok.



We have to do it because PF rules start to apply when you offer
installations where there's dozens of these in one location. Because of
the small space we must PFC via flyback first to get to a non-isolated
LV section as fast as possible. That LV are would contain post
converters and won't need so much in creepage distances. It'll be a
small and most of all very skinny product.




Good idea, although AFAIK Wuerth only does ferrites.

They also sell inductors and transformers. I bet John bought quite a
bunch of custom transformers. Last time I asked Wurth they had an MOQ
of 20k pieces.

 

[[email protected]]

They also sell inductors and transformers. I bet John bought quite a
bunch of custom transformers. Last time I asked Wurth they had an MOQ
of 20k pieces.

They also think they're wurth a lot!

 

[Joerg]

Do you even have enough volume available to get where you want at
100kHz? ...

Not for the usual ferrite that pegs at about 0.5T. Other more modern
stuff can do 1.5T or even better. Most of that is distributed gap.

... There's only so much energy that you can store in a cubic
millimeter of core (or air gap), and that energy times your switching
frequency is your power. If you want to violate that constraint you
don't need to come to us, you need to explain to God how he got the laws
of physics wrong.

As a kid I've redlined an iron powder core in a balun because I didn't
have enough money to stack two of them ... *PHOOMP* ... thwack ... and
the antenna lay on the ground. Now I was down to zero cores :-(

But back then dash-2 from Amidon was the best money could buy for use at
several MHz. It's been over 30 years.

You quote height and width -- does that mean it can be a foot long? Why
can't you take four or five of your 5W gapped ferrite transformers and
string them in parallel?

Not that way, because then you have to keep them at a distance to
fulfill creepage path regs. The core could be longer, within reason.
Like an inch or maybe more. But unfortunately only lengthwise, the
contacts can't be on the sides for discharge clearance reasons.

I don't have anything directly to help you, other than the observation
that you may be up against fundamental physics, and therefore you may
need to apply some of your cleverness in the direction of getting your
frequency up, rather than trying to stuff an impossible energy density
into whatever materials are available here in the real world.

The reason I am asking here is that there are materials with a
saturation flux density several times higher than ye olde ferrite. So
I'd like to see if a smaller flyback xfmr can be build with any of
those. From what I understand the core losses would be somewhat higher.
If that's not a whole lot of added loss it may be an option.

(I'd be looking at a microprocessor with PWM and ADC, but that's me, and
it doesn't necessarily mean that I'd come out the other side of the
exercise smelling like a rose, either. ADC -> software -> PWM -> gate
driver = joy. Or maybe lots of smoke).

uCs are very poor switch mode controllers. Every time that was done
(usually against my advice) they couldn't get the loop bandwidth high
enough because the thing plain ran out of MIPS. I have to live with
what's there in terms of ICs.

 

[Joerg]

They also sell inductors and transformers. I bet John bought quite a
bunch of custom transformers. Last time I asked Wurth they had an MOQ
of 20k pieces.

That MOQ is ok (after sampling), but their flyback stuff is huge, all
regular ferrite.

 

[Joerg]

I'm not sure what's your "lengthwise". Could you stack E-cores? Clearly
you won't get a bobbin if you do that, but if the $$ is there it may work.

We may have to stack. Ugly but ... man's gotta do what man's gotta do.
Problem will be to cajole a xfmr mfg into doing that. They are usually
very reluctant when it comes to unorthodox stuff.

BTW, table 3 in here is why I think there may be a change to shatter the
old size rules:

http://powerelectronics.com/passive...ite-inductor-cores-pol-converters/index2.html

Well, there is that. Hey! Get Jim to make you a custom chip!

There's also EMC. If you get above 150kHz all hell breaks loose in terms
of conducted EMI. The filters will easily eat up all save real estate.

 

[legg]

Believe it or not, this time cost is not the main concern. Size is
though. If the core loss amounts to a watt or so that'll be ok.



We have to do it because PF rules start to apply when you offer
installations where there's dozens of these in one location. Because of
the small space we must PFC via flyback first to get to a non-isolated
LV section as fast as possible. That LV are would contain post
converters and won't need so much in creepage distances. It'll be a
small and most of all very skinny product.

The creepage doesn't change until you've successfully isolated the
source - it's all hazardous until you provide a reinforced isolation
barrier.

If you're doing this in two stages of basic isolation, then the
line-input PFC section will need to be basic-isolated from the 24V
output, as a first stage.

Telecom stuff does this - basic isolation to battery bus, then basic
isolation to user ground-referenced outputs.

RL

 

[legg]

Can't do that because the other side is already full of parts and must
be low voltage. It wouldn't help anyhow because there's hardly any head
room. Plus we must remain around 100kHz.

This is unfortunate, because using the printed wiring is the most
commonly used method of jumping the low profile power barrier and
reducing or distributing surface temperature rise.

Your LT3798 app gives a table of parts illustrating what happens using
conventional shapes. None claim a power handling density greater than
3w/cm^3. You're asking for 15W/cm^3.

RL

 

[Nico Coesel]

We may have to stack. Ugly but ... man's gotta do what man's gotta do.
Problem will be to cajole a xfmr mfg into doing that. They are usually
very reluctant when it comes to unorthodox stuff.

In my experience I always had to do (most of) the transformer design
myself when I needed a custom transformer.

 

[Joerg]

This is unfortunate, because using the printed wiring is the most
commonly used method of jumping the low profile power barrier and
reducing or distributing surface temperature rise.

It isn't a popular method because that way you cannot reasonably get the
number of turns for a 180V peak primary. The only time I have seen
planar transformers is in signal isolation or pulse transformers.

Your LT3798 app gives a table of parts illustrating what happens using
conventional shapes. None claim a power handling density greater than
3w/cm^3. You're asking for 15W/cm^3.

In the core, yes.

 

[Joerg]

The creepage doesn't change until you've successfully isolated the
source - it's all hazardous until you provide a reinforced isolation
barrier.

Sure, that's why we want to do the flyback-PFC. Because that puts the
iso barrier as far forward as possible. The bulk caps are then on the
non-mains side which is a huge advantage versus boost-PFC when real
estate is of the essence.

If you're doing this in two stages of basic isolation, then the
line-input PFC section will need to be basic-isolated from the 24V
output, as a first stage.

Telecom stuff does this - basic isolation to battery bus, then basic
isolation to user ground-referenced outputs.

This is a hi-rel app so we have to be extra careful. Longer creepage
paths and so on. So we want to minimize the primary side circuitry.
Ideally without having to go above 150kHz.

 

[Joerg]

In my experience I always had to do (most of) the transformer design
myself when I needed a custom transformer.

Sure looks like it here as well :-(

But then where is the value added by domestic manufacturers? Then we
might as well have it produced in Asia.

 

[Joerg]

I've seen it done for 48 -> 12 power conversion. (something like 20W, I
think). It worked pretty good, except for the turns miscount on rev 0
and the itty bitty mod wires snaking under the core that resulted.

It can be done but it's expensive. 8-10 layers, board becomes brittle if
skinny, and so on. It does look elegant though.

It worked because we had a muy multi-layer board in which to pack the
windings -- I don't know how it would have flown if we were dealing with
higher than 48V.

And probably at a much higher frequency than 100kHz.

 

[notbob]

Here is the pickle: The transformer cannot be more than about 250mils
high and a little over 500mils wide.

Don't most CRTs have flyback xfrmr? Look fer one in an itty bitty tv
or o-scope. I've gotta tektonix 213 dmm w/ about a 2" CRT. The
flyback hasta be pretty small. (shrug)

nb

 

[Spehro Pefhany]

It isn't a popular method because that way you cannot reasonably get the
number of turns for a 180V peak primary. The only time I have seen
planar transformers is in signal isolation or pulse transformers.

They are used more often these days. For example, the Excelsys series
of modular supplies use planar transformers with mains volage
primaries. A bit pricey, but they have available good "medical"
isolation (4kV) high-rel and low noise options.

Here's the output module with the planar transformer:
http://media.digikey.com/Photos/Excelsys Tech Photos/POWERMOD SERIES.JPG

I have not attempted to determine how many layers are in the board.

 

[legg]

Sure, that's why we want to do the flyback-PFC. Because that puts the
iso barrier as far forward as possible. The bulk caps are then on the
non-mains side which is a huge advantage versus boost-PFC when real
estate is of the essence.

So the 13mmx13mmx6mm, 15-20W flyback transformer also requires
reinforced insulation?

RL

 

[Joerg]

So the 13mmx13mmx6mm, 15-20W flyback transformer also requires
reinforced insulation?

It doesn't follow the usual standards, this is a special hi-rel app. We
need certain breakdwon limits and creepage paths for everythin
line-side. Don't have them in my head right now but more than 0.150" to
chassis, for example. This goes into a narrow tube so that presents a
real estate issue.

The core itself can be on the isolated (LV) side.