Leakage inductance of "short spindle" flyback transformers is high?

[Flyback]

Hello,
My boss wants us to design an offline (230VAC) flyback (vout = 30v, Pout=40w) switch mode power supply. It will be sited inside a flat screen television, so therefore it will feature a very low profile flyback transformer, and therefore the ‘core spindle’ of this transformer will be very short. Is it true that the leakage inductance is always greater with such transformer profiles?.
(On other power supplies, I have previously always used long spindle transformer cores in the belief that this allowed better coupling of primary and secondary, and thus less leakage)

 

[ronsimpson]

Leakage inductance has to do with how close the two windings are to taking up the same area in space.
Adding a large amount of isolation (primary/secondary) will add leakage inductance.

 

[Dick Cappels]

If you provide an illustration, it would help us answer your question. Generally, (as ronsimpson indicated) leakage inductance is a function of how well the two windings are coupled.

 

[Nigel Goodwin]

Why not take an existing TV to pieces, and see how it's done? - both transformers and the larger electrolytics are commonly mounted THROUGH the board in order to reduce width.

 

[ronsimpson]

Flyback,
Look for DER276.PDF.
It is a design for a 40W switcher including transformer design.
You know how to change from 12V to 30V.

 

[Flyback]

Thanks, DER276 is good but only says that the transformer there is RM10 core. The transformer spec does not give a maximum value for leakage, so in my view, DER276 is incomplete in terms of transformer spec.

https://ac-dc.power.com/design-supp...s/der-276-high-efficiency-40-w-standby-power/

 

[kubeek]

Page 14 says it has 15uH primary leakage inductance. You even got there the actual Al value for the gapped core and primary inductance and whatnot. If that and the dozen pages just about the transformer after that are not enough for you, then I don´t know what is.

 

[ronsimpson]

Power Int does a very good job on application notes.

 

[Flyback]

thanks, though now having looked into it, that RM10 core isn't particularly "short and stubby". They only had to use (pretty much) no more than 1 full layer for each of the primary halves, so the bobbin is nice and wide. They could have worked the design to get exactly full layers of turns and not full layers plus a few extra turns, which makes manufacture that bit less simple and more error prone, plus less repeatable meaning leakage paramemeter is less near "set in stone" than it could be .
Also, RM10 from ferroxcube comes with a large number of off-the-shelf pre-gapped variants, so I am not sure why they chose a custom gap instead. They could easily have worked the design to accommodate one of the off-the-shelf supplied gapped rm10 core sets. This seems to be a feature of having a software program to calculate the transformer parameters...the software isn't practical enough to realise these things.

They got a k value (coupling) 0f 0.9945, which is very good as you know.
RM10 CORE DATASHEET:
https://www.ferroxcube.com/FerroxcubeCorporateReception/datasheet/rm10i.pdf

The effective path length in the RM10 is 44.6mm, so on page 17 of DER276, I don't know where they get the path length value of 3.39cm from?

DER276:
https://ac-dc.power.com/sites/default/files/PDFFiles/der276.pdf#pdfjs.action=download

 

[Blueteeth]

If you're after low profile, then the EFD series would be prudent, although it is rare to see these for >20W throughput. I suspect these have higher leakage inductance given the flatter core, meaning the top and bottom of the coil (horizontal in the EF series) are much wider than the sides (where the magnetic path is) compared with standard EF. I haven't seen half cover cores for these, as I guess it would negate the 'low profile' aspect of EFD, only to decrease leakage inductance slightly. There are of course very low profile EFD transformers, but rarely for flyback where leakage inductance isn't wanted, such as current fed royer converters used in CCFL drivers. If it was commonplace to have <10mm profile 40W+ offline PSU's, I'm sure Apple would have slapped one in an aluminium covered plastic case years ago

I cannot say I am seasoned in designing offline power supplies, I have only made a handful - but plenty of DC-DC, and these have not been tested for safety or reliability, but I believe you will be hard pushed to get the height of your board under 15mm. A quick google put the RM10 'low profile' core at ~13mm height, with 1.6 for standard PCB stock, that's your ~15mm, and after reading that app note, it looks like an ideal place to start. That version of the RM10 also has a 33.9mm path length:

https://www.farnell.com/datasheets/1571239.pdf (page 9).

I suspect they used the N87 type.

Standard RM10 will indeed have a ~44m path length, with a height of 18.5mm, as these are much more common, with gapped cores available for 400 - 630 Al (datasheet was 594 eh?) it should be much easier and cheaper at the cost of 5mm in height. I still don't know how people grind cores for gaps for one-offs. Adding spacers might be ok for E series, but not RM's..

 

[CapeCAD]

Have you considered a planar transformer design?

 

[ronsimpson]

Have you considered a planar transformer design?

I have never used planar for 220VAC switchers. Too many turns.

 

[CapeCAD]

I have never used planar for 220VAC switchers. Too many turns.

We use them all the time, especially when height is an issue.

 

[ronsimpson]

I know it is lots of work but:
can you show pictures of the PCBs area inside the transformer? How many layers.
I have only done low voltage planar transformers.

 

[CapeCAD]

I'll look for a good example tomorrow (Designs are on my other computer). High voltage is much easier than high current since you can use a lower copper weight and fit more turns per layer.

I've designed planars as high as 24 layer 3 Oz copper or 12 layer 6 Oz copper. This design could be done with 1 Oz or 2 Oz copper, since it's low current, depending on the turns ratio and core window chosen.

 

[ronsimpson]

CapeCAD,

a) "24 layers 3 Oz/12-6 Oz" There is part of my problem. I was using 4 layer 1 Oz because of cost.
b) My CAD did not like making 3 turns or more turns /layer. Real fight. One turn is easy!
c) I have amps flowing.

So I made one turn/layer. The winding area is 95% copper in each layer.
The main PCB had 4 turns. Then there are two break off crackers each with 4 more turns each. One cracker sits above and the other below the main PCB. The core holds all together. Stakes go through all three PCBs and are soldered in place. There are 3x4 turns (6T primary +6T secondary) Really it is 3T ct 3T to 3T ct 3T for noise consolation. The current runs through 2.5T--ct--2.5T to 3t--ct--3T. I know some people are really against half turns but I never had a problem.

It is a problem when the Gods declare the power supply shall not be taller than any other part. (2000V of isolation)

 

[Flyback]

Thanks,

On similar lines, we now wish to wind a transformer for a Flyback on a ETD54/28/19 core with 1mm gap (N87 material).
We have a split output rail of +/-40V, 4A. We will thus have two secondary windings, and these will be interleaved between the primary which will be split into three equal layers.
So we have a primary of 30 turns total, and a split secondary of 22 turns total. Each winding is made of three paralleled strands of SWG21 enamelled copper wire.
So we have this interleaved winding structure ,working from the core outwards.....

10 primary turns
11 secondary turns
10 primary turns
11 secondary turns
10 primary turns

We must have UK mains isolation between primary and secondary. (so 3mm margins)
The flyback is 330W peak, but average power is just 40W. (Guitar amplifier power supply)

Do you think we will be able to get a coupling factor of at least 0.994 with this interleaved winding strategy?
We fear that the 1mm gap size is just too wide to get this low leakage inductance?

The windings are each about 2.7mm wide (3 paralleled SWG21 strands) and this is wider than the core gap size so therefore we believe that this will help to reduce leakage inductance?


Epcos ETD54/28/19 core datasheet..(N87 material with 1mm gap)
https://www.italtras.com/filePDF/etd_54_28_19.pdf

 

[ronsimpson]

Flyback,
Look at triple insulated wire. teoflon
furukawa or rubadue I have used both. I have been to Rubadue.
Magnet wire "has no isolation". You can not wind to the edge of the transformer.
If you use triple insulated wire you can wind to the edges of the transformer. Better leakage.
I use only one layer of tape if the wire alone can do the isolation.

 

[CapeCAD]

Here is an example of a 16 layer planar transformer, DC-DC.

Most planar windings are designed using a MCAD software like AutoCAD or Draftsight and imported into the ECAD software using .DXF files as a copper shape.

 

[Flyback]

thanks, impressive planar xsection..planar txfmrs are always short and stubby, so I wonder if they suffer higher-than-normal leakage?