Transformer for 1kW SMPS.

[elektryk]

Hi.

I want to make power supply using SG3525 and IR2110, PCB has been already designed, but the worst problem is probably the output transformer.
I'm going to use ETD49 with divided coilformer, but theoretically primary winding should be above and below secondary like in computer power supply.
Unfortunately I want use first solution for safety reasons, will this SMPS work ok with it?
http://sklep.feryster.pl/index.php?...egory_id=603&option=com_virtuemart&Itemid=163

Which diameter of enamelled wire I have to use for this power? The best would be litzwire, but is a bit expensive and I'm planning to use several wires together, I think that for single wire and 80kHz 0.7mm diameter will be good. (EDIT: Rather for 50kHz.)


Primary will be controlled by half bridge at 280-320V, secondary will be regulated 2x50-70V.

 

[rikkitikkitavi]

By using a divided coilformer you will have a huge leakage inductance, and thatvwill be very bad in hardswitched converters , which is what you have with 3525 .

I suggest you look at LLC converters, or revise your bobbin, coilformer and go interleaved (sandwiched layers of prim secondary)

You use either triple insulated wire or plenty of insulatimg tape. Regarding wire diameter, you have to do a power loss calculation and estimate if temp rise will be to much.

 

[29285]

0.7mm copper wire @1kW/80Khz? You must be joking.
If litzwire is too expensive for you, start building an smps of let us say 30W.

 

[elektryk]

My previous SMPS based on IR2153+IRF840 for amplifier had about 300W and 700W momentary power.

A little overdone but if if I see depth will be about 0.25mm, so 0.5mm diameter at 80kHz seems to be ok, although I'd rather prefer slightly less frequency like 50kHz, then skin depth is about 0.4mm for Cu (unfortunately more windings will be needed).

Which value of flow density I can estimate for ETD49? In previous SMPS it was a bit more than 300mT.

 

[palstanturhin]

The core material data sheet will give you that maximum, which you will then derate by 30%...

 

[elektryk]

This items would be ok?
-wire:
http://sklep.feryster.pl/index.php?...egory_id=730&option=com_virtuemart&Itemid=163

-core:
http://sklep.feryster.pl/index.php?...egory_id=603&option=com_virtuemart&Itemid=163

-coilformer:
http://sklep.feryster.pl/index.php?...egory_id=603&option=com_virtuemart&Itemid=163

-insulation tape:
http://sklep.feryster.pl/index.php?...egory_id=679&option=com_virtuemart&Itemid=163

Which current density choose for sandwich windings?

For this material I have found this about flow density:
H=1kA/m(25°C) 390mT
H=1kA/m(100°C) 510mT
So about 350mT would be optimal?

 

[rikkitikkitavi]

0.7mm copper wire @1kW/80Khz? You must be joking.
If litzwire is too expensive for you, start building an smps of let us say 30W.

By all means, use 2 mm thick. But only single layers.

AC reistance does decrease with increasing wire diameter, despite skin effect. However it increases much much faster with increasing number of layers in each windingm due to various eddy current effects like proximity effect. So in the end it depends on a lot of factors.

So as thick wire as you can fit with minimum number of layers in each winding , interleave windings helps a lot.

But yes, even at 300 V primary, 0,7 mm will lead to execissve voltage drop and power loss.