SMPS toroid transformer question

[thoriated]

Hi -

I just found this thread, and perhaps it is timely for me. I may have a commercial application for about a 3KW push pull forward converter toroid transformer and was wondering if anybody here knew a good design and prototype house for this type of thing. The application is to step 16-40 volts up to 200-400 volts (input range, exact secondary center voltage not yet determined) and I'm thinking of runing the controller at maybe about 100khz, so was thinking that maybe a Litz or tape wound primary, at least, would be a good idea. I presume that type 77 ferrite or similar would be a good choice, but am open to other possibilities.

Any suggestions?

Thanks much in advance.

[mzzj]

Sorry, I dont know any protohouses as I am on different side of pond. And I make my own protos

But for toroids with foil winding I can give suggestion: Dont try


Are you sure you want toroid?

[thoriated]

Are you sure you want toroid?

Having toroid's desirable low radiation and small size is a plus in a competitive market. If you know of a good pot core or other type instead, I'll definitely consider such. As far as 'not trying' foil wound primaries, why not? How about Litz wire?

As far as having it prototyped, I can always do a 'one off' myself. The point in having it prototyped externally is to generate formal documentation and provide a path to go into production.

[Flodstroem]

If target is 3 kW output I would like to suggest Metglas cores. More efficient than Ferrite, easier to wind (if doing it by your self) and lower losses. Study this site: http://www.metglas.com/products/page5_1_6.htm

With those C-transformer you could easily make your own windings (on a coil former): foil for prim and foil or Litz for Sec (but Litz is a very inefficient winding material for a 3kW application)

Also, amorphous toroids are available if this is desirable
Otherwise you should use Planar technology to build your transformer. Farnell has a lot of these cores at the transformer section in their catalogue.

For a planar transformer you dont wind the coils. you make them of printed circuit boards. Very reliable method for to make transformer windings with small tolerances (material: 0.2-0.3 mm FR4 + 105µm Cu, double sided. Kapton in between layers. Also you could make windings of Cu-foil.

Im very interested to follow your development, though I am in the need of such a SMPS (input: 24-42V DC, output 4 x 300V at 1-2A . I hope you will post information further in this thread.

Regards

[mzzj]

Originally posted by Flodstroem

Also, amorphous toroids are available if this is desirable

VAC has some real nice amorphous toroid cores in materials like Vitroperm F. Gives you up to three times bigger usable flux density at same losses as ferrites if I got the graphs right.
Availlability and price are not the strong points I am affraid.

[N-Channel]

Amidon Associates makes an FT-240-77 toroid that I have used for half-bridge forward converters in the 500W-1kW range, with good results. After going over the nomographs provided by Amidon, I determined that the #77 material was the best choice for the frequencies and power levels I was using. At 20kHz, the 2.40" #77 core is good for ~1.2kW, and this figure goes up with frequency, though I wouldn't go too much over 100kHz, as the chart stops at 100kHz. I would think, for trhis type of material, 30-60kHz might work better.

Just my $0.02 worth,

Steve

[Flodstroem]

Hi thoriated

If you should feed the SMPS with 16 V, well if taking to account the efficiency could only be, say 88-90% then you must feed the switch with at least 3.4kW @ 213Amps??!! This should count for a foil wound transformer due to Litz is non efficient. This also counts for a Planar

You will need to parallel several foil windings to reach the target if fed by 16V @ 3.4kW input

Typical efficiency for a Planar is 97-99% (in this case 12W if 97% efficiency and if 400W of total power losses)

Prototyping a transformer (or making a small series)

Primaries: 0.2-0.3 mm Cu-foil, (1-4 turns depending on frequency)
Secondaries: FR4 0.2mm + double sided windings of 105-110µm (25-30turn)
(its very easy to paralleling windings in a Planar)

A benefit for to use Planar is that you could do the winding design your self in a PCB software program and then you could send the design to the factory as a Gerber file (I have done this)

You cant etch the boards at home due to the fact it must be plated 30-4µm Cu platings for the thru holes (for to connect the two windings sides together) and for to make the 70µm windings to become 105-110µm thick.

Windings for primaries could be etched at home using 0.2-0.25 mm Cu foil. After making a film or what ever needed, the pattern could be transfered to the Cu-foil by a chosen method.
Then go to etching the windings to become loose from foil.

Check with PAYTON, a very good supplier for ferrite-planar transformers. They stamp/punsh foil windings to custom design if not available as standard pre tooled windings.

web-site http://www.paytongroup.com/paytonplaner.htm

Another benefit for the Planar transformer design that is its very easy to mount and glue to a heat sink thus getting efficient heat sinking.

If your target is a very high efficiency SMPS (>90%) , then you must keep switching frequency low, say 20-40kHz to keep conduction losses small etc... On the other hand if target is physical format/ small size, then you should use high frequencies as for a planar could be from 100kHz up to 1Mhz.

Regards

[Eva]

There is a simple and cheaper solution that does not involve foils nor exotic core materials.

Use several smaller toroid transformers (36mm O.D. should suffice) wound with the required amount of paralleled 0.75mm or 1mm conventional magnet wires, each one contributing between 500W and 1000W to the output. Design the secondaries in such a way that you can obtain the desired output voltage by connecting all them in series. Drive each set of push-pull primaries from an independent bank of two, four or six low-cost TO-220 MOSFETs having their own set of low-ESR local supply decoupling electrolytics (series secondaries will ensure perfect current sharing).

That's all.

Using foil windings in such an application will cause a lot of trouble because primaries won't be perfectly symmetric and the transformer will tend to saturate. Also, higher flux exotic cores are useless for that application, as any 36mm O.D. toroid will allow for 16V 50Khz operation with just 4 primary turns or less, that are already hard to fit in such a long core (you end up using 5 or 6 turns to ease winding).

[N-Channel]

Eva,

You must be looking inside my 2.5kW inverter, 'cause that's exactly how it derives the high-voltage DC Bus.

Thiorated,

I would go with Eva's suggestion, as trying to parallel the secondaries will not guard against current hogging by one core, because no two cores can be wound exactly the same.

Steve

[mzzj]

Originally posted by Eva

Use several smaller toroid transformers (36mm O.D. should suffice) wound with the required amount of paralleled 0.75mm or 1mm conventional magnet wires, each one contributing between 500W and 1000W to the output. Design the secondaries in such a way that you can obtain the desired output voltage by connecting all them in series.

This is indeed potential alternative. Insulating pri-sec of 10 small toroids can be a less fun, depending on requirements. Fredos had nice looking toroids on some of his amps using UL-spec heat shrink tubing around litz bundle, that is maybe one option.

OT: Anyone knows where to buy triple-insulated wire in smaller quantities?