winding my own 300 watt switching transformer

[Rb827]

Hey guys, i need to wind a switching transformer for a power supply that I am building. The transformer will be used as a DC/DC push-pull transformer and it needs to be able to handle 300 watts, and it needs two primary windings to take in -15 and + 15 volts, and the secondary side will output 1000 volts. It will be switched at only 20kHz.

I am looking at two different cores, the EFD30 core, and the ETD34 core. What do you guys think about my choice of cores, and what all do i need to know about winding this transformer? impedances? number of turns? magnetic flux density?

Here are the links for the data sheet of the EDT34, EFD30, and my schematic for the power supply:

ETD34:

http://www.ferroxcube.com/prod/assets/etd34.pdf

EFD30:

http://www.ferroxcube.com/prod/assets/efd30.pdf

Figure 10, 1000V-300 mA BOOSTER: (pages 7-8)

http://www.national.com/an/AN/AN-272.pdf

 

[Skogsgurra]

Jim Williams, the Linear Technology application engineer that died recently, used to say: if it feels right and looks right and windings are around one ohm - then it will work.

Of course, that doesn't help. What Jim was trying to say is that you shouldn't try to optimize the first time. Just get it up and going. Then measure and refine. And keep leakage inductance as low as possible. That means high coupling and tight windings.

If you google Williams, switchmode, transformer you will find several appnotes on this topic. He was very much involved with switched power supplies also the last hours of his laife.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[ScottyUK]

Find a copy of 'Soft Ferrites' by Edgar Snelling. Everything you need to know. The Ferroxcube apps book is worth a look too.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[TurboXS]

I was in the same situation as yourself. It is all new and there doesn't seem to be a straight forward guide to designing these transformers. In the end I just started winding some transformers and as mentioned, measure and test. Mine worked first time but not across the specification. A few more trials and it was working but not optimised.

In other words give it a go and see. Eventually the theory and the practice will join and make more sense.

However given your voltages you will need to very carefully consider the best method for isolation between your windings.

 

[Comcokid]

Real world switch-mode transformer design is not straight forward, especially not as simple as engineering courses in college made it seem. Many companies that do a lot of power supply design will have Magnetics Engineers - engineers with a EE or Physics background who do nothing but do the design and design testing of inductors and transformers.

The best design guide and reference for transformers can be found in a series of books written by Colonel William T. McLyman such as Transformer and Inductor Design Handbook which has been published in several editions (I've tripped across scanned pdf versions of older editions online). It's full of step-by-step examples. However, the book is around $130 new.

Otherwise, go to coremaster.com and look at their application note AN109 which was written by McLyman for a push-pull application. A lot of what goes on in his approach are calculations to make sure the core size will support the wattage, the core window will be large enough to hold the winding needed, and what the approximate temperature rise will be. Just using this app note without addition examples you will miss some details (i.e. the skin effect calculated is as-if the excition is sine-wave, but reality will be switching with harmonics, and there is a unexplained SQRT2 factor applied to wattage in the secondary calcualtions that arises from the example being center-tapped that may be different if you have a 4-diode bridge). The fact that the Coremaster example is for a toroid will not change the design method, and following it and even making minor mistakes will result in a transformer that will work.

You didn't indicate what material your cores are, but you can assume a safe flux density of 0.06 Telsa for most ferrites (Note - Coremaster makes exotic nanocrystalline cores which support a huge flux density with a potential downside of huge inductance). A few quick calculations indicates that your ETD34 core is insufficient for 300 watts (doesn't meet the Kg core geometry miminum needed). Going to a much larger ETD54 core will meet the Kg minimum for a reasonable regulation (Alpha of slightly over 1%), but then you find your window factor is almost 0.6 (not enough room for the wire), so if you decrease your wire size (and increase the losses in the copper) you can make it fit, but It'll have a temp rise of 40 to 50C.

As you can see, transformer design can be a very iterative process. To complicate it, many different magnetics manufacturers use different units for their magnetic materials. Some follow metric units, some follow English units, some mix both, some publish theirs with a slightly different variable symbol where the units (not given) are in mm instead of cm just to be different from a competitor. It can be a daunting task just to make sense of it all.

Of course, I've just commented on the design of the transformer. Since you want to do push-pull (a very straight-forward method and very good choice for the power level your doing) you also have to consider the complexities of a switching circuit, and things like 'flux-walk', 'current mode' Vs 'voltage mode', and the extra parasitics you will have from a transformer that is stepping-up to a high output voltage.

Now that I've discouraged you, just go to Linear Tech, look at page 1 of Application Note 73, have a laugh, and proceed.

Good Luck!

 

[Rb827]

hey thanks guys.

i just went to linear tech, im having trouble getting past the "you're nuts!" part. haha.

so it looks like i basically have a buttload of reading to do. well i shall get started.

as far as the wattage on my ETD34 core, I also saw that im not going to be able to get 300 watts out of it at only 20 kHz, but look at this and tell me what you think:

http://www.wcmagnetics.com/catalog/switchmode-transformer.php

this is a link to west coast magnetic switched power transformers. the ETD44 core can handle 311 watts while being switched at 25kHz.

 

[Comcokid]

I don't know how West Cost Magnetics determines the numbers they show for wattage and switching frequency. It could depend upon the application. They do have a disclamer in the pdf data sheet.

Frequently data sheets like this are a starting-point-of-discussion that leads to the particulars of the custom transformer of the actual application. Just the fact that they start with 20 kHz and go up to 500 kHz indicates that at each of the frequencies indicated they will do a different custom design as the wire sizes and ferrites used will have to change several times over this frequency range.

From the coremaster app note I indicated you will see in step 1 that you calculate the power, and in step 2 you calculate the total apparent power which is double or slightly more than the power. Total apparent power arises from the switching nature of the primary from your transistors or MOSFETs, and the discontinuous nature of the currents in the secoondary from the rectification of the diodes. If the WCM transformer had imposed on the primary a sine-wave, and the output had a resistive load (no switching/current interruptions involved) it would handle about 300 watts. Maybe this is where they got the numbers.

Now, as to some of the power handling of the transformers at higher frequencies they indicated - yea, right! - you could do it for a moment, but parasitic losses (i.e. heating) in the wire and core will make the transformer glow cherry red if you tried to do it more than a short time. Not to mention that the losses in the ferrite core increase with temperature, so as the core heats up, the losses increase as well.

Which leads to some articles in magazines where industry experts have shown that the advertised saturation ratings on inductors are frequently bogus as they may not saturate immediately, but when they warm up after 30 to 90 seconds and the ferrite losses increase from the temperature and the flux handling level decreases, they saturate. Don't believe all that is advertised.