1:12, 200 KHz transformer

[akis]

Hello, I am new here, and am having trouble building a "simple" coil. I have asked here and there and have read pages upon pages but still basics seem to elude me.

The task is to build a step-up transformer from +/-4 Volts to +/- 50 Volts at 200 KHz and a secondary load of 500 Ohm to 2K Ohm (rough average 1K Ohm).

I have built 4 transformers with two types of ferrite cores and various results : they all have bad regulation, as much as 66% and it seems the greater the step-up the greater the regulation.

I have used RM10 and RM12 formers with N97 which seems to be the right material at 200 KHz having u=2300.

My knowledge is small. I know what currents will be used on the primary and the secondary, and the impedance seen on the primary when the secondary is 1Kohm with a 1:12.5 turns ratio.

So I typically wind 2.5 to 10 turns on the primary, and then wind 25 to 150 turns on the secondary (this bit is very laborious, and I usually lose count). I then plug the transformer in and observe its regulation, which is usually very poor.

I would appreciate any help to be able to calculate how many primary turns I need to be able to minimise regulation, so that I can get 50V on full load without jumping to 70V on no load.

[Davor]

You may try consulting this http://focus.ti.com/lit/ml/slup126/slup126.pdf

[Skippy]

A change from 50V to 70V does not surprise me at all, it could be plenty worse. I suspect most of your regulation woes are ferrite losses, with only a modest amount coming from the wire. To try and see if wire losses are your problem, rewind a core with multiple strands of finer gauge wire, for example instead of a single 0.2mm diameter wire, try 4 x 0.1mm diameter wires. This will reduce skin effect losses. the (possibly) better packing of the thinner wire may also allow more wire to be fitted in the same space, ie. maybe 5 x 0.1mm. Generally, a physically larger core should have lower losses for the same power throughput. I don't know whether it's worth trying paralleling up a pair of identical transformers - it may be educational, even if it doesn't work any better. I presume you are driving the primary with a sine-wave?

[akis]

Yes it is pure sinusoidal. I had thought to wind a second secondary to parallel the first secondary in case there are resistive losses on the secondary , but I do not think it is going to make any difference.

Also I am not sure do I need 1 turn on the primary or may be 10 turns? A practical problem is that each additional turn on the primary demands 12.5 turns on the secondary.

[Skippy]

One other thing worth considering is a powdered iron core transformer. These can be efficient, even at 200 KHz. They are often toroids, which allow quite a lot of space for the windings. The trouble is there's lots of types of core material, plus the different sizes, etc etc, but worth investigating. How many turns you use should be determined by such things as saturation flux in the core, you need to study the fine details of the core data sheet to determine what range of turns, etc is viable.
One other thing - you will find with transformers in general, the regulation is poorest when the load is minimal, and improves as the load is increased. So a cheat is to place a light permanent load on the secondary (maybe 25% of your actual load), then the change in output due to the actual load variation will be less.

[porkluvr]

Hello, I am new here, and am having trouble building a "simple" coil. I have asked here and there and have read pages upon pages but still basics seem to elude me.

First off, ain't nothing simple about building high frequency transformer.

Unless you already have a good deal of background, I would get as much information from Mag-inc.com as I could. I won't say that Magnetics' apps and data are the easiest to comprehend, but I have looked at data from some of the other big boys and could not even get a toe-hold. What I usually found was A completely disjointed, mostly useless collection of facts, but then maybe I am more stupid than what I already knew.

I'm curious why you apparently settled on N97 material for your frequency. I searched and could find almost nothing informative except for advertising hype.

You should perhaps look at some of the toroids at Surplus sales of Nebraska.
They do a poor job of cateloging many of the products they have on their "Ferrite and Powder" pages which means that many items are undervalued if you can see past the descriptions on their pages.

There are standard Ferrites to choose from, and they also have 'MPP' and other powder cores. Remember, you can glue two or three toroids together and arrive at almost 2x or 3x the Al value when that becomes necessary.

There is a lot of useful information at Mag-inc if you are willing to wade through it and that makes the little goodies at SSN possibly very useful.

Keep in mind that their A263104-21G
is near direct replacement for Magnetics' 55118.
330D55118 listed as iron? Nope that' Magnetics 55118
055206-L6-0, listed as Arnold Ferrite? No that's Magnetics 55206.
These should be worthy of your attention.

Remember, any 200kHz transformer you wind must have windings equally spaced as much as possible if they're in a single layer, but I won't tell you that you might not be able to wind a multi layer transformer, just plan the layout and be ready for some dissapointments.

A program called Pexprt would make your job go easier or maybe Aziz could devise you an appropriate device based on your requirements - maybe.

[moodz]

You want something like this ....
http://www.korpi.biz/balun%201%20to%2012%20single.pdf

You will need to change the ferrite material and the cap to extend the low frequency range
but the turns ratio should not alter.

Moodz

[akis]

I have re-run tests on the RM12 / N87 transformer with increased driver power to eliminate issues with the driver under load. The best regulation I got was 14%. I am not sure if 14% is "average" or "bad" or "great" for a 200 KHz transformer

Tomorrow I will try the RM14 cores to see if I can get any better, as they are larger than the RM12. I will follow the suggestions above, ie double 0.5mm for secondary wound in first and occupying the whole window, then tape, then multiple 0.5mm to effect one or two turns for the primary, depending on how many turns I managed to fit for the secondary.

The N87/N97 materials are shown in the list of available cores for the EPCOS RM10, RM12 and RM14. They are listed at 100KHz and the next core up is listed at 500KHz.
for example : http://www.epcos.com/inf/80/db/fer_07/rm_12.pdf

I live in the UK so my choices of suppliers are limited. I have not thought of using small toroids. Are they better than the EPCOS shapes? How would I wind one or two turns on a toroidal?

[porkluvr]

I live in the UK so my choices of suppliers are limited. I have not thought of using small toroids.

Moodz probably has shown a damn fine solution, even if all the devil details have not been handed to you.

If you were to try Surplus Sales of Nebraska you would find that they do make shipments to GB, perhaps even at a reasonable rates. Maybe you could select suitable toroids to experiment with, after doing some informative research.

I do not understand what depth of knowledge you could have possibly gleaned from your Epkos page, but since you are obviously some budding genius, maybe you should just keep beating your head against a wall until things fall into place.

Oh, and the best of luck to you!