Just wound a 1 to 1 transformer by hand in 1 minute.

A pair of paralleled 2N7002s. [1]

Sold it.

John

[1] and don't ask what I'm driving the gates with.

Hah, made me laugh!

I can guess. :-)

Me too.

Cheers, James Arthur

Harumph. Bet you can't guess the two extra tricks.

John

I think I've got one of 'em, but guessing out loud would be rude.

Besides, I'm thinking about slow stuff right now--stepper motors. They don't need no stinkin' edges.

Cheers, James Arthur

Was this thread supposed to be educational or instructional?

I vaguely recall that for lower frequencies a bigger iron core or E frame works better than ferrite cores.

Ferrite and air cores are better for high frequencies and this is the opposite of high frequency isn't it?

This transformer would be extremely inefficient, suffering from extremely high hysteresis, right?

It only works for this application because the needed output is all voltage and almost no current, right?

Did I remember the old conventional theory right?

Back before switching power supplies kind of rewrote the rules to drop the weight, right?

Microstepping? That's fun.

John

Any photo? Impedance?

Certainly no elevated temp compatibility. Hot melt glue on PCB assemblies has to be one of the most retarded choices on the planet, with all the RTVs and epoxies and other adhesives that are in the channel.

HMG is what they put the label on your 2 liter soda bottle with. Hot water softens it quite easily.

Bad choice, Johnny.

On a sunny day (Fri, 2 Jan 2009 20:06:46 -0800 (PST)) it happened Greegor wrote in :

Who cares, it was an example of a solution to a problem.

mmm

Well, you just need to define what is high and what is low. I use it at 15uS period time, last time I looked, and that makes about 66kHz. Not a good frequency for an iron core.

No. Have not noticed. Hysteresis is a property of the iron used in transformers. So, what is your brain picture of 'hysteresis'?

Ooops, not much current, but still > 1 Watt of power.

Dunno about your theories:-)

In this application. where the DC / DC converter supplies some PICs, one opamp, one big LCD display with LED backlight, several more LEDS, optocouplers, a MAX232, all in a small box, with 150V 27MHz RF at 2 cm distance, if using a switcher (I have tried) then harmonics are picked up by the antenna feed that is in the same box, as it passes through a SWR transformer circuit. That causes a S9 signal 100% of the time from the harmonics. Now you could perhaps shield that switcher, add a metal box in the box, add half a dozen feed through caps, plus half a dozen ferrite cores as filtering, and PERHAPS you would get your harmonics below microvolt level.

This little thing makes a positive and negative isolated supply (remember the opamp), without any noticeable interference on the S meter, without using a lot of needless power or causing weird supply currents, without all those feed through caps, without the extra ferrites, and without the extra metal box. Perhaps you can learn here something about RF design. Perhaps the idea is wasted on some.

Well, let's see. First, don't use too large a switching mosfet. The 2n7002 has Ron =3D 5 ohms, so two in parallel is 2.5 ohms, more than adequate for a 50-ohm load, and perhaps also for rapid switching of the mosfet and transformer capacitances. Larger mosfets would just add slowing capacitances.

Second, drive with lots of gate current. I like to go ahead and use a power-mosfet driver IC, but you may have used your favorite 1G or 2G cmos gates - they might be up to the task with, Third, add discrete n- and p-mosfets to make a fast power inverter to drive the gate, these should be big enough, but not too big, perhaps 0.2-ohm Fairchild parts in power sot-23 packages. Maybe fdn358P and fdn357N, except those are logic-level gate-drive parts.

Fourth, drive with a bipolar gate voltage, down to say -3V or more, to get the 2n7002 mosfets switched off rapidly, without suffering from the gate spreading-resistance slowing delays. The LM5110 and LM5112 are examples of fast mosfet driver ICs that have a say +8V and -5V gate-drive capability.

Tell us about your toroid core, and the miniature coax, that's hard stuff to find, where'd you get it?

And about as poorly wound as a person could do. My fellow student back in high school vocational electronics could wind better than that.

On a sunny day (Sat, 03 Jan 2009 10:21:27 GMT) it happened Jan Panteltje wrote in :

Since I dunno where this ringcore I am using came from, I did some searches...

Could be this core, but i am usin git at a much lower frequency: ftp://panteltje.com/pub/ringcore_calc.jpg

But the number of turns, the core size, the core color, and the tuning cap are in the same order of magnitude for 66 kHz.

I didn't do many of those things, but nice ideas.

But I'd expect the intermediate fets, or using a fet driver chip, to slow things down... I am getting 2 ns edges. But I don't want to publish my circuit, fiducial responsibilities and such, so...

John

It's a 5000u ferrite toroid with, as I recall, 4 turns of micro-coax. The cable is Joy Signal part number 14.

John

Do I have to give all the money back?

John

I tried some of the exotic/expensive metglas cores as fast pulse transformers. They have astounding low-frequency permeabilities,

100K-ish (1e6 with tricks) but u falls off with frequency, so ferrites are better for fast stuff.

John

We recently re-invented the flying capacitor dc-dc converter, using ssr's. Run in the sub-KHz range with big caps, it's potentially cheap and quiet.

John

If you were using it on an assembly that I was buying, hell yes.

Your knowledge of adhesives and potting materials must be right up there with your grasp of the operation of a vapor phase cleaning system. Right around nil.