30-40KHz sin power generetor needed

Hello, for a project of mine I need a way to generate a power SIN signal of somewhere between 30-40KHz (does not need to be variable. I emphasize power, the signal will be used to control other stuff, so a 555 or a chip generator is useless.
As an indication compare it with the output of a several watts 50Hz transformer, just to indicate what short of power needed (exactly not known)
Is there any easy way to do it?

I was thinking something like a power sinewave RF oscillator in the 30-40KHz range

I have found a suitable schematic using a tube, attached, I think it uses a dual anode triode (dual triode)?

here you just use a mosfet and switch thatmosfet on each 35kHz using microcontroller

nikhilt007 said:

here you just use a mosfet and switch thatmosfet on each 35kHz using microcontroller

Click to expand...

This will generate a square wave. I specified a sinusoidal wave is needed

A sinus generator + a power amplifier in A class with BJT or MOSFET. Power may go up to few kW... no need for tube, or need if you want tens of kW or MW power.

iop95 said:

A sinus generator + a power amplifier in A class with BJT or MOSFET. Power may go up to few kW... no need for tube, or need if you want tens of kW or MW power.

Click to expand...

You say the response of the power amplifier gan go up to 30-40KHz without attenuating too much?

Yes. With MOSFET and good design, may cover 0-80kHz with 3dB variation or if you need a specific band 30 - 40kHz for example, may cover it with close to 0dB variation.

I was thinking of the attached PSU, since it is more simple, but mine will run at 45KHz instead.

I have not figured out how it works
Any guess about the inductances/impedances of the output transformer?

That circuit is only useful for a manufacturer as it requires a special mains transformer and a special inductor. it would be useful to know what power roughly 0-10W?, 0-100W?. . . it is necessary to know the impedance of your load. . .do you need lots of voltage or lots of current? How much distortion can you tolerate or harmonic content?
If your load impedance is roughly the same as a loudspeaker ~ 3 - 15 ohms, then the oscillator + hi-fi amp is the way to go. Be wary of the ratings though, a 3 dB drop of at 45 KHZ, means 1/2 the power!
If you get a ferrite core then a transformer can be wound which will be resonated (tuned ) with a capacitor and will enable any load impedance to be used. An alternative would be to use a PI matching section, two capacitors and one inductor, which will need to be tuned to provide the necessary transformer action.
Frank

chuckey said:

That circuit is only useful for a manufacturer as it requires a special mains transformer and a special inductor. it would be useful to know what power roughly 0-10W?, 0-100W?. . . it is necessary to know the impedance of your load. . .do you need lots of voltage or lots of current? How much distortion can you tolerate or harmonic content?
If your load impedance is roughly the same as a loudspeaker ~ 3 - 15 ohms, then the oscillator + hi-fi amp is the way to go. Be wary of the ratings though, a 3 dB drop of at 45 KHZ, means 1/2 the power!
Frank

Click to expand...

It will drive a magnetic audio amplifier (experimental) that uses a transformer as an amplifier, so I guess current driving capability is needed? But I have no idea about the impedance. The transformer winding that it will drive has 25+25 turns on a TN36/23/15-3R1 Ferroxcube doroid. The power of the oscillator should be in the 5-10W of RF power or so to achieve a suitable amplification.
I avoid the amplifier solution because if one has already an amplifier, he would not need a magnetic amplifier at all. After all, all that is needed is a power sinewave oscillator at 45KHz.

The way I would go would be to use a Line Output Transformer (LOPT) from a TV set. These deliver the Extra High Voltage (EHT) for the Cathode Ray Tube (CRT). The problem you have is that most ferrite core have a square B/H curve so they are only good for square waves such as in power supplies. Your need a core that has a straight B/H curve, which LOPTs have, filter core have a straight B/H curve but in general are not rated for your sort of power. LOPTs run at TV line frequency (16 KHZ) but are tuned to three times this frequency with a power handling in the order of 20 - 100W.
The cheapest way to go , though not the easiest, would be to copy the LOPT stage of a TV set, run it with a rectifier straight from the mains and instead of using the line frequency drive, put on another feedback winding on the LOPT. Remove the EHT winding and replace it with your output winding. The whole circuit (except for your output winding) will be lethal and will need to be put in an insulated box.
The way I would go is to build /get a LV power supply (12 -50 V, 20 W) and build a push-pull power FET amplifier using the LOPT core with all the old winding removed, tune the primary to your 45 KHZ, put on feed back windings, put on a multitapped output winding.
Frank

chuckey said:

The problem you have is that most ferrite core have a square B/H curve so they are only good for square waves such as in power supplies. Your need a core that has a straight B/H curve
Frank

Click to expand...

I guess you refer to the line output transformer only (flyback transformer)? Ordinary ferrite toroids do not have a very square BH curve. There are a few special cores that are used in saturable reactors and magnetic amplifiers that have a stiff square BH curve and so they should be more efficient in these kind of circuits.
I was not aware that the flyback core has a more linear (relaxed) BH curve, that is new to me. By the way, why is this characteristic needed in flybacks?

No, I have been thinking about that all afternoon, and on this point I am wrong. The only thing I can say in my defence is that because the two sections are normally clamped, they can be taken apart and new windings put on, toroids are very difficult to wind and to modify the number of turns.
Frank