Push-pull rectangular wave oscillator

[928928]

Recently, I have learned something about oscillator circuit, and I read many articles about articles http://www.apogeeweb.net/article/146.html in the Internet, but I still think it's hard for me.
Now, I meet a problem.
How to achieve rectangular wave oscillation in the following figure, what are the requirements of the transformer, the requirements of the triode?

[vk6zgo]

That circuit, if it oscillates at all, will probably produce a sine wave at some frequency determined by the inductance of the windings & the stray circuit capacitance.

A rectangular wave can be produced by a discrete multivibrator where the RC time constant associated with one device is different from that of the other.

The more commonly known square wave is a "special case" of rectangular wave where both half cycles are the same duration.
Analysis of the two waveforms (which I did too long ago to remember), reveals that the harmonic structure of a rectangular wave differs from that of a square wave.

[nfmax]

that looks like a Royer oscillator, which relies on the magnetic properties of the transformer core material to generate a rectangular waveform. An old circuit, but in studying how it works, you will learn much!

[coromonadalix]

thanks  nfmax, was fun to read

@928928  what are your intentions for this kind of circuit,  i hope its not an dc-dc power supply   you need some voltage control and load and line regulation even pfc correction if needed, this circuit will not provide that.

And read the link nfmax sent, it talk about certain type of transformers cores, not necessarly the standard smps or nowadays 120-220-240v transformers

Example : if it drive or supply a circuit voltage and you put an 3 leg fixed regulator, you may end with somewhat good power supply

[T3sl4co1l]

The circuit shown will not begin oscillation, because no transistors are initially turned on.

Once started, it will continue to oscillate, at a frequency determined by the supply voltage, turns ratio (between collector and base windings), inductance, and hFE and to some extent fT of the transistors.

It may also destroy itself once started.

No capacitances are shown, so it is probably not a resonant circuit.  No supply inductance is shown, so it is not a Royer oscillator as one reply suggests.

Tim

[coromonadalix]

Its a crude schematic, maybe it miss some parts

[bugi]

No supply inductance is shown, so it is not a Royer oscillator as one reply suggests.

The linked wiki page on Royer oscillator does not have supply inductance, either. (No separate inductance at all, in fact.) Is that wiki page's circuit example wrong or perhaps oversimplified, too? (The "resonant" version does have the inductor on the supply..)
This page https://www.smps.us/inverters.html seems to also have the square-wave version without extra inductors. (And seems to explain the operation in quite detail, including how it could start.)

[chris_leyson]

No supply inductance is shown, so it is not a Royer oscillator as one reply suggests

The "classic" Royer oscillator is voltage fed and relies on a transformer with a square BH loop. In Royer's 1955 paper, "A Switching Transistor D-C to A-C Converter Having an Output Frequency Proportional to the D-C Input Voltage" he used a square loop core as part of a push pull oscillator and the output frequency f = V_in/4Nphi_m where phi_m is the saturation flux. Royer used Hipernik V as a core material with a saturation flux temperature coefficient of 0.07%/C.

The problem with the voltage fed Royer oscillator is the high collector current spike at the end of the transistor ON time, also sometimes at the start of the ON period. By feeding the transformer center tap with a small choke, i.e. making it current fed, fixes the current spikes and also improves the efficiency. Placing a capacitor across the transformer then turns it into a current fed resonant Royer which is not really the same as the original Royer which relied on saturation to get the switching action.

Small voltage fed Royer oscillators were often used in calculators to supply vacuum flourescent displays and also supply the heater voltage, usually the transformer was wound on a small pot core. I've got an old desktop calculator PCB somewhere so I will have to dig it out and have a look at the waveforms.

They are also used in some small fixed ratio DC-DC converter modules, usually 1-2W rating. If the data sheet specifies a minimum Vin risetime and recommends a lot of input filtering then it's probably a voltage fed Royer. Taken a few apart to see why they let the magic smoke out. Classic voltage fed Royer not really recommended if you want reliability.

Forget to mention. If you use a material with a very square BH loop like Metglas for example or where B_rem is very close to B_sat then you can also get a "latch up" condition where only one transistor conducts, see Pressman and Billings. One topology that used a smaller square loop core as a driver stage and a larger conventional core for the power stage is the Jensen converter, I think Severns and Bloom is the reference for that one but it's history.

[928928]

Thanks so much

[graybeard]

It is called a Royer oscillator.  It is a relaxation oscillator that is good for simple power converters.  For it to work well it requires a core with "hard" saturation.  It works becasue when the core saturates the field collapses and causes the non-conducting transistor to turn on, and the conducting one to turn off.

Your schematic is missing a few resistors, plus all the ones I have made I used NPNs not PNPs, but either can work.

Recommended cores are usually sold as "tape wound torriodal cores."   The best ones will have a very sharp transition to saturation and a vary flat saturation region.

When designing the transformer it is usually best to spit the restrictive wire loss and core loss about 50-50.

I built one 40 years ago to run a 12V headlight in my 6V motorcycle.  I have another I built in that era sitting in a box in my garage.

[T3sl4co1l]

I thought as much. IIRC, they're originally called Baxandall oscillators, after the famous engineer who published on them, among others:
http://www.sophia-electronica.com/Baxandall_parallel-resonant_Class-D_oscillator1.htm
Royer is the cheapened version; so it is everyone else* who is wrong, calling the resonant and inductor-supplied circuits as such, when that should refer only to the saturation-commutated type.

*Mostly the HV / resonant-fringe kind of side, of electronics hobbyists.

Tim

[Neomys Sapiens]

I built one 40 years ago to run a 12V headlight in my 6V motorcycle.  I have another I built in that era sitting in a box in my garage.

Before my inner Eye, I see an old and tired FCC guy, who can find his peace now after finally coming to know the fast-moving source of broadband interference, which to chase up and down the western coast he spent so much time during his active years.