Help With Royer Oscillator

Post the schematics, and we can better understand the situation.

John

You can post a schematic several ways. -- You can draw simple schematics in ASCII characters in fixed-width fonts, and post directly to the newsgroup. -- You can cut, paste, and directly post the (ACSII) contents of an LTSpice .asc file directly to the newsgroup. -- you can post a .JPG, .PNG, or other binary image to alt.binary.schematics.design -- you can post a graphic to your own website, or a 3rd-party photo site.

-- Cheers, James Arthur

A quick search on Royer oscillator got me this page:

That BJT version is a guaranteed croaker (Needs diodes to limit reverse Vbe). Did Larkin draw that up ?:-) ...Jim Thompson

"Tim Wescott"

At the start, both MOSFET gates are permanently biased hard on with a 12V zener regulated supply !!

Then somehow it starts ( ??) and each MOSFET is able to rob bias from the other when it's turn comes to be fully conducting.

It says also a 10 amp supply is needed and the 0.68uF, 250V cap "Must Be Good" .

Brutal.

.... Phil

The critical bit is L1. The op said that his circuit is "not current fed"; if he has no L1, it's not a Royer, it's a square-wave multivibrator. And if he's trying to resonate the primary, it will be very inefficient. We need a schematic.

John

Ok, thanks. How's this:

- This is part of a schematic from an old design I got my hands on, these are not the original component values but the circuit still works.

- The simulation run is from 0 to 8uS. Once it starts to oscillate it doesn't stop.

- I probed the base and collector of both transistors and the cap above Q1 (5 traces total). Maybe it's not a Royer but it does oscillate and produce sinewaves.

- Notice Vcap (C1) charges to about 9.5VDC and holds. This doesn't happen if I use mosfets.

- The other thing I notice is the connection from the 5V supply to the

With mosfets I don't get sinewaves, more like squarewaves. It has something to do with caps C1 and C2, they don't charge to a DC level with mosfets.

I really hope I can get this thing to produce sinewaves with fets.

Jason.

The first schematic is used for many CCFL drivers, and they never have diodes. I think the error is in the number of windings on the base drive.

dth

an

Hello Jason,

You are right, this is not a royer circuit.

When using BJT, you have a path via the BC diode, so this is a form of negative feedback resulting in lower average base voltage.

The DC base current is limited by R2, R3, hence the collector current (via HFE).

When using a mosfet, you don=92t have a BC diode and current is only limited by the mosfet itself (and wire resistance).

When (for example) Q2 (but now a mosfet) is on, it will not go off. This is because of the high input impedance, the gate voltage will not drop (as happens with the BJT. It will only go off when for some reason the current through the mosfet becomes soo high, that it will go into current saturation (So drain voltage starts to rise). When the drain voltage of Q2 starts to rise, you may have sufficient loop gain so that it will oscillate. You may probe the drain current, it may be high (depending on the mosfet you selected).

You can make a royer oscillator sith mosfets, but you need additional components to mimic BJT behavior.

Best regards,

Wim PA3DJS

BJT

=A0a

If it had an L1 - between the battery and the centre tap - it wouldn;t be a Royer oscillator but a variant of the Baxandall Class-D oscillator.

The original Royer circuit switched because the transformer core saturated, killing the base drive. Most practical Royer circuits rely on limiting the base drive so the "on" switching transistor comes out of saturation when its collector curent gets high enough.

In either case, resonance between the inductance of the transformer and the odd capacitance doesn't have a first order effect on the frequency of oscillation, though it can mess up switching times.

The Baxandall circuit seems to have been invented to cope with the relatively high inter-winding capacitances of high-raion step-up transformers.

Jim Williams re-invented it - many years after Baxandall had published the circuit in the Proceeding of the (British) IEE - and described it as Royer circuit in his series of application notes on driving cold- cathode back-light for lap-top computers.

-- Bill Sloman, Nijmegen

The misnaming is understandable - Jim Williams described it as a Royer circuit, as do LT app notes covering PWM control of the current feed.

RL

"Kooner" kirjoitti viestissä: snipped-for-privacy@v31g2000pri.googlegroups.com...

What is wrong with using BJT, if that already works?

Thanks for the link Bill... I had read Jim Williams' CCFL backlight driver app notes some years ago, and being interested in the oscillator he was using there did enough digging to realize that it wasn't really a classic Royer circuit, but I never did follow through enough to analyze it as a Baxandall.

---Joel

.com...

BJT

Good question...I'm working on a redesign that must work at an input voltage of 1.0VDC. The oriignal design also works at 1.0VDC but uses germanium transistors. I can't use them for several reasons. Other options are to boost the input and use silicons BJTs or to use Mosfets that have a very low Vth. I chose the Mosfet path but maybe I should try BJTs.

Jason.

ps.com...

e BJT

=A0a

The Baxandall circuit "squeggs" if you use too big an inductor and BJT switches. This doesn't show up in LTSpice simulations, probably because the Gummell-Poon model of the BJT isn't good enough. The are suggestions that using MOSFET switches avoids the "squegging".

-- Bill Sloman, Nijmegen

ps.com...

e BJT

=A0a

Look for the original royer circuit (with the emitter coupled BJT, center tapped resonant circuit between the collectors). This circuit starts to oscillate from the BE forward voltage (so about 0.7V) and provides a sinewave (while the transistors are switching). When properly designed it has good efficiency over a wide range of load variation (collector efficiency > 80%).

If you apply no load, it will take long time before the output is stable. This is because of the stored energy in the series inductor. If you make sure power supply doesn't come up quickly, this LF ripple behavior will be minimal.

For those who want to design these oscillators for higher voltages, the collector peak voltage is about 3*(supply voltage).

If you don't like the transformers, make a class-C collpits oscillator. This also starts from about 0.7V. Disadvantage of this circuit is that you need a higher loaded Q factor to avoid envelope instability (so you need slightly bigger magnetics).

Best regards,

Wim PA3DJS

It

is

he

ne BJT

th

=A0a

Hi Bill, Thanks, for the link. I'm running simulations with Simetrix and I will build something in the lab this week. Could you tell me how the Q of this circuit is defined?

If it's a parallel resonant circuit we're talking about, is it Q =3D R/ sqrt(L/C) ?

where sqrt(L/C) =3D sqrt (500uH / 100nH) =3D 70.7

and R =3D 10K x turns ratio squared ~ 10K x 100 =3D 1 Meg?

I can see that won't give a Q between 5 and 10 so I'm doing something wrong here....

Jason.

I'll buy that.

Isn't his turns ratio sqrt(500uH/50uH), so the 10k would actually reflect as

Of course, Q would still be much higher than 5-10, but I don't think his first circuit there is supposed to be a realistic example. Once he drops the resistor to 22ohms, the Q would be 22*10/70.7 or 3.11, which is on the low side of Tony's recommended values... which I think Bill did purposely to demonstrate the distorted waveforms.

Anyone have a PDF of Baxandall's paper around, per chance?

---Joel

fixed-wi=

photo

--------------------------------------- Posted through