Flyback converter

The outputs are of different voltages and loads...

Cross-regulation is limited by leakage inductance between the secondaries. It's usually in the 10-20% range, unless you take extreme care with the transformer winding. And it can be much worse, for badly designed transformers, or under light load conditions.

Tim

--
Seven Transistor Labs 
Electrical Engineering Consultation 
Website: http://seventransistorlabs.com

side regulation/ (if the link won't hold together with blank spaces)

RL

.

Or much better, if you design the system carefully. The extreme approach is to have a separate, light, un-loaded winding to sense the voltage being ge nerated across each turn on the core, and a feedback system that controls t he drive so that the drive is switched to fly-back when the voltage across the sense winding hits the desired level.

You'll get a voltage loss in each drive winding depending on the current be ing drawn from it, but that will be independent of the loads on the other w indings - give or take resistive heating of the windings as a whole.

--
Bill Sloman, Sydney

No, that's what I meant. Due to leakage. You can't control for that.

Not sure what you mean by "the drive is switched to fly-back when...". That sounds like the phrasing of a hysteretic (bang-bang) forward converter, but the flyback voltage in such a converter is inconsequential.

You also want a certain nominal load on the sense winding, I suppose ideally half the total rating(!), so that all the other secondaries are centered around their regulation error band -- 0% error at 50% load, +10% error at zero load, -10% at 100% load.

Minimal loading is also required on all secondaries, otherwise overshoot from the primary leaks in and causes the voltages to rise arbitrarily high (50-100%!).

Tim

--
Seven Transistor Labs 
Electrical Engineering Consultation 
Website: http://seventransistorlabs.com

Leakage inductance is flux that threads only some of the windings. It's inc onsequential in this context.

Each winding sees a voltage across it generated by the flux that links with it's windings. Some of this appears as output, some as voltage drop due to the current flowing in that winding across the resistance of that winding.

You were complaining about cross-regulation - where the current drawn from one winding affects the voltage appearing across the others. If you run you r circuit so that the flux variations through the core remain constant, thi s source of voltage error largely goes away.

.

A fly-back converter - in my dialect - is a converter where you build up th e current through the drive coil to some arbitrary level, limited by the he at dissipation in the drive coil, and cut off the drive current when it is a high as you can get away with. This initially produces a rapid change of flux through the core, which generates rapidly rising voltages across the d riven coils, until these voltages get high enough to start driving current through your rectifier diodes into the reservoir capacitors across each out put.

A heavily loaded output will start drawing current earlier than the other o utputs, and will lower the output voltages achieved at the less heavily loa ded outputs. Leakage inductance doesn't really come into it.

Since a flyback inverter has to store all the energy to be transferred befo re it starts transferring it, it's a totally rotten way of using your induc tive components. Forward converters make a lot more sense where you need mu ltiple outputs. You still have to monitor the current through your driven w inding, and kill it to allow the flux through the core to either decay or reverse so that you don't ask the windings to carry enough current to melt themselves (or heat the core to anywhere near its Curie temperature, which is the practical limit).

You've missed the point.

--
Bill Sloman, Sydney

Thanks people...

As far as cross-regulation is concerned, is fly-Buck a better choice for low power (3w) multiple output converter ?

At 3W primary-isolated, multi-output, flyback will make the most sense. Cross-regulation is more technique than topology. At that power level, post-regulators are a head-ache reducing option.

Using an integrated switch, with a pre-developed schematic and parts is probably your easiest solution.

RL

For a common E type flyback transformer, close to 2% leakage can be obtained (interleaved winding)

As for cross regulation, it can be approved by loading the winding, and another method is to stack the diodes of the output (if they are not galvanic isolated), instead of stacking the windings

Cheers

Klaus

SR) flyback converter ??

?

It doesn't sound like a job for a flyback converter. If you want good regul ation of several outputs, all controlled by the primary side, I'd go for a forward converter. Diode drops wreck regulation, so you presumably want act ive rectification.

3W is a low power design, and may not justify spending the kind of money th at it would take to do it right.

My thought would be a bunch of bifilar-wound centre-tapped secondaries, one for each output - with each centre-tap driving an LC filter, and power MOS

-FETs alternately pulling down one or other end of the winding.

You'd have a monitoring winding on the primary side, again a centre-tapped bifilar winding in relatively thin wire, with small power MOS-FETs to pull down one or other end of the winding, and you'd feed back voltage output fr om the centre-tap to control a power op-amp driving the primary side drive windings, again bifilar wound centre-tapped, but in heavy enough wire to ta ke the currents involved. You'd put a small current sensing resistor in ser ies with coil, and use it to switch all the power-MOS-FETs whenever the the current through the driving coil hit its upper limit.

The current being delivered to the secondary windings would still flow in t he same sense, but the excitation current, generating the changing flux in the core to sustain the output voltages, would reverse, so you'd drive a sa wtooth current waveform into the primary driven coils.

You'd need to engineer a break-before make interval between turning off the "active rectifier" MOS-FETs on one side before turning on the MOS-FETs on the other side - which is why you'd need the LC filters on each output.

The interwinding capacitances would be high enough to cope with the "both t ransistors off" period.

Messy - and high current op amps aren't all that cheap or nice. The LM675 c ame to mind, but it's more than ten years since I looked at it's data sheet . The SGS-Ates original, of which the National Semiconductor part was prett y much an exact copy, doesn't come to mind with a part number attached.

Since it's a uni-polar current drive, an emitter-follower boosted op-amp wo uld work, if you were careful to compensate for the extra phase-lag through the emitter-follower (not to mention through the transformer).

--
Bill Sloman, Sydney

hmm..thanks