LLC regulation over the load range?

[24vtingle]

Hi,
The attached is a 1kW Full Bridge LLC sim 390V to 60V, 80kHz.

The LLC is switched at 80kHz continuously, no matter if full load or no load.
When no loaded, the LLC output creeps up. Its as if it needs a 0.5W dummy load resistor
, or "regulator", to dissipate this "excess no load current".
Have you known LLC's which cause a higher rate of rise of vout when in no load?
What's the maximum dummy load that would be needed to keep such an LLC from rising its Vout
continuously higher?

We do not wish to go into burst mode in no load. The output regulation is very tight from full load to no load considering there's no feedback.
And the pre-charge on Vout means there is no overcurrent in the primary at startup.

 

[mtwieg]

For LLC, this will depend on the Q of the load + resonant circuit as well as the controller's frequency range.

In your simulation, the bridge is driven at a constant frequency, there's apparently no feedback at all. Not sure what you were expecting.

 

[Easy peasy]

All, LLC's have volt rise at the output for no load . . . look at the common graphs.

 

[24vtingle]

Thanks yes, and what would you say is typical for a power throughput for an LLC switching at upper res freq and no load?
(..OK thanks just seen Mtwieg's kind response about depends on Q)
I am thinking this is a small amount of power, and a linear regulator at the output , set to regulate the vout to "Normal vout" + 10V will dissipate max of around 2W, its difficult to see how it could be much more than that?

Just wondering whether to bother to put in a flyback to regenerate the no load overcharge back to the input...or just burn it up in a power resistor.

In your simulation, the bridge is driven at a constant frequency, there's apparently no feedback at all. Not sure what you were expecting.

Thanks yes it just uses a simple fixed freq oscillator. The pre-charge prevents overcurrent at startup...and a drop-dead comparator will latch it off very quickly if there is ever a short or big overload on the output.
As you can see, the LLC gives pretty good reg from no load to full load even with no feedback......but does need the "mothering" pre-charge at startup...that will be courtesy of an added flyback. Also, the loads have inrush limiters , and will not startup till 20 seconds after they are connected up. The max inrush current will always be lower than their max average current draw,

 

[Easy peasy]

@FlyBack: you may as well do an SRC, which is an LLC with very little to zero L series, ( use the low leakage in the Tx )

this can be run at fixed freq and 48.5 + 48.5 % duty cycle ( i.e. full power setting ), and gives Vout = fixed ratio of Vin, you'll still need a minor dummy load -

but it takes away all the LLC nonsense, you can even do a pwm sift start with little penalty.

 

[FvM]

Your requirements aren't clear. LLC is used to perform voltage variation and presumes feedback or possibly feedforward to adapt operation frequency.
For fixed voltage, fixed frequency converter, you can use CLLC topology (symmetrical series capacitors on primary and secondary) if cancellation of transformer leakage inductance is an objective. Unidirectional CLLC can still show some overvoltage in no-load operation due to parasitic effects.

 

[24vtingle]

you can even do a pwm sift start with little penalty.

Thanks yes i like that idea, but the loads (which may be sometimes be connected at startup time) may have large Cin's and so the soft start would have to run for a long time, and i cant say how long as i dont know the maximum C that the LLC output will see...as such, unfortunately i will stick with the pre-charge flyback to charge vout up high enough so i avoid high startup current surges in the LLC.

 

[Easy peasy]

Let us say the Vout is 60V and 100,000 uF, this is 180 Joule

starting this over 2 sec say is 90 watts, it appears there will be little risk to the converter with a phase shift soft start over 2 sec.

 

[24vtingle]

Thanks, in the end , went for an ungapped ETD54 Full bridge 1kW LLC transformer, and sendust torroids for Lmag and Lseries. Also, PQ3220 precharge flyback in COT. Also a PQ2020 flyback for regenerating the no Load power back to the input of the LLC.
Added a little bit of adjustment ability for the LLC frequency but only +/-10kHz. The error amplifier will take care of that.
Looked into SRC but needed to add the Lmag to get the required switching frequency away from the peak of the Vout vs Frequency graph.