Power Supply Topology Guide

[Error_Amplifier]

Please suggest me the right isolated topology to obtain:

320 volt at the input and 3 outputs having
30 V, 12 A
5 V, 4 A
18 V, 32 A

 

[BradtheRad]

I think they should be three separate transformers. Alternately you can make one primary, and 3 secondaries. However I think sudden load changes on one secondary will result in unexpected voltage fluctuations on the other two secondaries.

 

[treez]

5v,4A = offline flyback

30v,12A = 2 transistor forward with fullbridge sec rect using schottkys, or LLC

18v,32A = 2 transistor forward with synchronous secondary rectification.

three separate ones as its too high power for multiple outputs with any accuracy.

Of course, you could do the 5v,4a as a buck off the 30v output.

- - - Updated - - -

or you could do a 100v, 10A LLC to give an intermediary rail.

And then do buck off this for 5v,4a,
And 2tran forwards off the 100v for the other two, and do secondary sync rects, (with split secondaries) with the advantage that the sync rect drive doesnt have to go across the isolation barrier.
Arrange for the sync rects to have gnd referenced drive for ease.

 

[FlapJack]

In round numbers you need a 1000 watt supply, a half bridge or a full bridge would work, but as treez pointed out in post 3 you could have regulation problems, but a well thought out design may not. If there is a problem then a post regulator would be needed.

But one thing is for sure, this is no beginner project.

 

[schmitt trigger]

320VDC or 320VAC?

For AC input, at that power level, you will require active power factor correction.

 

[Error_Amplifier]

Thanks BradtheRad, treez, FlapJack and schmitt trigger for your help.

However I think sudden load changes on one secondary will result in unexpected voltage fluctuations on the other two secondaries.

Yes as pointed out. I have been trying to simulate to get the 3 different outputs using LLC full bridge converter. I have taken feedback of one of the outputs. I'm trying to observe the output voltage variations by changing the 3 output loads from full load to no load. The output regulation is surely getting disturbed.

Please suggest me is there anything which I can do to improve the regulation.

- - - Updated - - -

320VDC or 320VAC?

All voltage's are DC.

- - - Updated - - -

30v,12A = 2 transistor forward with fullbridge sec rect using schottkys, or LLC

I'm trying to built an LLC full bridge converter in simulation. Are the Lk (series) and Lm (parallel) inductors the part of transformer or are they externally connected.

How different is LLC from full bridge?

Just by adjusting the LLC components or the switching frequency is the efficiency improved using LLC converter or is there any other purpose ?

 

[treez]

if you do LLC, then you will have to ensure you get maximum coupling between the windings by interleave winding.
...and then you will have to implement the resonant inductor as a external inductor.
You can use an external magnetising inductor aswell if you want, youll probably find you need to at this power level.

i dont think a single LLC converter can do all those outputs. The frequency required for one output on low load would be too different to the frequency needed for another output simultaneously on heavy load.

- - - Updated - - -

At high power, it becomes easier to reduce your winding loss by having “sectioned” windings for primary and secondary…..besides sectioned bobbins giving the intrinsic leakage inductance, it is just easier to make for a low winding loss with sectioned bobbin if you are using round section ECW. –But such sectioned bobbins have such poor coupling that they aren’t good for multiple outputs.

 

[Error_Amplifier]

Thanks treez for your help.

How to find the ESR of an electrolytic capacitor 1000uF, 50V, at 80khz

Other than reducing the forward voltage of the diode, can I reduce the loss in diode using a snubber ?

 

[treez]

ESR....it should say in datasheet...or at least, should give the ripple current rating. Most el caps have that, if yours doesnt, then i would just pick an el cap that does have that info.

Most snubbers dont reduce diode loss.....mind you, if the snubber quenchs overvoltage transients such that you can use a low voltage schottky , then i suppose it helps lower diode loss. (schottkys dont like overvoltage spikes)

Synchronous rectifiers are the usual way to reduce diode loss.
Otherwise you can parallel sic diodes (PTC) , but often the increased capacitance just makes for more noisy operation.
Higher current rated diodes also have a lower vf but again, more capacitance too.

 

[Error_Amplifier]

LLC converter

Please tell me how can the frequency be changed using digital controller in LLC converter.

 

[treez]

I suppose that seamless change of frequency from a digital controller would be pretty difficult. –Far easier and more smoothly changing with an analog controller such as ICE2HSO1G….i am pretty certain that no digital controller could smoothly change the frequency like an analog controller does. (All the LLC’s that I have ever seen that use a digital controller (eg micro), just use one or two different frequencies, and use input voltage variation to regulate the load, or a kind of on/off control with the single switching frequency)

Of course, for a single output LLC converter, with a fixed vin, and a fixed vout, maybe you do not need a much varying switching frequency. This is because when operated in such a manner, always switching at the LLC’s resonant frequency, the current in the LLC converter just automatically varies to suit the load….ok , its not the most accurate output voltage regulation in the world, but you can kind of use ON/OFF control to kind of get your output voltage fairly tightly regulated…..i have seen a Class D guitar amplifier power supply which behaved in this exact way, using an LLC converter…it was using a fixed frequency driver chip. As long as you have overcurrent limiting then its ok…….the advantage though of a nice smoothly changing frequency is that when the LLC current peaks up, the controller can counteract this by increasing its frequency….so the analog controllers win out here.

So, all the digital LLC controllers I have seen have either been fixed switching frequency, or perhaps they had a main switching frequency, and another switching frequency about double that to cater for light loads…and they do ON/OFF control to approximate the regulation.

We saw a 3kW LLC battery charger which always switched at 100khz, and it handled the battery voltage change by simply varying its input voltage …(ie by varying the output voltage of the PFC stage which supplied the LLC converter).


I suppose if you insist on digital controller frequency changing, then you need to look at how the controllers for BLDC inverters work, as they change their frequency somehow. Unfortunately, I have never done that in software and don’t know how.
Of course, with the LLC, you need to pay attention to the dead time aswell, so its not just frequency change….ideally, you would increase your dead time as switching frequency increased in an LLC converter

 

[Error_Amplifier]

Thanks treez