How to use a Buck-Converter as a Pre-regulator stage

[JimmyCad]

I am designing a switched mode power supply. I currently have a fixed frequency half bridge resonant circuit as my main inverter, but want to operate it at a fixed frequency. It is my plan that the buck converter will also act as a current source to the main bridge.

I have seen that in this case in the application area I work in, this is typically achieved through the use of a pre-regulator stage such as a buck or a boost converter which regulates the DC bus input voltage to the fixed frequency inverter and achieves regulation that way. I have opted that the buck converter is the best way to do this.

I have an input voltage range of 240-300V that feeds the half bridge converter. From this point onwards I have absolutely no clue where to go. In a typical buck converter the output voltage is fixed (at lets say 200V) and we modulate the switches to keep it that way. However in this case the output voltage is not constant and therefore I am struggling to find material on how to design a buck converter this way.

It is to my understanding that the output capacitor of the buck converter will not exist in this configuration to provide current feed to the main bridge. Correct me if I am wrong.

I am also thinking about using a sychronous buck converter instead of using a free-wheel diode but after researching it seems like this is most important in low voltage, high current applications. In high voltage, lower current applications such as mine - is there any benefit to using a synchronous buck converter rather than an ayschronous one? Maybe the reduction in conduction loss for increased efficiency?

Any advice on how this pre-converter can be designed would be appreciated.

If any more information is required to answer appropriately I can provide it.

Thanks!

[MagicSmoker]

This isn't really a beginner's question, but okay... I assume the goal here is to run the half-bridge at a fixed duty cycle as well as a fixed frequency, so it can be considered a "DC transformer"? If so, then you can either take feedback from the output of the buck pre-regulator, or use an opto or other means of isolation and take feedback from the secondary of the half-bridge. There are pluses and minuses to both approaches.

Your assumption that a synchronous buck won't help much here is correct; it is most advantageous when dealing with a large step down ratio and/or low voltages. Presumably the buck pre-regulator here will provide a nominal 220V or so to the half bridge isolation stage even while the input ranges from 240 - 300V so neither a low voltage nor a large step down ratio.

Whether you can make the buck pre-regulator current-fed will depend on the topology and operation range (ie - above or below resonance) of the resonant half-bridge. Normally I'm a big fan of the buck current-fed (half) bridge in its non-resonant form, but you might have to go with the voltage fed version if your resonant half-bridge needs a choke in the output filter.

[David Hess]

The power supplies in the Tektronix 24xx series of oscilloscopes work the way you describe.  A line voltage input buck converter provides a constant current to an inverter which is clocked by the buck converter's oscillator.  All of the capacitors are on the secondary side and feedback is provided through an optocoupler from the secondary side.

[JimmyCad]

This isn't really a beginner's question, but okay... I assume the goal here is to run the half-bridge at a fixed duty cycle as well as a fixed frequency, so it can be considered a "DC transformer"? If so, then you can either take feedback from the output of the buck pre-regulator, or use an opto or other means of isolation and take feedback from the secondary of the half-bridge. There are pluses and minuses to both approaches.

Your assumption that a synchronous buck won't help much here is correct; it is most advantageous when dealing with a large step down ratio and/or low voltages. Presumably the buck pre-regulator here will provide a nominal 220V or so to the half bridge isolation stage even while the input ranges from 240 - 300V so neither a low voltage nor a large step down ratio.

Whether you can make the buck pre-regulator current-fed will depend on the topology and operation range (ie - above or below resonance) of the resonant half-bridge. Normally I'm a big fan of the buck current-fed (half) bridge in its non-resonant form, but you might have to go with the voltage fed version if your resonant half-bridge needs a choke in the output filter.

Apologies, I didn't realise I post in the wrong section of the forum.

Actually my application has a high output voltage therefore it's more desirable to not have the output choke, so moving that inductor to the input could prove advantageous. The resonant bridge is parallel loaded, and runs slightly above the peak resonant frequency and is fixed.

Is there any benefits to the hard-switched current fed bridge topologies? I assume they will be much simpler to design, however hard switching might be quite inefficient. I do plan on investigating the use of Gallium nitride switches so a hard-switched topology might not actually be a bad idea.

Could I ask, what are the benefits and drawbacks to the two methods you describe? My plan was to use current mode control of the buck inductor and VMC of the secondary side with the optocoupler as you described.

I understand your point about the buck converter, but isn't the efficiency of a synchronous buck always going to be more efficient? Since the conduction losses are related to I_o^2*Rds_on, if we choose a switch with low Rds_on.

[JimmyCad]

The power supplies in the Tektronix 24xx series of oscilloscopes work the way you describe.  A line voltage input buck converter provides a constant current to an inverter which is clocked by the buck converter's oscillator.  All of the capacitors are on the secondary side and feedback is provided through an optocoupler from the secondary side.

Interesting! Do you have one of these supplies? Or have you learnt about its operation elsewhere (any documents, schematics?)

Is the inverter hard switched or a resonant type topology, do you know?

Since the oscillator is shared by the buck and the main inverter bridge, they are operating at the same frequency? Is there any benefit to this at all?

[MagicSmoker]

...
Actually my application has a high output voltage therefore it's more desirable to not have the output choke, so moving that inductor to the input could prove advantageous. The resonant bridge is parallel loaded, and runs slightly above the peak resonant frequency and is fixed.

Everyone seems to think having a choke in the output filter of a HV supply is awful - usually assuming there will be a problem with insulation breakdown - but the same consideration applies to the secondary winding of the transformer. Maybe the choke will end up needing less turns - and therefore have to support a higher V/turn - but one usually has considerable freedom in selecting the core size and gap (distributed or discrete) for the choke, anyway.

At any rate, the phrase, "parallel-loaded," only partially describes the type of resonant topology; you also need to specify whether it is a series or parallel resonant network.

Is there any benefits to the hard-switched current fed bridge topologies? I assume they will be much simpler to design, however hard switching might be quite inefficient. I do plan on investigating the use of Gallium nitride switches so a hard-switched topology might not actually be a bad idea.

Hard-switched is easier to get design, easier to get working, and capable of a much wider load range than most (if not all) resonant topologies. That said, resonant converters can be immensely helpful in minimizing corona formation in high voltage supplies. Whether I'd strongly lean towards one or the other would depend on power level, output voltage, load current range, does it need to tolerate an open and/or short circuited load, etc.

Could I ask, what are the benefits and drawbacks to the two methods you describe? My plan was to use current mode control of the buck inductor and VMC of the secondary side with the optocoupler as you described.

Briefly (not comprehensively): taking feedback from the output of the buck converter means no isolation barrier is required for the feedback signal which means much faster transient response but less accurate control of the actual voltage on the secondary of the bridge converter; invert those pluses and minuses if taking feedback from the secondary of the bridge.

I understand your point about the buck converter, but isn't the efficiency of a synchronous buck always going to be more efficient? Since the conduction losses are related to I_o^2*Rds_on, if we choose a switch with low Rds_on.

Nope. There no guarantee a synchronous buck will be more efficient. In fact, the opposite is often the case, especially at partial loads, and/or when output voltage greater than half the input voltage. The two main reasons are higher total switching losses for the sync. rectifier, especially compared to a Schottky, and operation in "forced" continuous conduction mode all the way down to near zero load which increases conduction losses in the buck switch and inductor at light loading (when the buck inductor would go discontinuous).

[David Hess]

The power supplies in the Tektronix 24xx series of oscilloscopes work the way you describe.  A line voltage input buck converter provides a constant current to an inverter which is clocked by the buck converter's oscillator.  All of the capacitors are on the secondary side and feedback is provided through an optocoupler from the secondary side.

Interesting! Do you have one of these supplies? Or have you learnt about its operation elsewhere (any documents, schematics?)

Is the inverter hard switched or a resonant type topology, do you know?

Since the oscillator is shared by the buck and the main inverter bridge, they are operating at the same frequency? Is there any benefit to this at all?

The power supplies in these oscilloscopes are fully documented in the service manuals which can be found online.  The buck converter and inverter are hard switched.  The really interesting part on these and their earlier power supplies is that the line voltage input buck converter is floating so gate drive requirements for the n-channel MOSFET switch are trivial.  In other words, "ground" for the buck converter and high side switch is the positive supply voltage to the inverter which was about 42 volts.

The current fed inverter topology has the advantage of automatic flux balancing in the transformer.  Synchronous operation is not strictly required but I think it results in less noise.

Note that in theory feedback can be taken from the primary side of the transformer and Tektronix did this on earlier power supplies but they were voltage instead of current driven.  I think by the time of the 24xx series, the current requirements on the low voltage output were too high to allow good regulation with primary side feedback.