SMPS output coupled inductors

[MosK]

Hello All

I'm working on a new power supply design project. This is 2.1 kW , 400 V input (from PFC) and +/-85V output power supply design. I believe that UCC28950 (Phase-Shifted Full-Bridge Controller With Synchronous Rectification) is a good solution.

This power supply drive an audio amplifier and so need to provide two split rails output (please see attached schematic).

I do not have experience with power supply design for audio amplifier and split outputs. I found several full bridge power supply schematics with split rails output and they all had coupled inductors at the output. I rarely have seen these inductors with single output full bridges. I know coupled inductors provide better transient performance but I want to know using them in this application is a must or if it is not a must then what are the trade-off here.

I appreciate your response.

Thanks.

 

[greenocean]

in my experiment, coupled inductor balancing the output voltage if load imbalance

some manufacturers like yamaha, qsc, crest, crown prefer using unregulated smps for their amplifier, some do not even use inductors

and synchronous rectifier may be more suitable for low voltage high current smps, because power loss in secondary diode is Vf x I, I see TDK lambda using synchronous rectifier in their 12V 40A SMPS

maybe this could be a reference : (search in google)
-yamaha emx5000
-QSC plx series PLX-1202, PLX-1602,...
-crown cts 4200a
-crest audio pulse series
-lab gruppen fp-3400 (flyback)

 

[Pafi]

In case of separated inductors: when load of one rail is lower then the ripple current, then it will start operating in discontinuous conduction mode, which results in an elevated voltage. If the other rail is loaded more, then the controller can't give proper PW or phase shift value for both sides the same time -> depending on the feedback sensing point either 1 rail will go up, or the other go down, or both. At high current there will be no significant difference in DC. Common core solves this issue.

Theoretically synchonous rectification could solve the problem mentioned above also, but in practice it should have a very precise control, with very short dead time, otherwise voltage spikes (power loss) are created. I don't know such rectifier control that can manage this issue without creating an other.

By the way: be careful! 170V DC rail needs 400 V breakdown voltage rating for rectifiers in your schematic, otherwise ringing created by commutation reaches breakdown voltage and creates loss. This because synchronous rectification is not recommended here, diodes are more (cost) effective.

I wouldn't use Phase Shift PWM for this purpose, but frequency modulation and LLC topology. In LLC there is no DC inductor and rectifier only have to withstand output voltage, not more. There are other benefits also, like lower EMI, and lower price. The only thing that needs using special material is the low loss of leakage inductance.

 

[MosK]

I wouldn't use Phase Shift PWM for this purpose, but frequency modulation and LLC topology. In LLC there is no DC inductor and rectifier only have to withstand output voltage, not more. There are other benefits also, like lower EMI, and lower price. The only thing that needs using special material is the low loss of leakage inductance.

We are actually developing another 1.2 KW LLC here for another project. At first I wanted got with LLC full bridge or half bridge for this 2.1kW power supply also. But I was warned by other people regarding the difficulties of designing LLC transformer for application above 1kW. I calculated the magnetizing inductance to be 50uH. I was told it would need a large gap in core which comes with side effect. The efficiency will drop significantly too due to large circulating current. There would be lot of circulating current even at the idling mode.I was told phase shifted full bridge is the way to go for 2.1 kW power level. Is there any thing here that I'm missing so that I should reconsider my decision?

 

[Pafi]

I calculated the magnetizing inductance to be 50uH. I was told it would need a large gap in core which comes with side effect. The efficiency will drop significantly too due to large circulating current. There would be lot of circulating current even at the idling mode.

Magnetizing current requirement depends mainly on the required regulation. If input is stabilized, then theoretically you need 0 regulation, therefore almost 0 idle current (LLC degenerated to simple series LC resonant circuit). For charging parasitic capacitances a little current is still needed. For audio purposes even an unregulated PSU, fixed frequency is OK, especially since you use pre-regulator. But if you want some regulation, 10-20 % can be done easily. The same percentage of the nominal current for magnetising current at nominal freq is enough, and at light load the freq will be higher, therefore idle circulating current will be even lower. Practically no loss due to magnetising current. 50 uH is much lower then neccessary. I don't know how you could calculate this.

 

[MosK]

Good points. . The reason I came up with 50uH was that I calculated the LLC for wide input range (325 to 425V). A very conservative design!
Yes. you are right. Most of Power supplies which are designed to drive power amplifiers are unregulated.