My 4Kw smps finished

[Andy F]

Hi:

I have just finished to test my 4kw smps design, it is a commercial one made on purpose so I want to share some photos because this can be ilustrative for some forum users. It is a half bridge design using a pair of 60A Igbts working on quasi-resonant mode with the help of a series inductor and poliester capacitor.
The transformer is an EE65 3F3 core wound using Litz wire and interleaved windings to keep leakage inductance as low as possible (actually 1uH or so)
I use an small housekeeping non-isolated supply on primary to control the input relay and the protection and startup timing circuits. The Igbts are controlled by an IRS2110 and a pair of 9A peak totem pole drivers and it has provision for a pair of Igbts more if needed.
The duty cycle is fixed at around 120khz with some dead time added on the oscillator side.
The black residue on the Igbt pins is resin not smoke, I have tested it with different Ixys and Fairchild devices to find a compromise between cost and performance.
I am using a Coilcraft 30amps current sense coil for over current protection(the red one on the photo) it is in series with primary trafo to get fast response in case of short or any other anomaly to get cycle by cycle shutdown.On final board it will be on the top side of the board close to the control piggy board.
Now the input side has also provision for a PFC circuit, this one will be on a separate pcb as it was requested by customer to have 2 different versions of the supply. The final application is to feed class D modules, I will publish some photos of them as soon as they are fully tested.
All the oscillation circuitry and PWM control for fan is inside an small piggy board mounted vertically, this one will be epoxy filled by request to avoid easy copy. This is an one person job and I would like to preserve my efforts a bit more.

Regards.

[savu]

Hello Andy ...

Why "quasi-resonant mode with the help of a series inductor and poliester capacitor" ?

Best Regards,

Savu Silviu

[Andy F]

Hi:

Because to get complete resonance at any loading you must dynamically change your switching frequency using any kind of feedback, I want to avoid frequency changes or PWM control because this supply is for audio purposes so quasi-resonant means to select a fixed frequency with fixed duty cycle tuned to a closer frequency respect to resonant tank, that way you get a compromise on efficiency and regulation, the current keeps sinuoidal but the output voltage is not feedback regulated, anyway I dont care too much as I will have a PFC regulating voltage on primary side and my amps have no issues with some volts tolerance on supply on the intended application .

[81bas]

It is not regulated, right?

EDIT: you answered this during I typed my question Thanks!

[Andy F]

Not on a regular way, for feeding pro audio amps you dont need strict regulation, if not imagine the class H and AB monsters with heavy linear toroids on supply, they sag a lot during peaks, even big names like QSC or Crest Audio use non regulated smps on analog and class D amplifiers. The actual duty cycle is almost 50% so if you use a Pfc on primary side you have good and stable regulation for the main power stage and any variation on mains will be compensated on the PFC stage.Another advantage is that you can easily sync the supply with your class D amps if they use fixed frec. modulation.(Not mine)

[savu]

Originally Posted by Andy F

Hi:

Because to get complete resonance at any loading you must dynamically change your switching frequency using any kind of feedback, I want to avoid frequency changes or PWM control because this supply is for audio purposes so quasi-resonant means to select a fixed frequency with fixed duty cycle tuned to a closer frequency respect to resonant tank, that way you get a compromise on efficiency and regulation, the current keeps sinuoidal but the output voltage is not feedback regulated, anyway I dont care too much as I will have a PFC regulating voltage on primary side and my amps have no issues with some volts tolerance on supply on the intended application .

Cool.

any chance of seeing a schematic without component values?

Best Regards,

Savu Silviu

[81bas]

It is really cool!

Andy F, could you please say, how much the output voltage sags under load in your SMPS?

[Andy F]

Hi Savu:

No chance, I do this for living, I can answer questions but I cant expose my work completely, sorry. Anyway, the schematic alone does not mean too much, the hardest part for me was the right layout and some component selection based on my testing and past experience.

[savu]

Originally Posted by Andy F

Hi Savu:

No chance, I do this for living, I can answer questions but I cant expose my work completely, sorry. Anyway, the schematic alone does not mean too much, the hardest part for me was the right layout and some component selection based on my testing and past experience.

That's why I asked for schematic only without component values.
I was just curious if it was based on the same principle as Crest and other unregulated SMPS, or it is slightly different and it is a new concept).

Best regards,

Great work

[savu]

Hi Andy,

I am asking these questions because I am not familiar with this quasi-resonant stuff.
But as I understand it:
You drive the transformer with sine wave made from pwm+IGBT+L+C?

Sorry fro the noobish question

Best Regards,