Explanation required for current and voltage controlled switching supply

[adnan012]

hi,

Can any one explain the working of the feedback part of the given switching power supply (TL431 and Lm2094 and associated parts).

 

[FvM]

Separate PI controllers for voltage and current, with controller outputs OR-ed by a diode network. Usual current-mode flyback converter.

 

[adnan012]

What is the purpose of tl431 in this configuration?

 

[smijesh]

TL 431 act as a voltage regulator. Here Reference pin is connected to cathode so it will work as 2.5V(Internal reference voltage of TL431) shunt regulator like Zener diode

 

[adnan012]

Tl431 Pin 3 is connected to Shunt resistor RS via R17. Can i expect voltage greater than 2.5 across shunt resistor?

 

[FvM]

Can i expect voltage greater than 2.5 across shunt resistor?

Unlikely. The circuit around TL431 and the shunt doesn't look right, there are apparently drawing errors.

- - - Updated - - -

Should be corrected like below

 

[treez]

also, your secondary opto led is connected to the wrong side of the 5uh inductor.

Also, your LC post filter has a resonant frequency of 371Hz (5uh and 330uf)…this is too close to what your feedback crossover frequency is likely to be. You will possibly get oscillations in output.

Your LC post filter should have a resonant frequency that is at least 3 times the feedback loop crossover frequency.

- - - Updated - - -

of the schem in #6, I cant see how the current limiter/regulator opamp is ever going to work.

There is some funny connection at the fet gate too, with an rc that wouldn't want to be there

 

[adnan012]

Should LC filter frequency be greater or less than the feed back crossover frequency?

 

[treez]

it should be greater, preferably at least three times greater, otherwise the smps feedback loop will start reacting to the LC ring of the post filter, and instability may well ensue.
If the LC ring FREQUENCY is much greater than the feedback loop x'over frequency , then no problem, as the smps will not "see" it, and so will not react to it.

 

[FvM]

It's easy to make the feedback loop slow, unfortunately the regulated power supply can become almost useless this way, e.g. you get huge overshoot when causing a transition from constant V to constant I and vive versa.